wuggy/Dactyloidae
wuggy/Dactyloidae
1
0
Fork 0
mirror of https://repo.dactyloidae.xyz/Dactyloidae/UXP.git synced 2026-09-27 02:47:31 +09:00
Code Issues Projects Releases Packages Wiki Activity

Update aom to slightly newer commit ID

Browse source
This commit is contained in:
trav90 2018-10-17 05:59:08 -05:00 • committed by Roy Tam
commit 76f1e6edca
311 changed files with 55292 additions and 33790 deletions
Download patch file Download diff file Expand all files Collapse all files

2
dom/media/platforms/agnostic/AOMDecoder.cpp
View file

@ -218,7 +218,7 @@ AOMDecoder::IsSupportedCodec(const nsAString& aCodecType)
// for a specific aom commit hash so sites can check // for a specific aom commit hash so sites can check
// compatibility. // compatibility.
auto version = NS_ConvertASCIItoUTF16("av1.experimental."); auto version = NS_ConvertASCIItoUTF16("av1.experimental.");
version.AppendLiteral("4d668d7feb1f8abd809d1bca0418570a7f142a36"); version.AppendLiteral("aadbb0251996c8ebb8310567bea330ab7ae9abe4");
return aCodecType.EqualsLiteral("av1") || return aCodecType.EqualsLiteral("av1") ||
aCodecType.Equals(version); aCodecType.Equals(version);
} }

2
media/libaom/README_MCP
View file

@ -10,4 +10,4 @@ The upstream aom git repository is:
https://aomedia.googlesource.com/aom https://aomedia.googlesource.com/aom
The git commit ID used was 4d668d7feb1f8abd809d1bca0418570a7f142a36. The git commit ID used was aadbb0251996c8ebb8310567bea330ab7ae9abe4.

6
media/libaom/config/aom_version.h
View file

@ -1,7 +1,7 @@
#define VERSION_MAJOR 0 #define VERSION_MAJOR 0
#define VERSION_MINOR 0 #define VERSION_MINOR 1
#define VERSION_PATCH 0 #define VERSION_PATCH 0
#define VERSION_EXTRA "" #define VERSION_EXTRA ""
#define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH)) #define VERSION_PACKED ((VERSION_MAJOR<<16)|(VERSION_MINOR<<8)|(VERSION_PATCH))
#define VERSION_STRING_NOSP "Release" #define VERSION_STRING_NOSP "v0.1.0"
#define VERSION_STRING " Release" #define VERSION_STRING " v0.1.0"

39
media/libaom/config/generic/aom_config.asm
View file

@ -5,8 +5,6 @@
.equ ARCH_MIPS , 0 .equ ARCH_MIPS , 0
.equ ARCH_X86 , 0 .equ ARCH_X86 , 0
.equ ARCH_X86_64 , 0 .equ ARCH_X86_64 , 0
.equ HAVE_EDSP , 0
.equ HAVE_MEDIA , 0
.equ HAVE_NEON , 0 .equ HAVE_NEON , 0
.equ HAVE_NEON_ASM , 0 .equ HAVE_NEON_ASM , 0
.equ HAVE_MIPS32 , 0 .equ HAVE_MIPS32 , 0
@ -48,8 +46,6 @@
.equ CONFIG_AV1_ENCODER , 1 .equ CONFIG_AV1_ENCODER , 1
.equ CONFIG_AV1_DECODER , 1 .equ CONFIG_AV1_DECODER , 1
.equ CONFIG_AV1 , 1 .equ CONFIG_AV1 , 1
.equ CONFIG_ENCODERS , 1
.equ CONFIG_DECODERS , 1
.equ CONFIG_STATIC_MSVCRT , 0 .equ CONFIG_STATIC_MSVCRT , 0
.equ CONFIG_SPATIAL_RESAMPLING , 1 .equ CONFIG_SPATIAL_RESAMPLING , 1
.equ CONFIG_REALTIME_ONLY , 0 .equ CONFIG_REALTIME_ONLY , 0
@ -69,36 +65,35 @@
.equ CONFIG_ENCODE_PERF_TESTS , 0 .equ CONFIG_ENCODE_PERF_TESTS , 0
.equ CONFIG_COEFFICIENT_RANGE_CHECKING , 0 .equ CONFIG_COEFFICIENT_RANGE_CHECKING , 0
.equ CONFIG_LOWBITDEPTH , 1 .equ CONFIG_LOWBITDEPTH , 1
.equ CONFIG_HIGHBITDEPTH , 0 .equ CONFIG_HIGHBITDEPTH , 1
.equ CONFIG_EXPERIMENTAL , 0 .equ CONFIG_EXPERIMENTAL , 0
.equ CONFIG_SIZE_LIMIT , 1 .equ CONFIG_SIZE_LIMIT , 1
.equ CONFIG_FP_MB_STATS , 0 .equ CONFIG_FP_MB_STATS , 0
.equ CONFIG_CDEF , 1 .equ CONFIG_CDEF , 1
.equ CONFIG_VAR_TX , 0 .equ CONFIG_VAR_TX , 1
.equ CONFIG_RECT_TX , 1 .equ CONFIG_RECT_TX , 1
.equ CONFIG_REF_MV , 1 .equ CONFIG_RECT_TX_EXT , 0
.equ CONFIG_TPL_MV , 0 .equ CONFIG_TPL_MV , 0
.equ CONFIG_DUAL_FILTER , 1 .equ CONFIG_DUAL_FILTER , 1
.equ CONFIG_CONVOLVE_ROUND , 0 .equ CONFIG_CONVOLVE_ROUND , 0
.equ CONFIG_COMPOUND_ROUND , 0 .equ CONFIG_COMPOUND_ROUND , 0
.equ CONFIG_EXT_TX , 0 .equ CONFIG_EXT_TX , 1
.equ CONFIG_DPCM_INTRA , 0
.equ CONFIG_TX64X64 , 0 .equ CONFIG_TX64X64 , 0
.equ CONFIG_SUB8X8_MC , 0
.equ CONFIG_EXT_INTRA , 1 .equ CONFIG_EXT_INTRA , 1
.equ CONFIG_INTRA_INTERP , 0 .equ CONFIG_INTRA_INTERP , 0
.equ CONFIG_FILTER_INTRA , 0 .equ CONFIG_FILTER_INTRA , 0
.equ CONFIG_INTRA_EDGE , 0
.equ CONFIG_INTRABC , 0 .equ CONFIG_INTRABC , 0
.equ CONFIG_EXT_INTER , 0 .equ CONFIG_EXT_INTER , 1
.equ CONFIG_INTERINTRA , 0 .equ CONFIG_INTERINTRA , 1
.equ CONFIG_WEDGE , 0 .equ CONFIG_WEDGE , 1
.equ CONFIG_COMPOUND_SEGMENT , 0 .equ CONFIG_COMPOUND_SEGMENT , 1
.equ CONFIG_EXT_REFS , 1 .equ CONFIG_EXT_REFS , 1
.equ CONFIG_GLOBAL_MOTION , 1 .equ CONFIG_GLOBAL_MOTION , 1
.equ CONFIG_NEW_QUANT , 0 .equ CONFIG_NEW_QUANT , 0
.equ CONFIG_SUPERTX , 0 .equ CONFIG_SUPERTX , 0
.equ CONFIG_ANS , 0 .equ CONFIG_ANS , 0
.equ CONFIG_EC_MULTISYMBOL , 1
.equ CONFIG_NEW_TOKENSET , 1
.equ CONFIG_LOOP_RESTORATION , 0 .equ CONFIG_LOOP_RESTORATION , 0
.equ CONFIG_EXT_PARTITION , 0 .equ CONFIG_EXT_PARTITION , 0
.equ CONFIG_EXT_PARTITION_TYPES , 0 .equ CONFIG_EXT_PARTITION_TYPES , 0
@ -108,20 +103,18 @@
.equ CONFIG_NCOBMC , 0 .equ CONFIG_NCOBMC , 0
.equ CONFIG_WARPED_MOTION , 1 .equ CONFIG_WARPED_MOTION , 1
.equ CONFIG_Q_ADAPT_PROBS , 0 .equ CONFIG_Q_ADAPT_PROBS , 0
.equ CONFIG_SUBFRAME_PROB_UPDATE , 0
.equ CONFIG_BITSTREAM_DEBUG , 0 .equ CONFIG_BITSTREAM_DEBUG , 0
.equ CONFIG_ALT_INTRA , 1 .equ CONFIG_ALT_INTRA , 1
.equ CONFIG_PALETTE , 1 .equ CONFIG_PALETTE , 1
.equ CONFIG_PALETTE_DELTA_ENCODING , 0 .equ CONFIG_PALETTE_DELTA_ENCODING , 0
.equ CONFIG_DAALA_EC , 1
.equ CONFIG_RAWBITS , 0 .equ CONFIG_RAWBITS , 0
.equ CONFIG_EC_SMALLMUL , 0 .equ CONFIG_EC_SMALLMUL , 1
.equ CONFIG_PVQ , 0 .equ CONFIG_PVQ , 0
.equ CONFIG_CFL , 0 .equ CONFIG_CFL , 0
.equ CONFIG_XIPHRC , 0 .equ CONFIG_XIPHRC , 0
.equ CONFIG_CB4X4 , 1 .equ CONFIG_CB4X4 , 1
.equ CONFIG_CHROMA_2X2 , 0 .equ CONFIG_CHROMA_2X2 , 0
.equ CONFIG_CHROMA_SUB8X8 , 0 .equ CONFIG_CHROMA_SUB8X8 , 1
.equ CONFIG_FRAME_SIZE , 0 .equ CONFIG_FRAME_SIZE , 0
.equ CONFIG_DELTA_Q , 1 .equ CONFIG_DELTA_Q , 1
.equ CONFIG_EXT_DELTA_Q , 1 .equ CONFIG_EXT_DELTA_Q , 1
@ -129,10 +122,10 @@
.equ CONFIG_FILTER_7BIT , 1 .equ CONFIG_FILTER_7BIT , 1
.equ CONFIG_PARALLEL_DEBLOCKING , 0 .equ CONFIG_PARALLEL_DEBLOCKING , 0
.equ CONFIG_PARALLEL_DEBLOCKING_15TAP , 0 .equ CONFIG_PARALLEL_DEBLOCKING_15TAP , 0
.equ CONFIG_LOOPFILTERING_ACROSS_TILES , 0 .equ CONFIG_LOOPFILTERING_ACROSS_TILES , 1
.equ CONFIG_TILE_GROUPS , 1 .equ CONFIG_TILE_GROUPS , 1
.equ CONFIG_EC_ADAPT , 1 .equ CONFIG_EC_ADAPT , 1
.equ CONFIG_TEMPMV_SIGNALING , 0 .equ CONFIG_TEMPMV_SIGNALING , 1
.equ CONFIG_RD_DEBUG , 0 .equ CONFIG_RD_DEBUG , 0
.equ CONFIG_REFERENCE_BUFFER , 1 .equ CONFIG_REFERENCE_BUFFER , 1
.equ CONFIG_COEF_INTERLEAVE , 0 .equ CONFIG_COEF_INTERLEAVE , 0
@ -150,6 +143,8 @@
.equ CONFIG_NEW_MULTISYMBOL , 0 .equ CONFIG_NEW_MULTISYMBOL , 0
.equ CONFIG_COMPOUND_SINGLEREF , 0 .equ CONFIG_COMPOUND_SINGLEREF , 0
.equ CONFIG_AOM_QM , 0 .equ CONFIG_AOM_QM , 0
.equ CONFIG_LOWDELAY_COMPOUND , 0 .equ CONFIG_ONE_SIDED_COMPOUND , 1
.equ CONFIG_SMOOTH_HV , 0
.equ CONFIG_VAR_REFS , 0
.equ CONFIG_ANALYZER , 0 .equ CONFIG_ANALYZER , 0
.section .note.GNU-stack,"",%progbits .section .note.GNU-stack,"",%progbits

39
media/libaom/config/generic/aom_config.h
View file

@ -15,8 +15,6 @@
#define ARCH_MIPS 0 #define ARCH_MIPS 0
#define ARCH_X86 0 #define ARCH_X86 0
#define ARCH_X86_64 0 #define ARCH_X86_64 0
#define HAVE_EDSP 0
#define HAVE_MEDIA 0
#define HAVE_NEON 0 #define HAVE_NEON 0
#define HAVE_NEON_ASM 0 #define HAVE_NEON_ASM 0
#define HAVE_MIPS32 0 #define HAVE_MIPS32 0
@ -58,8 +56,6 @@
#define CONFIG_AV1_ENCODER 1 #define CONFIG_AV1_ENCODER 1
#define CONFIG_AV1_DECODER 1 #define CONFIG_AV1_DECODER 1
#define CONFIG_AV1 1 #define CONFIG_AV1 1
#define CONFIG_ENCODERS 1
#define CONFIG_DECODERS 1
#define CONFIG_STATIC_MSVCRT 0 #define CONFIG_STATIC_MSVCRT 0
#define CONFIG_SPATIAL_RESAMPLING 1 #define CONFIG_SPATIAL_RESAMPLING 1
#define CONFIG_REALTIME_ONLY 0 #define CONFIG_REALTIME_ONLY 0
@ -79,36 +75,35 @@
#define CONFIG_ENCODE_PERF_TESTS 0 #define CONFIG_ENCODE_PERF_TESTS 0
#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 #define CONFIG_COEFFICIENT_RANGE_CHECKING 0
#define CONFIG_LOWBITDEPTH 1 #define CONFIG_LOWBITDEPTH 1
#define CONFIG_HIGHBITDEPTH 0 #define CONFIG_HIGHBITDEPTH 1
#define CONFIG_EXPERIMENTAL 0 #define CONFIG_EXPERIMENTAL 0
#define CONFIG_SIZE_LIMIT 1 #define CONFIG_SIZE_LIMIT 1
#define CONFIG_FP_MB_STATS 0 #define CONFIG_FP_MB_STATS 0
#define CONFIG_CDEF 1 #define CONFIG_CDEF 1
#define CONFIG_VAR_TX 0 #define CONFIG_VAR_TX 1
#define CONFIG_RECT_TX 1 #define CONFIG_RECT_TX 1
#define CONFIG_REF_MV 1 #define CONFIG_RECT_TX_EXT 0
#define CONFIG_TPL_MV 0 #define CONFIG_TPL_MV 0
#define CONFIG_DUAL_FILTER 1 #define CONFIG_DUAL_FILTER 1
#define CONFIG_CONVOLVE_ROUND 0 #define CONFIG_CONVOLVE_ROUND 0
#define CONFIG_COMPOUND_ROUND 0 #define CONFIG_COMPOUND_ROUND 0
#define CONFIG_EXT_TX 0 #define CONFIG_EXT_TX 1
#define CONFIG_DPCM_INTRA 0
#define CONFIG_TX64X64 0 #define CONFIG_TX64X64 0
#define CONFIG_SUB8X8_MC 0
#define CONFIG_EXT_INTRA 1 #define CONFIG_EXT_INTRA 1
#define CONFIG_INTRA_INTERP 0 #define CONFIG_INTRA_INTERP 0
#define CONFIG_FILTER_INTRA 0 #define CONFIG_FILTER_INTRA 0
#define CONFIG_INTRA_EDGE 0
#define CONFIG_INTRABC 0 #define CONFIG_INTRABC 0
#define CONFIG_EXT_INTER 0 #define CONFIG_EXT_INTER 1
#define CONFIG_INTERINTRA 0 #define CONFIG_INTERINTRA 1
#define CONFIG_WEDGE 0 #define CONFIG_WEDGE 1
#define CONFIG_COMPOUND_SEGMENT 0 #define CONFIG_COMPOUND_SEGMENT 1
#define CONFIG_EXT_REFS 1 #define CONFIG_EXT_REFS 1
#define CONFIG_GLOBAL_MOTION 1 #define CONFIG_GLOBAL_MOTION 1
#define CONFIG_NEW_QUANT 0 #define CONFIG_NEW_QUANT 0
#define CONFIG_SUPERTX 0 #define CONFIG_SUPERTX 0
#define CONFIG_ANS 0 #define CONFIG_ANS 0
#define CONFIG_EC_MULTISYMBOL 1
#define CONFIG_NEW_TOKENSET 1
#define CONFIG_LOOP_RESTORATION 0 #define CONFIG_LOOP_RESTORATION 0
#define CONFIG_EXT_PARTITION 0 #define CONFIG_EXT_PARTITION 0
#define CONFIG_EXT_PARTITION_TYPES 0 #define CONFIG_EXT_PARTITION_TYPES 0
@ -118,20 +113,18 @@
#define CONFIG_NCOBMC 0 #define CONFIG_NCOBMC 0
#define CONFIG_WARPED_MOTION 1 #define CONFIG_WARPED_MOTION 1
#define CONFIG_Q_ADAPT_PROBS 0 #define CONFIG_Q_ADAPT_PROBS 0
#define CONFIG_SUBFRAME_PROB_UPDATE 0
#define CONFIG_BITSTREAM_DEBUG 0 #define CONFIG_BITSTREAM_DEBUG 0
#define CONFIG_ALT_INTRA 1 #define CONFIG_ALT_INTRA 1
#define CONFIG_PALETTE 1 #define CONFIG_PALETTE 1
#define CONFIG_PALETTE_DELTA_ENCODING 0 #define CONFIG_PALETTE_DELTA_ENCODING 0
#define CONFIG_DAALA_EC 1
#define CONFIG_RAWBITS 0 #define CONFIG_RAWBITS 0
#define CONFIG_EC_SMALLMUL 0 #define CONFIG_EC_SMALLMUL 1
#define CONFIG_PVQ 0 #define CONFIG_PVQ 0
#define CONFIG_CFL 0 #define CONFIG_CFL 0
#define CONFIG_XIPHRC 0 #define CONFIG_XIPHRC 0
#define CONFIG_CB4X4 1 #define CONFIG_CB4X4 1
#define CONFIG_CHROMA_2X2 0 #define CONFIG_CHROMA_2X2 0
#define CONFIG_CHROMA_SUB8X8 0 #define CONFIG_CHROMA_SUB8X8 1
#define CONFIG_FRAME_SIZE 0 #define CONFIG_FRAME_SIZE 0
#define CONFIG_DELTA_Q 1 #define CONFIG_DELTA_Q 1
#define CONFIG_EXT_DELTA_Q 1 #define CONFIG_EXT_DELTA_Q 1
@ -139,10 +132,10 @@
#define CONFIG_FILTER_7BIT 1 #define CONFIG_FILTER_7BIT 1
#define CONFIG_PARALLEL_DEBLOCKING 0 #define CONFIG_PARALLEL_DEBLOCKING 0
#define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 #define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
#define CONFIG_LOOPFILTERING_ACROSS_TILES 0 #define CONFIG_LOOPFILTERING_ACROSS_TILES 1
#define CONFIG_TILE_GROUPS 1 #define CONFIG_TILE_GROUPS 1
#define CONFIG_EC_ADAPT 1 #define CONFIG_EC_ADAPT 1
#define CONFIG_TEMPMV_SIGNALING 0 #define CONFIG_TEMPMV_SIGNALING 1
#define CONFIG_RD_DEBUG 0 #define CONFIG_RD_DEBUG 0
#define CONFIG_REFERENCE_BUFFER 1 #define CONFIG_REFERENCE_BUFFER 1
#define CONFIG_COEF_INTERLEAVE 0 #define CONFIG_COEF_INTERLEAVE 0
@ -160,7 +153,9 @@
#define CONFIG_NEW_MULTISYMBOL 0 #define CONFIG_NEW_MULTISYMBOL 0
#define CONFIG_COMPOUND_SINGLEREF 0 #define CONFIG_COMPOUND_SINGLEREF 0
#define CONFIG_AOM_QM 0 #define CONFIG_AOM_QM 0
#define CONFIG_LOWDELAY_COMPOUND 0 #define CONFIG_ONE_SIDED_COMPOUND 1
#define CONFIG_SMOOTH_HV 0
#define CONFIG_VAR_REFS 0
#define CONFIG_ANALYZER 0 #define CONFIG_ANALYZER 0
#define DECODE_WIDTH_LIMIT 8192 #define DECODE_WIDTH_LIMIT 8192
#define DECODE_HEIGHT_LIMIT 4608 #define DECODE_HEIGHT_LIMIT 4608

1476
media/libaom/config/generic/aom_dsp_rtcd.h
View file

File diff suppressed because it is too large Load diff

179
media/libaom/config/generic/av1_rtcd.h
View file

@ -50,9 +50,6 @@ void aom_clpf_hblock_hbd_c(uint16_t *dst, const uint16_t *src, int dstride, int
int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
#define av1_block_error av1_block_error_c #define av1_block_error av1_block_error_c
int64_t av1_block_error_fp_c(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
#define av1_block_error_fp av1_block_error_fp_c
void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
#define av1_convolve_horiz av1_convolve_horiz_c #define av1_convolve_horiz av1_convolve_horiz_c
@ -116,9 +113,156 @@ int av1_full_search_sad_c(const struct macroblock *x, const struct mv *ref_mv, i
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type); void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type);
#define av1_fwd_idtx av1_fwd_idtx_c #define av1_fwd_idtx av1_fwd_idtx_c
void av1_fwd_txfm2d_16x16_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_16x16 av1_fwd_txfm2d_16x16_c
void av1_fwd_txfm2d_32x32_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_32x32 av1_fwd_txfm2d_32x32_c
void av1_fwd_txfm2d_4x4_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_4x4 av1_fwd_txfm2d_4x4_c
void av1_fwd_txfm2d_64x64_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_64x64 av1_fwd_txfm2d_64x64_c
void av1_fwd_txfm2d_8x8_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_8x8 av1_fwd_txfm2d_8x8_c
void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_fwht4x4 av1_fwht4x4_c #define av1_fwht4x4 av1_fwht4x4_c
int64_t av1_highbd_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
#define av1_highbd_block_error av1_highbd_block_error_c
void av1_highbd_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8 av1_highbd_convolve8_c
void av1_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg av1_highbd_convolve8_avg_c
void av1_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_horiz av1_highbd_convolve8_avg_horiz_c
void av1_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_vert av1_highbd_convolve8_avg_vert_c
void av1_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_horiz av1_highbd_convolve8_horiz_c
void av1_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_vert av1_highbd_convolve8_vert_c
void av1_highbd_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_avg av1_highbd_convolve_avg_c
void av1_highbd_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_copy av1_highbd_convolve_copy_c
void av1_highbd_convolve_horiz_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
#define av1_highbd_convolve_horiz av1_highbd_convolve_horiz_c
void av1_highbd_convolve_init_c(void);
#define av1_highbd_convolve_init av1_highbd_convolve_init_c
void av1_highbd_convolve_vert_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
#define av1_highbd_convolve_vert av1_highbd_convolve_vert_c
void av1_highbd_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x16 av1_highbd_fht16x16_c
void av1_highbd_fht16x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x32 av1_highbd_fht16x32_c
void av1_highbd_fht16x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x4 av1_highbd_fht16x4_c
void av1_highbd_fht16x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x8 av1_highbd_fht16x8_c
void av1_highbd_fht32x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x16 av1_highbd_fht32x16_c
void av1_highbd_fht32x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x32 av1_highbd_fht32x32_c
void av1_highbd_fht32x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x8 av1_highbd_fht32x8_c
void av1_highbd_fht4x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x16 av1_highbd_fht4x16_c
void av1_highbd_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x4 av1_highbd_fht4x4_c
void av1_highbd_fht4x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x8 av1_highbd_fht4x8_c
void av1_highbd_fht8x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x16 av1_highbd_fht8x16_c
void av1_highbd_fht8x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x32 av1_highbd_fht8x32_c
void av1_highbd_fht8x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x4 av1_highbd_fht8x4_c
void av1_highbd_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x8 av1_highbd_fht8x8_c
void av1_highbd_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_highbd_fwht4x4 av1_highbd_fwht4x4_c
void av1_highbd_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd);
#define av1_highbd_iht16x16_256_add av1_highbd_iht16x16_256_add_c
void av1_highbd_iht16x32_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x32_512_add av1_highbd_iht16x32_512_add_c
void av1_highbd_iht16x4_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x4_64_add av1_highbd_iht16x4_64_add_c
void av1_highbd_iht16x8_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x8_128_add av1_highbd_iht16x8_128_add_c
void av1_highbd_iht32x16_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x16_512_add av1_highbd_iht32x16_512_add_c
void av1_highbd_iht32x8_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x8_256_add av1_highbd_iht32x8_256_add_c
void av1_highbd_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x16_64_add av1_highbd_iht4x16_64_add_c
void av1_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x4_16_add av1_highbd_iht4x4_16_add_c
void av1_highbd_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x8_32_add av1_highbd_iht4x8_32_add_c
void av1_highbd_iht8x16_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x16_128_add av1_highbd_iht8x16_128_add_c
void av1_highbd_iht8x32_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x32_256_add av1_highbd_iht8x32_256_add_c
void av1_highbd_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x4_32_add av1_highbd_iht8x4_32_add_c
void av1_highbd_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x8_64_add av1_highbd_iht8x8_64_add_c
void av1_highbd_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_b av1_highbd_quantize_b_c
void av1_highbd_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_fp av1_highbd_quantize_fp_c
void av1_highbd_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_highbd_temporal_filter_apply av1_highbd_temporal_filter_apply_c
void av1_highbd_warp_affine_c(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
#define av1_highbd_warp_affine av1_highbd_warp_affine_c
void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
#define av1_iht16x16_256_add av1_iht16x16_256_add_c #define av1_iht16x16_256_add av1_iht16x16_256_add_c
@ -161,6 +305,21 @@ void av1_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride
void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
#define av1_iht8x8_64_add av1_iht8x8_64_add_c #define av1_iht8x8_64_add av1_iht8x8_64_add_c
void av1_inv_txfm2d_add_16x16_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_16x16 av1_inv_txfm2d_add_16x16_c
void av1_inv_txfm2d_add_32x32_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_32x32 av1_inv_txfm2d_add_32x32_c
void av1_inv_txfm2d_add_4x4_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_4x4 av1_inv_txfm2d_add_4x4_c
void av1_inv_txfm2d_add_64x64_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_64x64 av1_inv_txfm2d_add_64x64_c
void av1_inv_txfm2d_add_8x8_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_8x8 av1_inv_txfm2d_add_8x8_c
void av1_lowbd_convolve_init_c(void); void av1_lowbd_convolve_init_c(void);
#define av1_lowbd_convolve_init av1_lowbd_convolve_init_c #define av1_lowbd_convolve_init av1_lowbd_convolve_init_c
@ -176,9 +335,21 @@ void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int
void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_temporal_filter_apply av1_temporal_filter_apply_c #define av1_temporal_filter_apply av1_temporal_filter_apply_c
void av1_warp_affine_c(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_c(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
#define av1_warp_affine av1_warp_affine_c #define av1_warp_affine av1_warp_affine_c
void av1_wedge_compute_delta_squares_c(int16_t *d, const int16_t *a, const int16_t *b, int N);
#define av1_wedge_compute_delta_squares av1_wedge_compute_delta_squares_c
int av1_wedge_sign_from_residuals_c(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
#define av1_wedge_sign_from_residuals av1_wedge_sign_from_residuals_c
uint64_t av1_wedge_sse_from_residuals_c(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
#define av1_wedge_sse_from_residuals av1_wedge_sse_from_residuals_c
double compute_cross_correlation_c(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
#define compute_cross_correlation compute_cross_correlation_c
void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
#define copy_4x4_16bit_to_16bit copy_4x4_16bit_to_16bit_c #define copy_4x4_16bit_to_16bit copy_4x4_16bit_to_16bit_c

39
media/libaom/config/linux/arm/aom_config.asm
View file

@ -5,8 +5,6 @@
.equ ARCH_MIPS , 0 .equ ARCH_MIPS , 0
.equ ARCH_X86 , 0 .equ ARCH_X86 , 0
.equ ARCH_X86_64 , 0 .equ ARCH_X86_64 , 0
.equ HAVE_EDSP , 0
.equ HAVE_MEDIA , 1
.equ HAVE_NEON , 1 .equ HAVE_NEON , 1
.equ HAVE_NEON_ASM , 1 .equ HAVE_NEON_ASM , 1
.equ HAVE_MIPS32 , 0 .equ HAVE_MIPS32 , 0
@ -48,8 +46,6 @@
.equ CONFIG_AV1_ENCODER , 1 .equ CONFIG_AV1_ENCODER , 1
.equ CONFIG_AV1_DECODER , 1 .equ CONFIG_AV1_DECODER , 1
.equ CONFIG_AV1 , 1 .equ CONFIG_AV1 , 1
.equ CONFIG_ENCODERS , 1
.equ CONFIG_DECODERS , 1
.equ CONFIG_STATIC_MSVCRT , 0 .equ CONFIG_STATIC_MSVCRT , 0
.equ CONFIG_SPATIAL_RESAMPLING , 1 .equ CONFIG_SPATIAL_RESAMPLING , 1
.equ CONFIG_REALTIME_ONLY , 1 .equ CONFIG_REALTIME_ONLY , 1
@ -69,36 +65,35 @@
.equ CONFIG_ENCODE_PERF_TESTS , 0 .equ CONFIG_ENCODE_PERF_TESTS , 0
.equ CONFIG_COEFFICIENT_RANGE_CHECKING , 0 .equ CONFIG_COEFFICIENT_RANGE_CHECKING , 0
.equ CONFIG_LOWBITDEPTH , 1 .equ CONFIG_LOWBITDEPTH , 1
.equ CONFIG_HIGHBITDEPTH , 0 .equ CONFIG_HIGHBITDEPTH , 1
.equ CONFIG_EXPERIMENTAL , 0 .equ CONFIG_EXPERIMENTAL , 0
.equ CONFIG_SIZE_LIMIT , 1 .equ CONFIG_SIZE_LIMIT , 1
.equ CONFIG_FP_MB_STATS , 0 .equ CONFIG_FP_MB_STATS , 0
.equ CONFIG_CDEF , 1 .equ CONFIG_CDEF , 1
.equ CONFIG_VAR_TX , 0 .equ CONFIG_VAR_TX , 1
.equ CONFIG_RECT_TX , 1 .equ CONFIG_RECT_TX , 1
.equ CONFIG_REF_MV , 1 .equ CONFIG_RECT_TX_EXT , 0
.equ CONFIG_TPL_MV , 0 .equ CONFIG_TPL_MV , 0
.equ CONFIG_DUAL_FILTER , 1 .equ CONFIG_DUAL_FILTER , 1
.equ CONFIG_CONVOLVE_ROUND , 0 .equ CONFIG_CONVOLVE_ROUND , 0
.equ CONFIG_COMPOUND_ROUND , 0 .equ CONFIG_COMPOUND_ROUND , 0
.equ CONFIG_EXT_TX , 0 .equ CONFIG_EXT_TX , 1
.equ CONFIG_DPCM_INTRA , 0
.equ CONFIG_TX64X64 , 0 .equ CONFIG_TX64X64 , 0
.equ CONFIG_SUB8X8_MC , 0
.equ CONFIG_EXT_INTRA , 1 .equ CONFIG_EXT_INTRA , 1
.equ CONFIG_INTRA_INTERP , 0 .equ CONFIG_INTRA_INTERP , 0
.equ CONFIG_FILTER_INTRA , 0 .equ CONFIG_FILTER_INTRA , 0
.equ CONFIG_INTRA_EDGE , 0
.equ CONFIG_INTRABC , 0 .equ CONFIG_INTRABC , 0
.equ CONFIG_EXT_INTER , 0 .equ CONFIG_EXT_INTER , 1
.equ CONFIG_INTERINTRA , 0 .equ CONFIG_INTERINTRA , 1
.equ CONFIG_WEDGE , 0 .equ CONFIG_WEDGE , 1
.equ CONFIG_COMPOUND_SEGMENT , 0 .equ CONFIG_COMPOUND_SEGMENT , 1
.equ CONFIG_EXT_REFS , 1 .equ CONFIG_EXT_REFS , 1
.equ CONFIG_GLOBAL_MOTION , 1 .equ CONFIG_GLOBAL_MOTION , 1
.equ CONFIG_NEW_QUANT , 0 .equ CONFIG_NEW_QUANT , 0
.equ CONFIG_SUPERTX , 0 .equ CONFIG_SUPERTX , 0
.equ CONFIG_ANS , 0 .equ CONFIG_ANS , 0
.equ CONFIG_EC_MULTISYMBOL , 1
.equ CONFIG_NEW_TOKENSET , 1
.equ CONFIG_LOOP_RESTORATION , 0 .equ CONFIG_LOOP_RESTORATION , 0
.equ CONFIG_EXT_PARTITION , 0 .equ CONFIG_EXT_PARTITION , 0
.equ CONFIG_EXT_PARTITION_TYPES , 0 .equ CONFIG_EXT_PARTITION_TYPES , 0
@ -108,20 +103,18 @@
.equ CONFIG_NCOBMC , 0 .equ CONFIG_NCOBMC , 0
.equ CONFIG_WARPED_MOTION , 1 .equ CONFIG_WARPED_MOTION , 1
.equ CONFIG_Q_ADAPT_PROBS , 0 .equ CONFIG_Q_ADAPT_PROBS , 0
.equ CONFIG_SUBFRAME_PROB_UPDATE , 0
.equ CONFIG_BITSTREAM_DEBUG , 0 .equ CONFIG_BITSTREAM_DEBUG , 0
.equ CONFIG_ALT_INTRA , 1 .equ CONFIG_ALT_INTRA , 1
.equ CONFIG_PALETTE , 1 .equ CONFIG_PALETTE , 1
.equ CONFIG_PALETTE_DELTA_ENCODING , 0 .equ CONFIG_PALETTE_DELTA_ENCODING , 0
.equ CONFIG_DAALA_EC , 1
.equ CONFIG_RAWBITS , 0 .equ CONFIG_RAWBITS , 0
.equ CONFIG_EC_SMALLMUL , 0 .equ CONFIG_EC_SMALLMUL , 1
.equ CONFIG_PVQ , 0 .equ CONFIG_PVQ , 0
.equ CONFIG_CFL , 0 .equ CONFIG_CFL , 0
.equ CONFIG_XIPHRC , 0 .equ CONFIG_XIPHRC , 0
.equ CONFIG_CB4X4 , 1 .equ CONFIG_CB4X4 , 1
.equ CONFIG_CHROMA_2X2 , 0 .equ CONFIG_CHROMA_2X2 , 0
.equ CONFIG_CHROMA_SUB8X8 , 0 .equ CONFIG_CHROMA_SUB8X8 , 1
.equ CONFIG_FRAME_SIZE , 0 .equ CONFIG_FRAME_SIZE , 0
.equ CONFIG_DELTA_Q , 1 .equ CONFIG_DELTA_Q , 1
.equ CONFIG_EXT_DELTA_Q , 1 .equ CONFIG_EXT_DELTA_Q , 1
@ -129,10 +122,10 @@
.equ CONFIG_FILTER_7BIT , 1 .equ CONFIG_FILTER_7BIT , 1
.equ CONFIG_PARALLEL_DEBLOCKING , 0 .equ CONFIG_PARALLEL_DEBLOCKING , 0
.equ CONFIG_PARALLEL_DEBLOCKING_15TAP , 0 .equ CONFIG_PARALLEL_DEBLOCKING_15TAP , 0
.equ CONFIG_LOOPFILTERING_ACROSS_TILES , 0 .equ CONFIG_LOOPFILTERING_ACROSS_TILES , 1
.equ CONFIG_TILE_GROUPS , 1 .equ CONFIG_TILE_GROUPS , 1
.equ CONFIG_EC_ADAPT , 1 .equ CONFIG_EC_ADAPT , 1
.equ CONFIG_TEMPMV_SIGNALING , 0 .equ CONFIG_TEMPMV_SIGNALING , 1
.equ CONFIG_RD_DEBUG , 0 .equ CONFIG_RD_DEBUG , 0
.equ CONFIG_REFERENCE_BUFFER , 1 .equ CONFIG_REFERENCE_BUFFER , 1
.equ CONFIG_COEF_INTERLEAVE , 0 .equ CONFIG_COEF_INTERLEAVE , 0
@ -150,6 +143,8 @@
.equ CONFIG_NEW_MULTISYMBOL , 0 .equ CONFIG_NEW_MULTISYMBOL , 0
.equ CONFIG_COMPOUND_SINGLEREF , 0 .equ CONFIG_COMPOUND_SINGLEREF , 0
.equ CONFIG_AOM_QM , 0 .equ CONFIG_AOM_QM , 0
.equ CONFIG_LOWDELAY_COMPOUND , 0 .equ CONFIG_ONE_SIDED_COMPOUND , 1
.equ CONFIG_SMOOTH_HV , 0
.equ CONFIG_VAR_REFS , 0
.equ CONFIG_ANALYZER , 0 .equ CONFIG_ANALYZER , 0
.section .note.GNU-stack,"",%progbits .section .note.GNU-stack,"",%progbits

39
media/libaom/config/linux/arm/aom_config.h
View file

@ -15,8 +15,6 @@
#define ARCH_MIPS 0 #define ARCH_MIPS 0
#define ARCH_X86 0 #define ARCH_X86 0
#define ARCH_X86_64 0 #define ARCH_X86_64 0
#define HAVE_EDSP 0
#define HAVE_MEDIA 1
#define HAVE_NEON 1 #define HAVE_NEON 1
#define HAVE_NEON_ASM 1 #define HAVE_NEON_ASM 1
#define HAVE_MIPS32 0 #define HAVE_MIPS32 0
@ -58,8 +56,6 @@
#define CONFIG_AV1_ENCODER 1 #define CONFIG_AV1_ENCODER 1
#define CONFIG_AV1_DECODER 1 #define CONFIG_AV1_DECODER 1
#define CONFIG_AV1 1 #define CONFIG_AV1 1
#define CONFIG_ENCODERS 1
#define CONFIG_DECODERS 1
#define CONFIG_STATIC_MSVCRT 0 #define CONFIG_STATIC_MSVCRT 0
#define CONFIG_SPATIAL_RESAMPLING 1 #define CONFIG_SPATIAL_RESAMPLING 1
#define CONFIG_REALTIME_ONLY 1 #define CONFIG_REALTIME_ONLY 1
@ -79,36 +75,35 @@
#define CONFIG_ENCODE_PERF_TESTS 0 #define CONFIG_ENCODE_PERF_TESTS 0
#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 #define CONFIG_COEFFICIENT_RANGE_CHECKING 0
#define CONFIG_LOWBITDEPTH 1 #define CONFIG_LOWBITDEPTH 1
#define CONFIG_HIGHBITDEPTH 0 #define CONFIG_HIGHBITDEPTH 1
#define CONFIG_EXPERIMENTAL 0 #define CONFIG_EXPERIMENTAL 0
#define CONFIG_SIZE_LIMIT 1 #define CONFIG_SIZE_LIMIT 1
#define CONFIG_FP_MB_STATS 0 #define CONFIG_FP_MB_STATS 0
#define CONFIG_CDEF 1 #define CONFIG_CDEF 1
#define CONFIG_VAR_TX 0 #define CONFIG_VAR_TX 1
#define CONFIG_RECT_TX 1 #define CONFIG_RECT_TX 1
#define CONFIG_REF_MV 1 #define CONFIG_RECT_TX_EXT 0
#define CONFIG_TPL_MV 0 #define CONFIG_TPL_MV 0
#define CONFIG_DUAL_FILTER 1 #define CONFIG_DUAL_FILTER 1
#define CONFIG_CONVOLVE_ROUND 0 #define CONFIG_CONVOLVE_ROUND 0
#define CONFIG_COMPOUND_ROUND 0 #define CONFIG_COMPOUND_ROUND 0
#define CONFIG_EXT_TX 0 #define CONFIG_EXT_TX 1
#define CONFIG_DPCM_INTRA 0
#define CONFIG_TX64X64 0 #define CONFIG_TX64X64 0
#define CONFIG_SUB8X8_MC 0
#define CONFIG_EXT_INTRA 1 #define CONFIG_EXT_INTRA 1
#define CONFIG_INTRA_INTERP 0 #define CONFIG_INTRA_INTERP 0
#define CONFIG_FILTER_INTRA 0 #define CONFIG_FILTER_INTRA 0
#define CONFIG_INTRA_EDGE 0
#define CONFIG_INTRABC 0 #define CONFIG_INTRABC 0
#define CONFIG_EXT_INTER 0 #define CONFIG_EXT_INTER 1
#define CONFIG_INTERINTRA 0 #define CONFIG_INTERINTRA 1
#define CONFIG_WEDGE 0 #define CONFIG_WEDGE 1
#define CONFIG_COMPOUND_SEGMENT 0 #define CONFIG_COMPOUND_SEGMENT 1
#define CONFIG_EXT_REFS 1 #define CONFIG_EXT_REFS 1
#define CONFIG_GLOBAL_MOTION 1 #define CONFIG_GLOBAL_MOTION 1
#define CONFIG_NEW_QUANT 0 #define CONFIG_NEW_QUANT 0
#define CONFIG_SUPERTX 0 #define CONFIG_SUPERTX 0
#define CONFIG_ANS 0 #define CONFIG_ANS 0
#define CONFIG_EC_MULTISYMBOL 1
#define CONFIG_NEW_TOKENSET 1
#define CONFIG_LOOP_RESTORATION 0 #define CONFIG_LOOP_RESTORATION 0
#define CONFIG_EXT_PARTITION 0 #define CONFIG_EXT_PARTITION 0
#define CONFIG_EXT_PARTITION_TYPES 0 #define CONFIG_EXT_PARTITION_TYPES 0
@ -118,20 +113,18 @@
#define CONFIG_NCOBMC 0 #define CONFIG_NCOBMC 0
#define CONFIG_WARPED_MOTION 1 #define CONFIG_WARPED_MOTION 1
#define CONFIG_Q_ADAPT_PROBS 0 #define CONFIG_Q_ADAPT_PROBS 0
#define CONFIG_SUBFRAME_PROB_UPDATE 0
#define CONFIG_BITSTREAM_DEBUG 0 #define CONFIG_BITSTREAM_DEBUG 0
#define CONFIG_ALT_INTRA 1 #define CONFIG_ALT_INTRA 1
#define CONFIG_PALETTE 1 #define CONFIG_PALETTE 1
#define CONFIG_PALETTE_DELTA_ENCODING 0 #define CONFIG_PALETTE_DELTA_ENCODING 0
#define CONFIG_DAALA_EC 1
#define CONFIG_RAWBITS 0 #define CONFIG_RAWBITS 0
#define CONFIG_EC_SMALLMUL 0 #define CONFIG_EC_SMALLMUL 1
#define CONFIG_PVQ 0 #define CONFIG_PVQ 0
#define CONFIG_CFL 0 #define CONFIG_CFL 0
#define CONFIG_XIPHRC 0 #define CONFIG_XIPHRC 0
#define CONFIG_CB4X4 1 #define CONFIG_CB4X4 1
#define CONFIG_CHROMA_2X2 0 #define CONFIG_CHROMA_2X2 0
#define CONFIG_CHROMA_SUB8X8 0 #define CONFIG_CHROMA_SUB8X8 1
#define CONFIG_FRAME_SIZE 0 #define CONFIG_FRAME_SIZE 0
#define CONFIG_DELTA_Q 1 #define CONFIG_DELTA_Q 1
#define CONFIG_EXT_DELTA_Q 1 #define CONFIG_EXT_DELTA_Q 1
@ -139,10 +132,10 @@
#define CONFIG_FILTER_7BIT 1 #define CONFIG_FILTER_7BIT 1
#define CONFIG_PARALLEL_DEBLOCKING 0 #define CONFIG_PARALLEL_DEBLOCKING 0
#define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 #define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
#define CONFIG_LOOPFILTERING_ACROSS_TILES 0 #define CONFIG_LOOPFILTERING_ACROSS_TILES 1
#define CONFIG_TILE_GROUPS 1 #define CONFIG_TILE_GROUPS 1
#define CONFIG_EC_ADAPT 1 #define CONFIG_EC_ADAPT 1
#define CONFIG_TEMPMV_SIGNALING 0 #define CONFIG_TEMPMV_SIGNALING 1
#define CONFIG_RD_DEBUG 0 #define CONFIG_RD_DEBUG 0
#define CONFIG_REFERENCE_BUFFER 1 #define CONFIG_REFERENCE_BUFFER 1
#define CONFIG_COEF_INTERLEAVE 0 #define CONFIG_COEF_INTERLEAVE 0
@ -160,7 +153,9 @@
#define CONFIG_NEW_MULTISYMBOL 0 #define CONFIG_NEW_MULTISYMBOL 0
#define CONFIG_COMPOUND_SINGLEREF 0 #define CONFIG_COMPOUND_SINGLEREF 0
#define CONFIG_AOM_QM 0 #define CONFIG_AOM_QM 0
#define CONFIG_LOWDELAY_COMPOUND 0 #define CONFIG_ONE_SIDED_COMPOUND 1
#define CONFIG_SMOOTH_HV 0
#define CONFIG_VAR_REFS 0
#define CONFIG_ANALYZER 0 #define CONFIG_ANALYZER 0
#define DECODE_WIDTH_LIMIT 8192 #define DECODE_WIDTH_LIMIT 8192
#define DECODE_HEIGHT_LIMIT 4608 #define DECODE_HEIGHT_LIMIT 4608

1579
media/libaom/config/linux/arm/aom_dsp_rtcd.h
View file

File diff suppressed because it is too large Load diff

202
media/libaom/config/linux/arm/av1_rtcd.h
View file

@ -54,10 +54,6 @@ RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int
int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
#define av1_block_error av1_block_error_c #define av1_block_error av1_block_error_c
int64_t av1_block_error_fp_c(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
int64_t av1_block_error_fp_neon(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
RTCD_EXTERN int64_t (*av1_block_error_fp)(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
#define av1_convolve_horiz av1_convolve_horiz_c #define av1_convolve_horiz av1_convolve_horiz_c
@ -68,8 +64,7 @@ int av1_diamond_search_sad_c(struct macroblock *x, const struct search_site_conf
#define av1_diamond_search_sad av1_diamond_search_sad_c #define av1_diamond_search_sad av1_diamond_search_sad_c
void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fdct8x8_quant_neon(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_fdct8x8_quant av1_fdct8x8_quant_c
RTCD_EXTERN void (*av1_fdct8x8_quant)(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_fht16x16 av1_fht16x16_c #define av1_fht16x16 av1_fht16x16_c
@ -122,9 +117,156 @@ int av1_full_search_sad_c(const struct macroblock *x, const struct mv *ref_mv, i
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type); void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type);
#define av1_fwd_idtx av1_fwd_idtx_c #define av1_fwd_idtx av1_fwd_idtx_c
void av1_fwd_txfm2d_16x16_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_16x16 av1_fwd_txfm2d_16x16_c
void av1_fwd_txfm2d_32x32_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_32x32 av1_fwd_txfm2d_32x32_c
void av1_fwd_txfm2d_4x4_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_4x4 av1_fwd_txfm2d_4x4_c
void av1_fwd_txfm2d_64x64_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_64x64 av1_fwd_txfm2d_64x64_c
void av1_fwd_txfm2d_8x8_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
#define av1_fwd_txfm2d_8x8 av1_fwd_txfm2d_8x8_c
void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_fwht4x4 av1_fwht4x4_c #define av1_fwht4x4 av1_fwht4x4_c
int64_t av1_highbd_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
#define av1_highbd_block_error av1_highbd_block_error_c
void av1_highbd_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8 av1_highbd_convolve8_c
void av1_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg av1_highbd_convolve8_avg_c
void av1_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_horiz av1_highbd_convolve8_avg_horiz_c
void av1_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_vert av1_highbd_convolve8_avg_vert_c
void av1_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_horiz av1_highbd_convolve8_horiz_c
void av1_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_vert av1_highbd_convolve8_vert_c
void av1_highbd_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_avg av1_highbd_convolve_avg_c
void av1_highbd_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_copy av1_highbd_convolve_copy_c
void av1_highbd_convolve_horiz_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
#define av1_highbd_convolve_horiz av1_highbd_convolve_horiz_c
void av1_highbd_convolve_init_c(void);
#define av1_highbd_convolve_init av1_highbd_convolve_init_c
void av1_highbd_convolve_vert_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
#define av1_highbd_convolve_vert av1_highbd_convolve_vert_c
void av1_highbd_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x16 av1_highbd_fht16x16_c
void av1_highbd_fht16x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x32 av1_highbd_fht16x32_c
void av1_highbd_fht16x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x4 av1_highbd_fht16x4_c
void av1_highbd_fht16x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x8 av1_highbd_fht16x8_c
void av1_highbd_fht32x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x16 av1_highbd_fht32x16_c
void av1_highbd_fht32x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x32 av1_highbd_fht32x32_c
void av1_highbd_fht32x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x8 av1_highbd_fht32x8_c
void av1_highbd_fht4x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x16 av1_highbd_fht4x16_c
void av1_highbd_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x4 av1_highbd_fht4x4_c
void av1_highbd_fht4x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x8 av1_highbd_fht4x8_c
void av1_highbd_fht8x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x16 av1_highbd_fht8x16_c
void av1_highbd_fht8x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x32 av1_highbd_fht8x32_c
void av1_highbd_fht8x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x4 av1_highbd_fht8x4_c
void av1_highbd_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x8 av1_highbd_fht8x8_c
void av1_highbd_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_highbd_fwht4x4 av1_highbd_fwht4x4_c
void av1_highbd_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd);
#define av1_highbd_iht16x16_256_add av1_highbd_iht16x16_256_add_c
void av1_highbd_iht16x32_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x32_512_add av1_highbd_iht16x32_512_add_c
void av1_highbd_iht16x4_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x4_64_add av1_highbd_iht16x4_64_add_c
void av1_highbd_iht16x8_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x8_128_add av1_highbd_iht16x8_128_add_c
void av1_highbd_iht32x16_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x16_512_add av1_highbd_iht32x16_512_add_c
void av1_highbd_iht32x8_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x8_256_add av1_highbd_iht32x8_256_add_c
void av1_highbd_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x16_64_add av1_highbd_iht4x16_64_add_c
void av1_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x4_16_add av1_highbd_iht4x4_16_add_c
void av1_highbd_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x8_32_add av1_highbd_iht4x8_32_add_c
void av1_highbd_iht8x16_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x16_128_add av1_highbd_iht8x16_128_add_c
void av1_highbd_iht8x32_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x32_256_add av1_highbd_iht8x32_256_add_c
void av1_highbd_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x4_32_add av1_highbd_iht8x4_32_add_c
void av1_highbd_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x8_64_add av1_highbd_iht8x8_64_add_c
void av1_highbd_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_b av1_highbd_quantize_b_c
void av1_highbd_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_fp av1_highbd_quantize_fp_c
void av1_highbd_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_highbd_temporal_filter_apply av1_highbd_temporal_filter_apply_c
void av1_highbd_warp_affine_c(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
#define av1_highbd_warp_affine av1_highbd_warp_affine_c
void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
#define av1_iht16x16_256_add av1_iht16x16_256_add_c #define av1_iht16x16_256_add av1_iht16x16_256_add_c
@ -150,8 +292,7 @@ void av1_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_strid
#define av1_iht4x16_64_add av1_iht4x16_64_add_c #define av1_iht4x16_64_add av1_iht4x16_64_add_c
void av1_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
void av1_iht4x4_16_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); #define av1_iht4x4_16_add av1_iht4x4_16_add_c
RTCD_EXTERN void (*av1_iht4x4_16_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
void av1_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
#define av1_iht4x8_32_add av1_iht4x8_32_add_c #define av1_iht4x8_32_add av1_iht4x8_32_add_c
@ -166,8 +307,22 @@ void av1_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride
#define av1_iht8x4_32_add av1_iht8x4_32_add_c #define av1_iht8x4_32_add av1_iht8x4_32_add_c
void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
void av1_iht8x8_64_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); #define av1_iht8x8_64_add av1_iht8x8_64_add_c
RTCD_EXTERN void (*av1_iht8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
void av1_inv_txfm2d_add_16x16_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_16x16 av1_inv_txfm2d_add_16x16_c
void av1_inv_txfm2d_add_32x32_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_32x32 av1_inv_txfm2d_add_32x32_c
void av1_inv_txfm2d_add_4x4_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_4x4 av1_inv_txfm2d_add_4x4_c
void av1_inv_txfm2d_add_64x64_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_64x64 av1_inv_txfm2d_add_64x64_c
void av1_inv_txfm2d_add_8x8_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_8x8 av1_inv_txfm2d_add_8x8_c
void av1_lowbd_convolve_init_c(void); void av1_lowbd_convolve_init_c(void);
#define av1_lowbd_convolve_init av1_lowbd_convolve_init_c #define av1_lowbd_convolve_init av1_lowbd_convolve_init_c
@ -176,8 +331,7 @@ void av1_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_b
#define av1_quantize_b av1_quantize_b_c #define av1_quantize_b av1_quantize_b_c
void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_neon(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_quantize_fp av1_quantize_fp_c
RTCD_EXTERN void (*av1_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
#define av1_quantize_fp_32x32 av1_quantize_fp_32x32_c #define av1_quantize_fp_32x32 av1_quantize_fp_32x32_c
@ -185,9 +339,21 @@ void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int
void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_temporal_filter_apply av1_temporal_filter_apply_c #define av1_temporal_filter_apply av1_temporal_filter_apply_c
void av1_warp_affine_c(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_c(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
#define av1_warp_affine av1_warp_affine_c #define av1_warp_affine av1_warp_affine_c
void av1_wedge_compute_delta_squares_c(int16_t *d, const int16_t *a, const int16_t *b, int N);
#define av1_wedge_compute_delta_squares av1_wedge_compute_delta_squares_c
int av1_wedge_sign_from_residuals_c(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
#define av1_wedge_sign_from_residuals av1_wedge_sign_from_residuals_c
uint64_t av1_wedge_sse_from_residuals_c(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
#define av1_wedge_sse_from_residuals av1_wedge_sse_from_residuals_c
double compute_cross_correlation_c(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
#define compute_cross_correlation compute_cross_correlation_c
void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
void copy_4x4_16bit_to_16bit_neon(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_neon(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
RTCD_EXTERN void (*copy_4x4_16bit_to_16bit)(uint16_t *dst, int dstride, const uint16_t *src, int sstride); RTCD_EXTERN void (*copy_4x4_16bit_to_16bit)(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
@ -244,16 +410,6 @@ static void setup_rtcd_internal(void)
if (flags & HAS_NEON) aom_clpf_hblock = aom_clpf_hblock_neon; if (flags & HAS_NEON) aom_clpf_hblock = aom_clpf_hblock_neon;
aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_c; aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_c;
if (flags & HAS_NEON) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_neon; if (flags & HAS_NEON) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_neon;
av1_block_error_fp = av1_block_error_fp_c;
if (flags & HAS_NEON) av1_block_error_fp = av1_block_error_fp_neon;
av1_fdct8x8_quant = av1_fdct8x8_quant_c;
if (flags & HAS_NEON) av1_fdct8x8_quant = av1_fdct8x8_quant_neon;
av1_iht4x4_16_add = av1_iht4x4_16_add_c;
if (flags & HAS_NEON) av1_iht4x4_16_add = av1_iht4x4_16_add_neon;
av1_iht8x8_64_add = av1_iht8x8_64_add_c;
if (flags & HAS_NEON) av1_iht8x8_64_add = av1_iht8x8_64_add_neon;
av1_quantize_fp = av1_quantize_fp_c;
if (flags & HAS_NEON) av1_quantize_fp = av1_quantize_fp_neon;
copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_c; copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_c;
if (flags & HAS_NEON) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_neon; if (flags & HAS_NEON) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_neon;
copy_4x4_16bit_to_8bit = copy_4x4_16bit_to_8bit_c; copy_4x4_16bit_to_8bit = copy_4x4_16bit_to_8bit_c;

39
media/libaom/config/linux/ia32/aom_config.asm
View file

@ -2,8 +2,6 @@
%define ARCH_MIPS 0 %define ARCH_MIPS 0
%define ARCH_X86 1 %define ARCH_X86 1
%define ARCH_X86_64 0 %define ARCH_X86_64 0
%define HAVE_EDSP 0
%define HAVE_MEDIA 0
%define HAVE_NEON 0 %define HAVE_NEON 0
%define HAVE_NEON_ASM 0 %define HAVE_NEON_ASM 0
%define HAVE_MIPS32 0 %define HAVE_MIPS32 0
@ -45,8 +43,6 @@
%define CONFIG_AV1_ENCODER 1 %define CONFIG_AV1_ENCODER 1
%define CONFIG_AV1_DECODER 1 %define CONFIG_AV1_DECODER 1
%define CONFIG_AV1 1 %define CONFIG_AV1 1
%define CONFIG_ENCODERS 1
%define CONFIG_DECODERS 1
%define CONFIG_STATIC_MSVCRT 0 %define CONFIG_STATIC_MSVCRT 0
%define CONFIG_SPATIAL_RESAMPLING 1 %define CONFIG_SPATIAL_RESAMPLING 1
%define CONFIG_REALTIME_ONLY 0 %define CONFIG_REALTIME_ONLY 0
@ -66,36 +62,35 @@
%define CONFIG_ENCODE_PERF_TESTS 0 %define CONFIG_ENCODE_PERF_TESTS 0
%define CONFIG_COEFFICIENT_RANGE_CHECKING 0 %define CONFIG_COEFFICIENT_RANGE_CHECKING 0
%define CONFIG_LOWBITDEPTH 1 %define CONFIG_LOWBITDEPTH 1
%define CONFIG_HIGHBITDEPTH 0 %define CONFIG_HIGHBITDEPTH 1
%define CONFIG_EXPERIMENTAL 0 %define CONFIG_EXPERIMENTAL 0
%define CONFIG_SIZE_LIMIT 1 %define CONFIG_SIZE_LIMIT 1
%define CONFIG_FP_MB_STATS 0 %define CONFIG_FP_MB_STATS 0
%define CONFIG_CDEF 1 %define CONFIG_CDEF 1
%define CONFIG_VAR_TX 0 %define CONFIG_VAR_TX 1
%define CONFIG_RECT_TX 1 %define CONFIG_RECT_TX 1
%define CONFIG_REF_MV 1 %define CONFIG_RECT_TX_EXT 0
%define CONFIG_TPL_MV 0 %define CONFIG_TPL_MV 0
%define CONFIG_DUAL_FILTER 1 %define CONFIG_DUAL_FILTER 1
%define CONFIG_CONVOLVE_ROUND 0 %define CONFIG_CONVOLVE_ROUND 0
%define CONFIG_COMPOUND_ROUND 0 %define CONFIG_COMPOUND_ROUND 0
%define CONFIG_EXT_TX 0 %define CONFIG_EXT_TX 1
%define CONFIG_DPCM_INTRA 0
%define CONFIG_TX64X64 0 %define CONFIG_TX64X64 0
%define CONFIG_SUB8X8_MC 0
%define CONFIG_EXT_INTRA 1 %define CONFIG_EXT_INTRA 1
%define CONFIG_INTRA_INTERP 0 %define CONFIG_INTRA_INTERP 0
%define CONFIG_FILTER_INTRA 0 %define CONFIG_FILTER_INTRA 0
%define CONFIG_INTRA_EDGE 0
%define CONFIG_INTRABC 0 %define CONFIG_INTRABC 0
%define CONFIG_EXT_INTER 0 %define CONFIG_EXT_INTER 1
%define CONFIG_INTERINTRA 0 %define CONFIG_INTERINTRA 1
%define CONFIG_WEDGE 0 %define CONFIG_WEDGE 1
%define CONFIG_COMPOUND_SEGMENT 0 %define CONFIG_COMPOUND_SEGMENT 1
%define CONFIG_EXT_REFS 1 %define CONFIG_EXT_REFS 1
%define CONFIG_GLOBAL_MOTION 1 %define CONFIG_GLOBAL_MOTION 1
%define CONFIG_NEW_QUANT 0 %define CONFIG_NEW_QUANT 0
%define CONFIG_SUPERTX 0 %define CONFIG_SUPERTX 0
%define CONFIG_ANS 0 %define CONFIG_ANS 0
%define CONFIG_EC_MULTISYMBOL 1
%define CONFIG_NEW_TOKENSET 1
%define CONFIG_LOOP_RESTORATION 0 %define CONFIG_LOOP_RESTORATION 0
%define CONFIG_EXT_PARTITION 0 %define CONFIG_EXT_PARTITION 0
%define CONFIG_EXT_PARTITION_TYPES 0 %define CONFIG_EXT_PARTITION_TYPES 0
@ -105,20 +100,18 @@
%define CONFIG_NCOBMC 0 %define CONFIG_NCOBMC 0
%define CONFIG_WARPED_MOTION 1 %define CONFIG_WARPED_MOTION 1
%define CONFIG_Q_ADAPT_PROBS 0 %define CONFIG_Q_ADAPT_PROBS 0
%define CONFIG_SUBFRAME_PROB_UPDATE 0
%define CONFIG_BITSTREAM_DEBUG 0 %define CONFIG_BITSTREAM_DEBUG 0
%define CONFIG_ALT_INTRA 1 %define CONFIG_ALT_INTRA 1
%define CONFIG_PALETTE 1 %define CONFIG_PALETTE 1
%define CONFIG_PALETTE_DELTA_ENCODING 0 %define CONFIG_PALETTE_DELTA_ENCODING 0
%define CONFIG_DAALA_EC 1
%define CONFIG_RAWBITS 0 %define CONFIG_RAWBITS 0
%define CONFIG_EC_SMALLMUL 0 %define CONFIG_EC_SMALLMUL 1
%define CONFIG_PVQ 0 %define CONFIG_PVQ 0
%define CONFIG_CFL 0 %define CONFIG_CFL 0
%define CONFIG_XIPHRC 0 %define CONFIG_XIPHRC 0
%define CONFIG_CB4X4 1 %define CONFIG_CB4X4 1
%define CONFIG_CHROMA_2X2 0 %define CONFIG_CHROMA_2X2 0
%define CONFIG_CHROMA_SUB8X8 0 %define CONFIG_CHROMA_SUB8X8 1
%define CONFIG_FRAME_SIZE 0 %define CONFIG_FRAME_SIZE 0
%define CONFIG_DELTA_Q 1 %define CONFIG_DELTA_Q 1
%define CONFIG_EXT_DELTA_Q 1 %define CONFIG_EXT_DELTA_Q 1
@ -126,10 +119,10 @@
%define CONFIG_FILTER_7BIT 1 %define CONFIG_FILTER_7BIT 1
%define CONFIG_PARALLEL_DEBLOCKING 0 %define CONFIG_PARALLEL_DEBLOCKING 0
%define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 %define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
%define CONFIG_LOOPFILTERING_ACROSS_TILES 0 %define CONFIG_LOOPFILTERING_ACROSS_TILES 1
%define CONFIG_TILE_GROUPS 1 %define CONFIG_TILE_GROUPS 1
%define CONFIG_EC_ADAPT 1 %define CONFIG_EC_ADAPT 1
%define CONFIG_TEMPMV_SIGNALING 0 %define CONFIG_TEMPMV_SIGNALING 1
%define CONFIG_RD_DEBUG 0 %define CONFIG_RD_DEBUG 0
%define CONFIG_REFERENCE_BUFFER 1 %define CONFIG_REFERENCE_BUFFER 1
%define CONFIG_COEF_INTERLEAVE 0 %define CONFIG_COEF_INTERLEAVE 0
@ -147,5 +140,7 @@
%define CONFIG_NEW_MULTISYMBOL 0 %define CONFIG_NEW_MULTISYMBOL 0
%define CONFIG_COMPOUND_SINGLEREF 0 %define CONFIG_COMPOUND_SINGLEREF 0
%define CONFIG_AOM_QM 0 %define CONFIG_AOM_QM 0
%define CONFIG_LOWDELAY_COMPOUND 0 %define CONFIG_ONE_SIDED_COMPOUND 1
%define CONFIG_SMOOTH_HV 0
%define CONFIG_VAR_REFS 0
%define CONFIG_ANALYZER 0 %define CONFIG_ANALYZER 0

39
media/libaom/config/linux/ia32/aom_config.h
View file

@ -15,8 +15,6 @@
#define ARCH_MIPS 0 #define ARCH_MIPS 0
#define ARCH_X86 1 #define ARCH_X86 1
#define ARCH_X86_64 0 #define ARCH_X86_64 0
#define HAVE_EDSP 0
#define HAVE_MEDIA 0
#define HAVE_NEON 0 #define HAVE_NEON 0
#define HAVE_NEON_ASM 0 #define HAVE_NEON_ASM 0
#define HAVE_MIPS32 0 #define HAVE_MIPS32 0
@ -58,8 +56,6 @@
#define CONFIG_AV1_ENCODER 1 #define CONFIG_AV1_ENCODER 1
#define CONFIG_AV1_DECODER 1 #define CONFIG_AV1_DECODER 1
#define CONFIG_AV1 1 #define CONFIG_AV1 1
#define CONFIG_ENCODERS 1
#define CONFIG_DECODERS 1
#define CONFIG_STATIC_MSVCRT 0 #define CONFIG_STATIC_MSVCRT 0
#define CONFIG_SPATIAL_RESAMPLING 1 #define CONFIG_SPATIAL_RESAMPLING 1
#define CONFIG_REALTIME_ONLY 0 #define CONFIG_REALTIME_ONLY 0
@ -79,36 +75,35 @@
#define CONFIG_ENCODE_PERF_TESTS 0 #define CONFIG_ENCODE_PERF_TESTS 0
#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 #define CONFIG_COEFFICIENT_RANGE_CHECKING 0
#define CONFIG_LOWBITDEPTH 1 #define CONFIG_LOWBITDEPTH 1
#define CONFIG_HIGHBITDEPTH 0 #define CONFIG_HIGHBITDEPTH 1
#define CONFIG_EXPERIMENTAL 0 #define CONFIG_EXPERIMENTAL 0
#define CONFIG_SIZE_LIMIT 1 #define CONFIG_SIZE_LIMIT 1
#define CONFIG_FP_MB_STATS 0 #define CONFIG_FP_MB_STATS 0
#define CONFIG_CDEF 1 #define CONFIG_CDEF 1
#define CONFIG_VAR_TX 0 #define CONFIG_VAR_TX 1
#define CONFIG_RECT_TX 1 #define CONFIG_RECT_TX 1
#define CONFIG_REF_MV 1 #define CONFIG_RECT_TX_EXT 0
#define CONFIG_TPL_MV 0 #define CONFIG_TPL_MV 0
#define CONFIG_DUAL_FILTER 1 #define CONFIG_DUAL_FILTER 1
#define CONFIG_CONVOLVE_ROUND 0 #define CONFIG_CONVOLVE_ROUND 0
#define CONFIG_COMPOUND_ROUND 0 #define CONFIG_COMPOUND_ROUND 0
#define CONFIG_EXT_TX 0 #define CONFIG_EXT_TX 1
#define CONFIG_DPCM_INTRA 0
#define CONFIG_TX64X64 0 #define CONFIG_TX64X64 0
#define CONFIG_SUB8X8_MC 0
#define CONFIG_EXT_INTRA 1 #define CONFIG_EXT_INTRA 1
#define CONFIG_INTRA_INTERP 0 #define CONFIG_INTRA_INTERP 0
#define CONFIG_FILTER_INTRA 0 #define CONFIG_FILTER_INTRA 0
#define CONFIG_INTRA_EDGE 0
#define CONFIG_INTRABC 0 #define CONFIG_INTRABC 0
#define CONFIG_EXT_INTER 0 #define CONFIG_EXT_INTER 1
#define CONFIG_INTERINTRA 0 #define CONFIG_INTERINTRA 1
#define CONFIG_WEDGE 0 #define CONFIG_WEDGE 1
#define CONFIG_COMPOUND_SEGMENT 0 #define CONFIG_COMPOUND_SEGMENT 1
#define CONFIG_EXT_REFS 1 #define CONFIG_EXT_REFS 1
#define CONFIG_GLOBAL_MOTION 1 #define CONFIG_GLOBAL_MOTION 1
#define CONFIG_NEW_QUANT 0 #define CONFIG_NEW_QUANT 0
#define CONFIG_SUPERTX 0 #define CONFIG_SUPERTX 0
#define CONFIG_ANS 0 #define CONFIG_ANS 0
#define CONFIG_EC_MULTISYMBOL 1
#define CONFIG_NEW_TOKENSET 1
#define CONFIG_LOOP_RESTORATION 0 #define CONFIG_LOOP_RESTORATION 0
#define CONFIG_EXT_PARTITION 0 #define CONFIG_EXT_PARTITION 0
#define CONFIG_EXT_PARTITION_TYPES 0 #define CONFIG_EXT_PARTITION_TYPES 0
@ -118,20 +113,18 @@
#define CONFIG_NCOBMC 0 #define CONFIG_NCOBMC 0
#define CONFIG_WARPED_MOTION 1 #define CONFIG_WARPED_MOTION 1
#define CONFIG_Q_ADAPT_PROBS 0 #define CONFIG_Q_ADAPT_PROBS 0
#define CONFIG_SUBFRAME_PROB_UPDATE 0
#define CONFIG_BITSTREAM_DEBUG 0 #define CONFIG_BITSTREAM_DEBUG 0
#define CONFIG_ALT_INTRA 1 #define CONFIG_ALT_INTRA 1
#define CONFIG_PALETTE 1 #define CONFIG_PALETTE 1
#define CONFIG_PALETTE_DELTA_ENCODING 0 #define CONFIG_PALETTE_DELTA_ENCODING 0
#define CONFIG_DAALA_EC 1
#define CONFIG_RAWBITS 0 #define CONFIG_RAWBITS 0
#define CONFIG_EC_SMALLMUL 0 #define CONFIG_EC_SMALLMUL 1
#define CONFIG_PVQ 0 #define CONFIG_PVQ 0
#define CONFIG_CFL 0 #define CONFIG_CFL 0
#define CONFIG_XIPHRC 0 #define CONFIG_XIPHRC 0
#define CONFIG_CB4X4 1 #define CONFIG_CB4X4 1
#define CONFIG_CHROMA_2X2 0 #define CONFIG_CHROMA_2X2 0
#define CONFIG_CHROMA_SUB8X8 0 #define CONFIG_CHROMA_SUB8X8 1
#define CONFIG_FRAME_SIZE 0 #define CONFIG_FRAME_SIZE 0
#define CONFIG_DELTA_Q 1 #define CONFIG_DELTA_Q 1
#define CONFIG_EXT_DELTA_Q 1 #define CONFIG_EXT_DELTA_Q 1
@ -139,10 +132,10 @@
#define CONFIG_FILTER_7BIT 1 #define CONFIG_FILTER_7BIT 1
#define CONFIG_PARALLEL_DEBLOCKING 0 #define CONFIG_PARALLEL_DEBLOCKING 0
#define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 #define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
#define CONFIG_LOOPFILTERING_ACROSS_TILES 0 #define CONFIG_LOOPFILTERING_ACROSS_TILES 1
#define CONFIG_TILE_GROUPS 1 #define CONFIG_TILE_GROUPS 1
#define CONFIG_EC_ADAPT 1 #define CONFIG_EC_ADAPT 1
#define CONFIG_TEMPMV_SIGNALING 0 #define CONFIG_TEMPMV_SIGNALING 1
#define CONFIG_RD_DEBUG 0 #define CONFIG_RD_DEBUG 0
#define CONFIG_REFERENCE_BUFFER 1 #define CONFIG_REFERENCE_BUFFER 1
#define CONFIG_COEF_INTERLEAVE 0 #define CONFIG_COEF_INTERLEAVE 0
@ -160,7 +153,9 @@
#define CONFIG_NEW_MULTISYMBOL 0 #define CONFIG_NEW_MULTISYMBOL 0
#define CONFIG_COMPOUND_SINGLEREF 0 #define CONFIG_COMPOUND_SINGLEREF 0
#define CONFIG_AOM_QM 0 #define CONFIG_AOM_QM 0
#define CONFIG_LOWDELAY_COMPOUND 0 #define CONFIG_ONE_SIDED_COMPOUND 1
#define CONFIG_SMOOTH_HV 0
#define CONFIG_VAR_REFS 0
#define CONFIG_ANALYZER 0 #define CONFIG_ANALYZER 0
#define DECODE_WIDTH_LIMIT 8192 #define DECODE_WIDTH_LIMIT 8192
#define DECODE_HEIGHT_LIMIT 4608 #define DECODE_HEIGHT_LIMIT 4608

2507
media/libaom/config/linux/ia32/aom_dsp_rtcd.h
View file

File diff suppressed because it is too large Load diff

262
media/libaom/config/linux/ia32/av1_rtcd.h
View file

@ -60,13 +60,7 @@ void aom_clpf_hblock_hbd_sse4_1(uint16_t *dst, const uint16_t *src, int dstride,
RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd); RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd);
int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); #define av1_block_error av1_block_error_c
int64_t av1_block_error_avx2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
RTCD_EXTERN int64_t (*av1_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_fp_c(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
int64_t av1_block_error_fp_sse2(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
RTCD_EXTERN int64_t (*av1_block_error_fp)(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
@ -80,9 +74,7 @@ int av1_diamond_search_sad_c(struct macroblock *x, const struct search_site_conf
#define av1_diamond_search_sad av1_diamond_search_sad_c #define av1_diamond_search_sad av1_diamond_search_sad_c
void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fdct8x8_quant_sse2(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_fdct8x8_quant av1_fdct8x8_quant_c
void av1_fdct8x8_quant_ssse3(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
RTCD_EXTERN void (*av1_fdct8x8_quant)(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type);
@ -149,9 +141,168 @@ RTCD_EXTERN int (*av1_full_search_sad)(const struct macroblock *x, const struct
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type); void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type);
#define av1_fwd_idtx av1_fwd_idtx_c #define av1_fwd_idtx av1_fwd_idtx_c
void av1_fwd_txfm2d_16x16_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_16x16_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_16x16)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_32x32)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_4x4)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_64x64)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_8x8)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_fwht4x4 av1_fwht4x4_c #define av1_fwht4x4 av1_fwht4x4_c
int64_t av1_highbd_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
int64_t av1_highbd_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
RTCD_EXTERN int64_t (*av1_highbd_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
void av1_highbd_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8 av1_highbd_convolve8_c
void av1_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg av1_highbd_convolve8_avg_c
void av1_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_horiz av1_highbd_convolve8_avg_horiz_c
void av1_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_vert av1_highbd_convolve8_avg_vert_c
void av1_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_horiz av1_highbd_convolve8_horiz_c
void av1_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_vert av1_highbd_convolve8_vert_c
void av1_highbd_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_avg av1_highbd_convolve_avg_c
void av1_highbd_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_copy av1_highbd_convolve_copy_c
void av1_highbd_convolve_horiz_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_horiz_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_horiz)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_init_c(void);
void av1_highbd_convolve_init_sse4_1(void);
RTCD_EXTERN void (*av1_highbd_convolve_init)(void);
void av1_highbd_convolve_vert_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_vert_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_vert)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x16 av1_highbd_fht16x16_c
void av1_highbd_fht16x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x32 av1_highbd_fht16x32_c
void av1_highbd_fht16x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x4 av1_highbd_fht16x4_c
void av1_highbd_fht16x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x8 av1_highbd_fht16x8_c
void av1_highbd_fht32x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x16 av1_highbd_fht32x16_c
void av1_highbd_fht32x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x32 av1_highbd_fht32x32_c
void av1_highbd_fht32x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x8 av1_highbd_fht32x8_c
void av1_highbd_fht4x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x16 av1_highbd_fht4x16_c
void av1_highbd_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x4_sse4_1(const int16_t *input, tran_low_t *output, int stride, int tx_type);
RTCD_EXTERN void (*av1_highbd_fht4x4)(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x8 av1_highbd_fht4x8_c
void av1_highbd_fht8x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x16 av1_highbd_fht8x16_c
void av1_highbd_fht8x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x32 av1_highbd_fht8x32_c
void av1_highbd_fht8x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x4 av1_highbd_fht8x4_c
void av1_highbd_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x8 av1_highbd_fht8x8_c
void av1_highbd_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_highbd_fwht4x4 av1_highbd_fwht4x4_c
void av1_highbd_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd);
#define av1_highbd_iht16x16_256_add av1_highbd_iht16x16_256_add_c
void av1_highbd_iht16x32_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x32_512_add av1_highbd_iht16x32_512_add_c
void av1_highbd_iht16x4_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x4_64_add av1_highbd_iht16x4_64_add_c
void av1_highbd_iht16x8_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x8_128_add av1_highbd_iht16x8_128_add_c
void av1_highbd_iht32x16_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x16_512_add av1_highbd_iht32x16_512_add_c
void av1_highbd_iht32x8_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x8_256_add av1_highbd_iht32x8_256_add_c
void av1_highbd_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x16_64_add av1_highbd_iht4x16_64_add_c
void av1_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x4_16_add av1_highbd_iht4x4_16_add_c
void av1_highbd_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x8_32_add av1_highbd_iht4x8_32_add_c
void av1_highbd_iht8x16_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x16_128_add av1_highbd_iht8x16_128_add_c
void av1_highbd_iht8x32_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x32_256_add av1_highbd_iht8x32_256_add_c
void av1_highbd_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x4_32_add av1_highbd_iht8x4_32_add_c
void av1_highbd_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x8_64_add av1_highbd_iht8x8_64_add_c
void av1_highbd_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_b av1_highbd_quantize_b_c
void av1_highbd_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_quantize_fp_sse4_1(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
RTCD_EXTERN void (*av1_highbd_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_highbd_temporal_filter_apply av1_highbd_temporal_filter_apply_c
void av1_highbd_warp_affine_c(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_highbd_warp_affine_ssse3(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_highbd_warp_affine)(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
@ -204,6 +355,25 @@ void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride
void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
RTCD_EXTERN void (*av1_iht8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); RTCD_EXTERN void (*av1_iht8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
void av1_inv_txfm2d_add_16x16_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_16x16_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_16x16)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_avx2(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_32x32)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_4x4)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_64x64_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_64x64 av1_inv_txfm2d_add_64x64_c
void av1_inv_txfm2d_add_8x8_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_8x8_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_8x8)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_lowbd_convolve_init_c(void); void av1_lowbd_convolve_init_c(void);
void av1_lowbd_convolve_init_ssse3(void); void av1_lowbd_convolve_init_ssse3(void);
RTCD_EXTERN void (*av1_lowbd_convolve_init)(void); RTCD_EXTERN void (*av1_lowbd_convolve_init)(void);
@ -222,9 +392,26 @@ void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *
void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
RTCD_EXTERN void (*av1_temporal_filter_apply)(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); RTCD_EXTERN void (*av1_temporal_filter_apply)(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
void av1_warp_affine_c(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_c(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_warp_affine_sse2(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_sse2(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_warp_affine)(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_ssse3(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_warp_affine)(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_wedge_compute_delta_squares_c(int16_t *d, const int16_t *a, const int16_t *b, int N);
void av1_wedge_compute_delta_squares_sse2(int16_t *d, const int16_t *a, const int16_t *b, int N);
RTCD_EXTERN void (*av1_wedge_compute_delta_squares)(int16_t *d, const int16_t *a, const int16_t *b, int N);
int av1_wedge_sign_from_residuals_c(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
int av1_wedge_sign_from_residuals_sse2(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
RTCD_EXTERN int (*av1_wedge_sign_from_residuals)(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
uint64_t av1_wedge_sse_from_residuals_c(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
uint64_t av1_wedge_sse_from_residuals_sse2(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
RTCD_EXTERN uint64_t (*av1_wedge_sse_from_residuals)(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
double compute_cross_correlation_c(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
double compute_cross_correlation_sse4_1(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
RTCD_EXTERN double (*compute_cross_correlation)(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
@ -306,18 +493,10 @@ static void setup_rtcd_internal(void)
if (flags & HAS_SSE2) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2; if (flags & HAS_SSE2) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2;
if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3; if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3;
if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1; if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1;
av1_block_error = av1_block_error_c;
if (flags & HAS_SSE2) av1_block_error = av1_block_error_sse2;
if (flags & HAS_AVX2) av1_block_error = av1_block_error_avx2;
av1_block_error_fp = av1_block_error_fp_c;
if (flags & HAS_SSE2) av1_block_error_fp = av1_block_error_fp_sse2;
av1_convolve_horiz = av1_convolve_horiz_c; av1_convolve_horiz = av1_convolve_horiz_c;
if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3; if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3;
av1_convolve_vert = av1_convolve_vert_c; av1_convolve_vert = av1_convolve_vert_c;
if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3; if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3;
av1_fdct8x8_quant = av1_fdct8x8_quant_c;
if (flags & HAS_SSE2) av1_fdct8x8_quant = av1_fdct8x8_quant_sse2;
if (flags & HAS_SSSE3) av1_fdct8x8_quant = av1_fdct8x8_quant_ssse3;
av1_fht16x16 = av1_fht16x16_c; av1_fht16x16 = av1_fht16x16_c;
if (flags & HAS_SSE2) av1_fht16x16 = av1_fht16x16_sse2; if (flags & HAS_SSE2) av1_fht16x16 = av1_fht16x16_sse2;
if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2; if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2;
@ -343,6 +522,30 @@ static void setup_rtcd_internal(void)
av1_full_search_sad = av1_full_search_sad_c; av1_full_search_sad = av1_full_search_sad_c;
if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3; if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3;
if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8; if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8;
av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_sse4_1;
av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_sse4_1;
av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_sse4_1;
av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_sse4_1;
av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_sse4_1;
av1_highbd_block_error = av1_highbd_block_error_c;
if (flags & HAS_SSE2) av1_highbd_block_error = av1_highbd_block_error_sse2;
av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_sse4_1;
av1_highbd_convolve_init = av1_highbd_convolve_init_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_init = av1_highbd_convolve_init_sse4_1;
av1_highbd_convolve_vert = av1_highbd_convolve_vert_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_vert = av1_highbd_convolve_vert_sse4_1;
av1_highbd_fht4x4 = av1_highbd_fht4x4_c;
if (flags & HAS_SSE4_1) av1_highbd_fht4x4 = av1_highbd_fht4x4_sse4_1;
av1_highbd_quantize_fp = av1_highbd_quantize_fp_c;
if (flags & HAS_SSE4_1) av1_highbd_quantize_fp = av1_highbd_quantize_fp_sse4_1;
av1_highbd_warp_affine = av1_highbd_warp_affine_c;
if (flags & HAS_SSSE3) av1_highbd_warp_affine = av1_highbd_warp_affine_ssse3;
av1_iht16x16_256_add = av1_iht16x16_256_add_c; av1_iht16x16_256_add = av1_iht16x16_256_add_c;
if (flags & HAS_SSE2) av1_iht16x16_256_add = av1_iht16x16_256_add_sse2; if (flags & HAS_SSE2) av1_iht16x16_256_add = av1_iht16x16_256_add_sse2;
if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2; if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2;
@ -362,6 +565,14 @@ static void setup_rtcd_internal(void)
if (flags & HAS_SSE2) av1_iht8x4_32_add = av1_iht8x4_32_add_sse2; if (flags & HAS_SSE2) av1_iht8x4_32_add = av1_iht8x4_32_add_sse2;
av1_iht8x8_64_add = av1_iht8x8_64_add_c; av1_iht8x8_64_add = av1_iht8x8_64_add_c;
if (flags & HAS_SSE2) av1_iht8x8_64_add = av1_iht8x8_64_add_sse2; if (flags & HAS_SSE2) av1_iht8x8_64_add = av1_iht8x8_64_add_sse2;
av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_sse4_1;
av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_c;
if (flags & HAS_AVX2) av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_avx2;
av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_sse4_1;
av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_sse4_1;
av1_lowbd_convolve_init = av1_lowbd_convolve_init_c; av1_lowbd_convolve_init = av1_lowbd_convolve_init_c;
if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3; if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3;
av1_quantize_fp = av1_quantize_fp_c; av1_quantize_fp = av1_quantize_fp_c;
@ -370,6 +581,15 @@ static void setup_rtcd_internal(void)
if (flags & HAS_SSE2) av1_temporal_filter_apply = av1_temporal_filter_apply_sse2; if (flags & HAS_SSE2) av1_temporal_filter_apply = av1_temporal_filter_apply_sse2;
av1_warp_affine = av1_warp_affine_c; av1_warp_affine = av1_warp_affine_c;
if (flags & HAS_SSE2) av1_warp_affine = av1_warp_affine_sse2; if (flags & HAS_SSE2) av1_warp_affine = av1_warp_affine_sse2;
if (flags & HAS_SSSE3) av1_warp_affine = av1_warp_affine_ssse3;
av1_wedge_compute_delta_squares = av1_wedge_compute_delta_squares_c;
if (flags & HAS_SSE2) av1_wedge_compute_delta_squares = av1_wedge_compute_delta_squares_sse2;
av1_wedge_sign_from_residuals = av1_wedge_sign_from_residuals_c;
if (flags & HAS_SSE2) av1_wedge_sign_from_residuals = av1_wedge_sign_from_residuals_sse2;
av1_wedge_sse_from_residuals = av1_wedge_sse_from_residuals_c;
if (flags & HAS_SSE2) av1_wedge_sse_from_residuals = av1_wedge_sse_from_residuals_sse2;
compute_cross_correlation = compute_cross_correlation_c;
if (flags & HAS_SSE4_1) compute_cross_correlation = compute_cross_correlation_sse4_1;
copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_c; copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_c;
if (flags & HAS_SSE2) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2; if (flags & HAS_SSE2) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2;
if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3; if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3;

39
media/libaom/config/linux/x64/aom_config.asm
View file

@ -2,8 +2,6 @@
%define ARCH_MIPS 0 %define ARCH_MIPS 0
%define ARCH_X86 0 %define ARCH_X86 0
%define ARCH_X86_64 1 %define ARCH_X86_64 1
%define HAVE_EDSP 0
%define HAVE_MEDIA 0
%define HAVE_NEON 0 %define HAVE_NEON 0
%define HAVE_NEON_ASM 0 %define HAVE_NEON_ASM 0
%define HAVE_MIPS32 0 %define HAVE_MIPS32 0
@ -45,8 +43,6 @@
%define CONFIG_AV1_ENCODER 1 %define CONFIG_AV1_ENCODER 1
%define CONFIG_AV1_DECODER 1 %define CONFIG_AV1_DECODER 1
%define CONFIG_AV1 1 %define CONFIG_AV1 1
%define CONFIG_ENCODERS 1
%define CONFIG_DECODERS 1
%define CONFIG_STATIC_MSVCRT 0 %define CONFIG_STATIC_MSVCRT 0
%define CONFIG_SPATIAL_RESAMPLING 1 %define CONFIG_SPATIAL_RESAMPLING 1
%define CONFIG_REALTIME_ONLY 0 %define CONFIG_REALTIME_ONLY 0
@ -66,36 +62,35 @@
%define CONFIG_ENCODE_PERF_TESTS 0 %define CONFIG_ENCODE_PERF_TESTS 0
%define CONFIG_COEFFICIENT_RANGE_CHECKING 0 %define CONFIG_COEFFICIENT_RANGE_CHECKING 0
%define CONFIG_LOWBITDEPTH 1 %define CONFIG_LOWBITDEPTH 1
%define CONFIG_HIGHBITDEPTH 0 %define CONFIG_HIGHBITDEPTH 1
%define CONFIG_EXPERIMENTAL 0 %define CONFIG_EXPERIMENTAL 0
%define CONFIG_SIZE_LIMIT 1 %define CONFIG_SIZE_LIMIT 1
%define CONFIG_FP_MB_STATS 0 %define CONFIG_FP_MB_STATS 0
%define CONFIG_CDEF 1 %define CONFIG_CDEF 1
%define CONFIG_VAR_TX 0 %define CONFIG_VAR_TX 1
%define CONFIG_RECT_TX 1 %define CONFIG_RECT_TX 1
%define CONFIG_REF_MV 1 %define CONFIG_RECT_TX_EXT 0
%define CONFIG_TPL_MV 0 %define CONFIG_TPL_MV 0
%define CONFIG_DUAL_FILTER 1 %define CONFIG_DUAL_FILTER 1
%define CONFIG_CONVOLVE_ROUND 0 %define CONFIG_CONVOLVE_ROUND 0
%define CONFIG_COMPOUND_ROUND 0 %define CONFIG_COMPOUND_ROUND 0
%define CONFIG_EXT_TX 0 %define CONFIG_EXT_TX 1
%define CONFIG_DPCM_INTRA 0
%define CONFIG_TX64X64 0 %define CONFIG_TX64X64 0
%define CONFIG_SUB8X8_MC 0
%define CONFIG_EXT_INTRA 1 %define CONFIG_EXT_INTRA 1
%define CONFIG_INTRA_INTERP 0 %define CONFIG_INTRA_INTERP 0
%define CONFIG_FILTER_INTRA 0 %define CONFIG_FILTER_INTRA 0
%define CONFIG_INTRA_EDGE 0
%define CONFIG_INTRABC 0 %define CONFIG_INTRABC 0
%define CONFIG_EXT_INTER 0 %define CONFIG_EXT_INTER 1
%define CONFIG_INTERINTRA 0 %define CONFIG_INTERINTRA 1
%define CONFIG_WEDGE 0 %define CONFIG_WEDGE 1
%define CONFIG_COMPOUND_SEGMENT 0 %define CONFIG_COMPOUND_SEGMENT 1
%define CONFIG_EXT_REFS 1 %define CONFIG_EXT_REFS 1
%define CONFIG_GLOBAL_MOTION 1 %define CONFIG_GLOBAL_MOTION 1
%define CONFIG_NEW_QUANT 0 %define CONFIG_NEW_QUANT 0
%define CONFIG_SUPERTX 0 %define CONFIG_SUPERTX 0
%define CONFIG_ANS 0 %define CONFIG_ANS 0
%define CONFIG_EC_MULTISYMBOL 1
%define CONFIG_NEW_TOKENSET 1
%define CONFIG_LOOP_RESTORATION 0 %define CONFIG_LOOP_RESTORATION 0
%define CONFIG_EXT_PARTITION 0 %define CONFIG_EXT_PARTITION 0
%define CONFIG_EXT_PARTITION_TYPES 0 %define CONFIG_EXT_PARTITION_TYPES 0
@ -105,20 +100,18 @@
%define CONFIG_NCOBMC 0 %define CONFIG_NCOBMC 0
%define CONFIG_WARPED_MOTION 1 %define CONFIG_WARPED_MOTION 1
%define CONFIG_Q_ADAPT_PROBS 0 %define CONFIG_Q_ADAPT_PROBS 0
%define CONFIG_SUBFRAME_PROB_UPDATE 0
%define CONFIG_BITSTREAM_DEBUG 0 %define CONFIG_BITSTREAM_DEBUG 0
%define CONFIG_ALT_INTRA 1 %define CONFIG_ALT_INTRA 1
%define CONFIG_PALETTE 1 %define CONFIG_PALETTE 1
%define CONFIG_PALETTE_DELTA_ENCODING 0 %define CONFIG_PALETTE_DELTA_ENCODING 0
%define CONFIG_DAALA_EC 1
%define CONFIG_RAWBITS 0 %define CONFIG_RAWBITS 0
%define CONFIG_EC_SMALLMUL 0 %define CONFIG_EC_SMALLMUL 1
%define CONFIG_PVQ 0 %define CONFIG_PVQ 0
%define CONFIG_CFL 0 %define CONFIG_CFL 0
%define CONFIG_XIPHRC 0 %define CONFIG_XIPHRC 0
%define CONFIG_CB4X4 1 %define CONFIG_CB4X4 1
%define CONFIG_CHROMA_2X2 0 %define CONFIG_CHROMA_2X2 0
%define CONFIG_CHROMA_SUB8X8 0 %define CONFIG_CHROMA_SUB8X8 1
%define CONFIG_FRAME_SIZE 0 %define CONFIG_FRAME_SIZE 0
%define CONFIG_DELTA_Q 1 %define CONFIG_DELTA_Q 1
%define CONFIG_EXT_DELTA_Q 1 %define CONFIG_EXT_DELTA_Q 1
@ -126,10 +119,10 @@
%define CONFIG_FILTER_7BIT 1 %define CONFIG_FILTER_7BIT 1
%define CONFIG_PARALLEL_DEBLOCKING 0 %define CONFIG_PARALLEL_DEBLOCKING 0
%define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 %define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
%define CONFIG_LOOPFILTERING_ACROSS_TILES 0 %define CONFIG_LOOPFILTERING_ACROSS_TILES 1
%define CONFIG_TILE_GROUPS 1 %define CONFIG_TILE_GROUPS 1
%define CONFIG_EC_ADAPT 1 %define CONFIG_EC_ADAPT 1
%define CONFIG_TEMPMV_SIGNALING 0 %define CONFIG_TEMPMV_SIGNALING 1
%define CONFIG_RD_DEBUG 0 %define CONFIG_RD_DEBUG 0
%define CONFIG_REFERENCE_BUFFER 1 %define CONFIG_REFERENCE_BUFFER 1
%define CONFIG_COEF_INTERLEAVE 0 %define CONFIG_COEF_INTERLEAVE 0
@ -147,5 +140,7 @@
%define CONFIG_NEW_MULTISYMBOL 0 %define CONFIG_NEW_MULTISYMBOL 0
%define CONFIG_COMPOUND_SINGLEREF 0 %define CONFIG_COMPOUND_SINGLEREF 0
%define CONFIG_AOM_QM 0 %define CONFIG_AOM_QM 0
%define CONFIG_LOWDELAY_COMPOUND 0 %define CONFIG_ONE_SIDED_COMPOUND 1
%define CONFIG_SMOOTH_HV 0
%define CONFIG_VAR_REFS 0
%define CONFIG_ANALYZER 0 %define CONFIG_ANALYZER 0

39
media/libaom/config/linux/x64/aom_config.h
View file

@ -15,8 +15,6 @@
#define ARCH_MIPS 0 #define ARCH_MIPS 0
#define ARCH_X86 0 #define ARCH_X86 0
#define ARCH_X86_64 1 #define ARCH_X86_64 1
#define HAVE_EDSP 0
#define HAVE_MEDIA 0
#define HAVE_NEON 0 #define HAVE_NEON 0
#define HAVE_NEON_ASM 0 #define HAVE_NEON_ASM 0
#define HAVE_MIPS32 0 #define HAVE_MIPS32 0
@ -58,8 +56,6 @@
#define CONFIG_AV1_ENCODER 1 #define CONFIG_AV1_ENCODER 1
#define CONFIG_AV1_DECODER 1 #define CONFIG_AV1_DECODER 1
#define CONFIG_AV1 1 #define CONFIG_AV1 1
#define CONFIG_ENCODERS 1
#define CONFIG_DECODERS 1
#define CONFIG_STATIC_MSVCRT 0 #define CONFIG_STATIC_MSVCRT 0
#define CONFIG_SPATIAL_RESAMPLING 1 #define CONFIG_SPATIAL_RESAMPLING 1
#define CONFIG_REALTIME_ONLY 0 #define CONFIG_REALTIME_ONLY 0
@ -79,36 +75,35 @@
#define CONFIG_ENCODE_PERF_TESTS 0 #define CONFIG_ENCODE_PERF_TESTS 0
#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 #define CONFIG_COEFFICIENT_RANGE_CHECKING 0
#define CONFIG_LOWBITDEPTH 1 #define CONFIG_LOWBITDEPTH 1
#define CONFIG_HIGHBITDEPTH 0 #define CONFIG_HIGHBITDEPTH 1
#define CONFIG_EXPERIMENTAL 0 #define CONFIG_EXPERIMENTAL 0
#define CONFIG_SIZE_LIMIT 1 #define CONFIG_SIZE_LIMIT 1
#define CONFIG_FP_MB_STATS 0 #define CONFIG_FP_MB_STATS 0
#define CONFIG_CDEF 1 #define CONFIG_CDEF 1
#define CONFIG_VAR_TX 0 #define CONFIG_VAR_TX 1
#define CONFIG_RECT_TX 1 #define CONFIG_RECT_TX 1
#define CONFIG_REF_MV 1 #define CONFIG_RECT_TX_EXT 0
#define CONFIG_TPL_MV 0 #define CONFIG_TPL_MV 0
#define CONFIG_DUAL_FILTER 1 #define CONFIG_DUAL_FILTER 1
#define CONFIG_CONVOLVE_ROUND 0 #define CONFIG_CONVOLVE_ROUND 0
#define CONFIG_COMPOUND_ROUND 0 #define CONFIG_COMPOUND_ROUND 0
#define CONFIG_EXT_TX 0 #define CONFIG_EXT_TX 1
#define CONFIG_DPCM_INTRA 0
#define CONFIG_TX64X64 0 #define CONFIG_TX64X64 0
#define CONFIG_SUB8X8_MC 0
#define CONFIG_EXT_INTRA 1 #define CONFIG_EXT_INTRA 1
#define CONFIG_INTRA_INTERP 0 #define CONFIG_INTRA_INTERP 0
#define CONFIG_FILTER_INTRA 0 #define CONFIG_FILTER_INTRA 0
#define CONFIG_INTRA_EDGE 0
#define CONFIG_INTRABC 0 #define CONFIG_INTRABC 0
#define CONFIG_EXT_INTER 0 #define CONFIG_EXT_INTER 1
#define CONFIG_INTERINTRA 0 #define CONFIG_INTERINTRA 1
#define CONFIG_WEDGE 0 #define CONFIG_WEDGE 1
#define CONFIG_COMPOUND_SEGMENT 0 #define CONFIG_COMPOUND_SEGMENT 1
#define CONFIG_EXT_REFS 1 #define CONFIG_EXT_REFS 1
#define CONFIG_GLOBAL_MOTION 1 #define CONFIG_GLOBAL_MOTION 1
#define CONFIG_NEW_QUANT 0 #define CONFIG_NEW_QUANT 0
#define CONFIG_SUPERTX 0 #define CONFIG_SUPERTX 0
#define CONFIG_ANS 0 #define CONFIG_ANS 0
#define CONFIG_EC_MULTISYMBOL 1
#define CONFIG_NEW_TOKENSET 1
#define CONFIG_LOOP_RESTORATION 0 #define CONFIG_LOOP_RESTORATION 0
#define CONFIG_EXT_PARTITION 0 #define CONFIG_EXT_PARTITION 0
#define CONFIG_EXT_PARTITION_TYPES 0 #define CONFIG_EXT_PARTITION_TYPES 0
@ -118,20 +113,18 @@
#define CONFIG_NCOBMC 0 #define CONFIG_NCOBMC 0
#define CONFIG_WARPED_MOTION 1 #define CONFIG_WARPED_MOTION 1
#define CONFIG_Q_ADAPT_PROBS 0 #define CONFIG_Q_ADAPT_PROBS 0
#define CONFIG_SUBFRAME_PROB_UPDATE 0
#define CONFIG_BITSTREAM_DEBUG 0 #define CONFIG_BITSTREAM_DEBUG 0
#define CONFIG_ALT_INTRA 1 #define CONFIG_ALT_INTRA 1
#define CONFIG_PALETTE 1 #define CONFIG_PALETTE 1
#define CONFIG_PALETTE_DELTA_ENCODING 0 #define CONFIG_PALETTE_DELTA_ENCODING 0
#define CONFIG_DAALA_EC 1
#define CONFIG_RAWBITS 0 #define CONFIG_RAWBITS 0
#define CONFIG_EC_SMALLMUL 0 #define CONFIG_EC_SMALLMUL 1
#define CONFIG_PVQ 0 #define CONFIG_PVQ 0
#define CONFIG_CFL 0 #define CONFIG_CFL 0
#define CONFIG_XIPHRC 0 #define CONFIG_XIPHRC 0
#define CONFIG_CB4X4 1 #define CONFIG_CB4X4 1
#define CONFIG_CHROMA_2X2 0 #define CONFIG_CHROMA_2X2 0
#define CONFIG_CHROMA_SUB8X8 0 #define CONFIG_CHROMA_SUB8X8 1
#define CONFIG_FRAME_SIZE 0 #define CONFIG_FRAME_SIZE 0
#define CONFIG_DELTA_Q 1 #define CONFIG_DELTA_Q 1
#define CONFIG_EXT_DELTA_Q 1 #define CONFIG_EXT_DELTA_Q 1
@ -139,10 +132,10 @@
#define CONFIG_FILTER_7BIT 1 #define CONFIG_FILTER_7BIT 1
#define CONFIG_PARALLEL_DEBLOCKING 0 #define CONFIG_PARALLEL_DEBLOCKING 0
#define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 #define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
#define CONFIG_LOOPFILTERING_ACROSS_TILES 0 #define CONFIG_LOOPFILTERING_ACROSS_TILES 1
#define CONFIG_TILE_GROUPS 1 #define CONFIG_TILE_GROUPS 1
#define CONFIG_EC_ADAPT 1 #define CONFIG_EC_ADAPT 1
#define CONFIG_TEMPMV_SIGNALING 0 #define CONFIG_TEMPMV_SIGNALING 1
#define CONFIG_RD_DEBUG 0 #define CONFIG_RD_DEBUG 0
#define CONFIG_REFERENCE_BUFFER 1 #define CONFIG_REFERENCE_BUFFER 1
#define CONFIG_COEF_INTERLEAVE 0 #define CONFIG_COEF_INTERLEAVE 0
@ -160,7 +153,9 @@
#define CONFIG_NEW_MULTISYMBOL 0 #define CONFIG_NEW_MULTISYMBOL 0
#define CONFIG_COMPOUND_SINGLEREF 0 #define CONFIG_COMPOUND_SINGLEREF 0
#define CONFIG_AOM_QM 0 #define CONFIG_AOM_QM 0
#define CONFIG_LOWDELAY_COMPOUND 0 #define CONFIG_ONE_SIDED_COMPOUND 1
#define CONFIG_SMOOTH_HV 0
#define CONFIG_VAR_REFS 0
#define CONFIG_ANALYZER 0 #define CONFIG_ANALYZER 0
#define DECODE_WIDTH_LIMIT 8192 #define DECODE_WIDTH_LIMIT 8192
#define DECODE_HEIGHT_LIMIT 4608 #define DECODE_HEIGHT_LIMIT 4608

2207
media/libaom/config/linux/x64/aom_dsp_rtcd.h
View file

File diff suppressed because it is too large Load diff

267
media/libaom/config/linux/x64/av1_rtcd.h
View file

@ -60,13 +60,7 @@ void aom_clpf_hblock_hbd_sse4_1(uint16_t *dst, const uint16_t *src, int dstride,
RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd); RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd);
int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); #define av1_block_error av1_block_error_c
int64_t av1_block_error_avx2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
RTCD_EXTERN int64_t (*av1_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_fp_c(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
int64_t av1_block_error_fp_sse2(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
#define av1_block_error_fp av1_block_error_fp_sse2
void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
@ -80,9 +74,7 @@ int av1_diamond_search_sad_c(struct macroblock *x, const struct search_site_conf
#define av1_diamond_search_sad av1_diamond_search_sad_c #define av1_diamond_search_sad av1_diamond_search_sad_c
void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fdct8x8_quant_sse2(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_fdct8x8_quant av1_fdct8x8_quant_c
void av1_fdct8x8_quant_ssse3(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
RTCD_EXTERN void (*av1_fdct8x8_quant)(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type);
@ -149,9 +141,174 @@ RTCD_EXTERN int (*av1_full_search_sad)(const struct macroblock *x, const struct
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type); void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type);
#define av1_fwd_idtx av1_fwd_idtx_c #define av1_fwd_idtx av1_fwd_idtx_c
void av1_fwd_txfm2d_16x16_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_16x16_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_16x16)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_32x32)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_4x4)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_64x64)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_8x8)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_fwht4x4 av1_fwht4x4_c #define av1_fwht4x4 av1_fwht4x4_c
int64_t av1_highbd_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
int64_t av1_highbd_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
#define av1_highbd_block_error av1_highbd_block_error_sse2
void av1_highbd_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8 av1_highbd_convolve8_sse2
void av1_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg av1_highbd_convolve8_avg_sse2
void av1_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_horiz av1_highbd_convolve8_avg_horiz_sse2
void av1_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_vert av1_highbd_convolve8_avg_vert_sse2
void av1_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_horiz av1_highbd_convolve8_horiz_sse2
void av1_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_vert av1_highbd_convolve8_vert_sse2
void av1_highbd_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_avg av1_highbd_convolve_avg_c
void av1_highbd_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_copy av1_highbd_convolve_copy_c
void av1_highbd_convolve_horiz_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_horiz_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_horiz)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_init_c(void);
void av1_highbd_convolve_init_sse4_1(void);
RTCD_EXTERN void (*av1_highbd_convolve_init)(void);
void av1_highbd_convolve_vert_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_vert_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_vert)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x16 av1_highbd_fht16x16_c
void av1_highbd_fht16x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x32 av1_highbd_fht16x32_c
void av1_highbd_fht16x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x4 av1_highbd_fht16x4_c
void av1_highbd_fht16x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x8 av1_highbd_fht16x8_c
void av1_highbd_fht32x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x16 av1_highbd_fht32x16_c
void av1_highbd_fht32x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x32 av1_highbd_fht32x32_c
void av1_highbd_fht32x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x8 av1_highbd_fht32x8_c
void av1_highbd_fht4x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x16 av1_highbd_fht4x16_c
void av1_highbd_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x4_sse4_1(const int16_t *input, tran_low_t *output, int stride, int tx_type);
RTCD_EXTERN void (*av1_highbd_fht4x4)(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x8 av1_highbd_fht4x8_c
void av1_highbd_fht8x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x16 av1_highbd_fht8x16_c
void av1_highbd_fht8x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x32 av1_highbd_fht8x32_c
void av1_highbd_fht8x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x4 av1_highbd_fht8x4_c
void av1_highbd_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x8 av1_highbd_fht8x8_c
void av1_highbd_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_highbd_fwht4x4 av1_highbd_fwht4x4_c
void av1_highbd_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd);
#define av1_highbd_iht16x16_256_add av1_highbd_iht16x16_256_add_c
void av1_highbd_iht16x32_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x32_512_add av1_highbd_iht16x32_512_add_c
void av1_highbd_iht16x4_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x4_64_add av1_highbd_iht16x4_64_add_c
void av1_highbd_iht16x8_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x8_128_add av1_highbd_iht16x8_128_add_c
void av1_highbd_iht32x16_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x16_512_add av1_highbd_iht32x16_512_add_c
void av1_highbd_iht32x8_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x8_256_add av1_highbd_iht32x8_256_add_c
void av1_highbd_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x16_64_add av1_highbd_iht4x16_64_add_c
void av1_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x4_16_add av1_highbd_iht4x4_16_add_c
void av1_highbd_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x8_32_add av1_highbd_iht4x8_32_add_c
void av1_highbd_iht8x16_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x16_128_add av1_highbd_iht8x16_128_add_c
void av1_highbd_iht8x32_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x32_256_add av1_highbd_iht8x32_256_add_c
void av1_highbd_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x4_32_add av1_highbd_iht8x4_32_add_c
void av1_highbd_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x8_64_add av1_highbd_iht8x8_64_add_c
void av1_highbd_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_b av1_highbd_quantize_b_c
void av1_highbd_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_quantize_fp_sse4_1(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
RTCD_EXTERN void (*av1_highbd_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_highbd_temporal_filter_apply av1_highbd_temporal_filter_apply_c
void av1_highbd_warp_affine_c(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_highbd_warp_affine_ssse3(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_highbd_warp_affine)(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
@ -204,6 +361,25 @@ void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride
void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
#define av1_iht8x8_64_add av1_iht8x8_64_add_sse2 #define av1_iht8x8_64_add av1_iht8x8_64_add_sse2
void av1_inv_txfm2d_add_16x16_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_16x16_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_16x16)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_avx2(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_32x32)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_4x4)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_64x64_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_64x64 av1_inv_txfm2d_add_64x64_c
void av1_inv_txfm2d_add_8x8_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_8x8_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_8x8)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_lowbd_convolve_init_c(void); void av1_lowbd_convolve_init_c(void);
void av1_lowbd_convolve_init_ssse3(void); void av1_lowbd_convolve_init_ssse3(void);
RTCD_EXTERN void (*av1_lowbd_convolve_init)(void); RTCD_EXTERN void (*av1_lowbd_convolve_init)(void);
@ -213,20 +389,35 @@ void av1_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_b
void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_sse2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_sse2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_ssse3(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_quantize_fp av1_quantize_fp_sse2
RTCD_EXTERN void (*av1_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_32x32_ssse3(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_quantize_fp_32x32 av1_quantize_fp_32x32_c
RTCD_EXTERN void (*av1_quantize_fp_32x32)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_temporal_filter_apply av1_temporal_filter_apply_sse2 #define av1_temporal_filter_apply av1_temporal_filter_apply_sse2
void av1_warp_affine_c(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_c(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_warp_affine_sse2(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_sse2(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
#define av1_warp_affine av1_warp_affine_sse2 void av1_warp_affine_ssse3(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_warp_affine)(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_wedge_compute_delta_squares_c(int16_t *d, const int16_t *a, const int16_t *b, int N);
void av1_wedge_compute_delta_squares_sse2(int16_t *d, const int16_t *a, const int16_t *b, int N);
#define av1_wedge_compute_delta_squares av1_wedge_compute_delta_squares_sse2
int av1_wedge_sign_from_residuals_c(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
int av1_wedge_sign_from_residuals_sse2(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
#define av1_wedge_sign_from_residuals av1_wedge_sign_from_residuals_sse2
uint64_t av1_wedge_sse_from_residuals_c(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
uint64_t av1_wedge_sse_from_residuals_sse2(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
#define av1_wedge_sse_from_residuals av1_wedge_sse_from_residuals_sse2
double compute_cross_correlation_c(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
double compute_cross_correlation_sse4_1(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
RTCD_EXTERN double (*compute_cross_correlation)(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
@ -304,14 +495,10 @@ static void setup_rtcd_internal(void)
aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2; aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2;
if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3; if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3;
if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1; if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1;
av1_block_error = av1_block_error_sse2;
if (flags & HAS_AVX2) av1_block_error = av1_block_error_avx2;
av1_convolve_horiz = av1_convolve_horiz_c; av1_convolve_horiz = av1_convolve_horiz_c;
if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3; if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3;
av1_convolve_vert = av1_convolve_vert_c; av1_convolve_vert = av1_convolve_vert_c;
if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3; if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3;
av1_fdct8x8_quant = av1_fdct8x8_quant_sse2;
if (flags & HAS_SSSE3) av1_fdct8x8_quant = av1_fdct8x8_quant_ssse3;
av1_fht16x16 = av1_fht16x16_sse2; av1_fht16x16 = av1_fht16x16_sse2;
if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2; if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2;
av1_fht32x32 = av1_fht32x32_sse2; av1_fht32x32 = av1_fht32x32_sse2;
@ -319,14 +506,44 @@ static void setup_rtcd_internal(void)
av1_full_search_sad = av1_full_search_sad_c; av1_full_search_sad = av1_full_search_sad_c;
if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3; if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3;
if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8; if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8;
av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_sse4_1;
av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_sse4_1;
av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_sse4_1;
av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_sse4_1;
av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_sse4_1;
av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_sse4_1;
av1_highbd_convolve_init = av1_highbd_convolve_init_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_init = av1_highbd_convolve_init_sse4_1;
av1_highbd_convolve_vert = av1_highbd_convolve_vert_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_vert = av1_highbd_convolve_vert_sse4_1;
av1_highbd_fht4x4 = av1_highbd_fht4x4_c;
if (flags & HAS_SSE4_1) av1_highbd_fht4x4 = av1_highbd_fht4x4_sse4_1;
av1_highbd_quantize_fp = av1_highbd_quantize_fp_c;
if (flags & HAS_SSE4_1) av1_highbd_quantize_fp = av1_highbd_quantize_fp_sse4_1;
av1_highbd_warp_affine = av1_highbd_warp_affine_c;
if (flags & HAS_SSSE3) av1_highbd_warp_affine = av1_highbd_warp_affine_ssse3;
av1_iht16x16_256_add = av1_iht16x16_256_add_sse2; av1_iht16x16_256_add = av1_iht16x16_256_add_sse2;
if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2; if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2;
av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_sse4_1;
av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_c;
if (flags & HAS_AVX2) av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_avx2;
av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_sse4_1;
av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_sse4_1;
av1_lowbd_convolve_init = av1_lowbd_convolve_init_c; av1_lowbd_convolve_init = av1_lowbd_convolve_init_c;
if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3; if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3;
av1_quantize_fp = av1_quantize_fp_sse2; av1_warp_affine = av1_warp_affine_sse2;
if (flags & HAS_SSSE3) av1_quantize_fp = av1_quantize_fp_ssse3; if (flags & HAS_SSSE3) av1_warp_affine = av1_warp_affine_ssse3;
av1_quantize_fp_32x32 = av1_quantize_fp_32x32_c; compute_cross_correlation = compute_cross_correlation_c;
if (flags & HAS_SSSE3) av1_quantize_fp_32x32 = av1_quantize_fp_32x32_ssse3; if (flags & HAS_SSE4_1) compute_cross_correlation = compute_cross_correlation_sse4_1;
copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2; copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2;
if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3; if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3;
if (flags & HAS_SSE4_1) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse4_1; if (flags & HAS_SSE4_1) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse4_1;

39
media/libaom/config/mac/x64/aom_config.asm
View file

@ -2,8 +2,6 @@
%define ARCH_MIPS 0 %define ARCH_MIPS 0
%define ARCH_X86 0 %define ARCH_X86 0
%define ARCH_X86_64 1 %define ARCH_X86_64 1
%define HAVE_EDSP 0
%define HAVE_MEDIA 0
%define HAVE_NEON 0 %define HAVE_NEON 0
%define HAVE_NEON_ASM 0 %define HAVE_NEON_ASM 0
%define HAVE_MIPS32 0 %define HAVE_MIPS32 0
@ -45,8 +43,6 @@
%define CONFIG_AV1_ENCODER 1 %define CONFIG_AV1_ENCODER 1
%define CONFIG_AV1_DECODER 1 %define CONFIG_AV1_DECODER 1
%define CONFIG_AV1 1 %define CONFIG_AV1 1
%define CONFIG_ENCODERS 1
%define CONFIG_DECODERS 1
%define CONFIG_STATIC_MSVCRT 0 %define CONFIG_STATIC_MSVCRT 0
%define CONFIG_SPATIAL_RESAMPLING 1 %define CONFIG_SPATIAL_RESAMPLING 1
%define CONFIG_REALTIME_ONLY 0 %define CONFIG_REALTIME_ONLY 0
@ -66,36 +62,35 @@
%define CONFIG_ENCODE_PERF_TESTS 0 %define CONFIG_ENCODE_PERF_TESTS 0
%define CONFIG_COEFFICIENT_RANGE_CHECKING 0 %define CONFIG_COEFFICIENT_RANGE_CHECKING 0
%define CONFIG_LOWBITDEPTH 1 %define CONFIG_LOWBITDEPTH 1
%define CONFIG_HIGHBITDEPTH 0 %define CONFIG_HIGHBITDEPTH 1
%define CONFIG_EXPERIMENTAL 0 %define CONFIG_EXPERIMENTAL 0
%define CONFIG_SIZE_LIMIT 1 %define CONFIG_SIZE_LIMIT 1
%define CONFIG_FP_MB_STATS 0 %define CONFIG_FP_MB_STATS 0
%define CONFIG_CDEF 1 %define CONFIG_CDEF 1
%define CONFIG_VAR_TX 0 %define CONFIG_VAR_TX 1
%define CONFIG_RECT_TX 1 %define CONFIG_RECT_TX 1
%define CONFIG_REF_MV 1 %define CONFIG_RECT_TX_EXT 0
%define CONFIG_TPL_MV 0 %define CONFIG_TPL_MV 0
%define CONFIG_DUAL_FILTER 1 %define CONFIG_DUAL_FILTER 1
%define CONFIG_CONVOLVE_ROUND 0 %define CONFIG_CONVOLVE_ROUND 0
%define CONFIG_COMPOUND_ROUND 0 %define CONFIG_COMPOUND_ROUND 0
%define CONFIG_EXT_TX 0 %define CONFIG_EXT_TX 1
%define CONFIG_DPCM_INTRA 0
%define CONFIG_TX64X64 0 %define CONFIG_TX64X64 0
%define CONFIG_SUB8X8_MC 0
%define CONFIG_EXT_INTRA 1 %define CONFIG_EXT_INTRA 1
%define CONFIG_INTRA_INTERP 0 %define CONFIG_INTRA_INTERP 0
%define CONFIG_FILTER_INTRA 0 %define CONFIG_FILTER_INTRA 0
%define CONFIG_INTRA_EDGE 0
%define CONFIG_INTRABC 0 %define CONFIG_INTRABC 0
%define CONFIG_EXT_INTER 0 %define CONFIG_EXT_INTER 1
%define CONFIG_INTERINTRA 0 %define CONFIG_INTERINTRA 1
%define CONFIG_WEDGE 0 %define CONFIG_WEDGE 1
%define CONFIG_COMPOUND_SEGMENT 0 %define CONFIG_COMPOUND_SEGMENT 1
%define CONFIG_EXT_REFS 1 %define CONFIG_EXT_REFS 1
%define CONFIG_GLOBAL_MOTION 1 %define CONFIG_GLOBAL_MOTION 1
%define CONFIG_NEW_QUANT 0 %define CONFIG_NEW_QUANT 0
%define CONFIG_SUPERTX 0 %define CONFIG_SUPERTX 0
%define CONFIG_ANS 0 %define CONFIG_ANS 0
%define CONFIG_EC_MULTISYMBOL 1
%define CONFIG_NEW_TOKENSET 1
%define CONFIG_LOOP_RESTORATION 0 %define CONFIG_LOOP_RESTORATION 0
%define CONFIG_EXT_PARTITION 0 %define CONFIG_EXT_PARTITION 0
%define CONFIG_EXT_PARTITION_TYPES 0 %define CONFIG_EXT_PARTITION_TYPES 0
@ -105,20 +100,18 @@
%define CONFIG_NCOBMC 0 %define CONFIG_NCOBMC 0
%define CONFIG_WARPED_MOTION 1 %define CONFIG_WARPED_MOTION 1
%define CONFIG_Q_ADAPT_PROBS 0 %define CONFIG_Q_ADAPT_PROBS 0
%define CONFIG_SUBFRAME_PROB_UPDATE 0
%define CONFIG_BITSTREAM_DEBUG 0 %define CONFIG_BITSTREAM_DEBUG 0
%define CONFIG_ALT_INTRA 1 %define CONFIG_ALT_INTRA 1
%define CONFIG_PALETTE 1 %define CONFIG_PALETTE 1
%define CONFIG_PALETTE_DELTA_ENCODING 0 %define CONFIG_PALETTE_DELTA_ENCODING 0
%define CONFIG_DAALA_EC 1
%define CONFIG_RAWBITS 0 %define CONFIG_RAWBITS 0
%define CONFIG_EC_SMALLMUL 0 %define CONFIG_EC_SMALLMUL 1
%define CONFIG_PVQ 0 %define CONFIG_PVQ 0
%define CONFIG_CFL 0 %define CONFIG_CFL 0
%define CONFIG_XIPHRC 0 %define CONFIG_XIPHRC 0
%define CONFIG_CB4X4 1 %define CONFIG_CB4X4 1
%define CONFIG_CHROMA_2X2 0 %define CONFIG_CHROMA_2X2 0
%define CONFIG_CHROMA_SUB8X8 0 %define CONFIG_CHROMA_SUB8X8 1
%define CONFIG_FRAME_SIZE 0 %define CONFIG_FRAME_SIZE 0
%define CONFIG_DELTA_Q 1 %define CONFIG_DELTA_Q 1
%define CONFIG_EXT_DELTA_Q 1 %define CONFIG_EXT_DELTA_Q 1
@ -126,10 +119,10 @@
%define CONFIG_FILTER_7BIT 1 %define CONFIG_FILTER_7BIT 1
%define CONFIG_PARALLEL_DEBLOCKING 0 %define CONFIG_PARALLEL_DEBLOCKING 0
%define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 %define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
%define CONFIG_LOOPFILTERING_ACROSS_TILES 0 %define CONFIG_LOOPFILTERING_ACROSS_TILES 1
%define CONFIG_TILE_GROUPS 1 %define CONFIG_TILE_GROUPS 1
%define CONFIG_EC_ADAPT 1 %define CONFIG_EC_ADAPT 1
%define CONFIG_TEMPMV_SIGNALING 0 %define CONFIG_TEMPMV_SIGNALING 1
%define CONFIG_RD_DEBUG 0 %define CONFIG_RD_DEBUG 0
%define CONFIG_REFERENCE_BUFFER 1 %define CONFIG_REFERENCE_BUFFER 1
%define CONFIG_COEF_INTERLEAVE 0 %define CONFIG_COEF_INTERLEAVE 0
@ -147,5 +140,7 @@
%define CONFIG_NEW_MULTISYMBOL 0 %define CONFIG_NEW_MULTISYMBOL 0
%define CONFIG_COMPOUND_SINGLEREF 0 %define CONFIG_COMPOUND_SINGLEREF 0
%define CONFIG_AOM_QM 0 %define CONFIG_AOM_QM 0
%define CONFIG_LOWDELAY_COMPOUND 0 %define CONFIG_ONE_SIDED_COMPOUND 1
%define CONFIG_SMOOTH_HV 0
%define CONFIG_VAR_REFS 0
%define CONFIG_ANALYZER 0 %define CONFIG_ANALYZER 0

39
media/libaom/config/mac/x64/aom_config.h
View file

@ -15,8 +15,6 @@
#define ARCH_MIPS 0 #define ARCH_MIPS 0
#define ARCH_X86 0 #define ARCH_X86 0
#define ARCH_X86_64 1 #define ARCH_X86_64 1
#define HAVE_EDSP 0
#define HAVE_MEDIA 0
#define HAVE_NEON 0 #define HAVE_NEON 0
#define HAVE_NEON_ASM 0 #define HAVE_NEON_ASM 0
#define HAVE_MIPS32 0 #define HAVE_MIPS32 0
@ -58,8 +56,6 @@
#define CONFIG_AV1_ENCODER 1 #define CONFIG_AV1_ENCODER 1
#define CONFIG_AV1_DECODER 1 #define CONFIG_AV1_DECODER 1
#define CONFIG_AV1 1 #define CONFIG_AV1 1
#define CONFIG_ENCODERS 1
#define CONFIG_DECODERS 1
#define CONFIG_STATIC_MSVCRT 0 #define CONFIG_STATIC_MSVCRT 0
#define CONFIG_SPATIAL_RESAMPLING 1 #define CONFIG_SPATIAL_RESAMPLING 1
#define CONFIG_REALTIME_ONLY 0 #define CONFIG_REALTIME_ONLY 0
@ -79,36 +75,35 @@
#define CONFIG_ENCODE_PERF_TESTS 0 #define CONFIG_ENCODE_PERF_TESTS 0
#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 #define CONFIG_COEFFICIENT_RANGE_CHECKING 0
#define CONFIG_LOWBITDEPTH 1 #define CONFIG_LOWBITDEPTH 1
#define CONFIG_HIGHBITDEPTH 0 #define CONFIG_HIGHBITDEPTH 1
#define CONFIG_EXPERIMENTAL 0 #define CONFIG_EXPERIMENTAL 0
#define CONFIG_SIZE_LIMIT 1 #define CONFIG_SIZE_LIMIT 1
#define CONFIG_FP_MB_STATS 0 #define CONFIG_FP_MB_STATS 0
#define CONFIG_CDEF 1 #define CONFIG_CDEF 1
#define CONFIG_VAR_TX 0 #define CONFIG_VAR_TX 1
#define CONFIG_RECT_TX 1 #define CONFIG_RECT_TX 1
#define CONFIG_REF_MV 1 #define CONFIG_RECT_TX_EXT 0
#define CONFIG_TPL_MV 0 #define CONFIG_TPL_MV 0
#define CONFIG_DUAL_FILTER 1 #define CONFIG_DUAL_FILTER 1
#define CONFIG_CONVOLVE_ROUND 0 #define CONFIG_CONVOLVE_ROUND 0
#define CONFIG_COMPOUND_ROUND 0 #define CONFIG_COMPOUND_ROUND 0
#define CONFIG_EXT_TX 0 #define CONFIG_EXT_TX 1
#define CONFIG_DPCM_INTRA 0
#define CONFIG_TX64X64 0 #define CONFIG_TX64X64 0
#define CONFIG_SUB8X8_MC 0
#define CONFIG_EXT_INTRA 1 #define CONFIG_EXT_INTRA 1
#define CONFIG_INTRA_INTERP 0 #define CONFIG_INTRA_INTERP 0
#define CONFIG_FILTER_INTRA 0 #define CONFIG_FILTER_INTRA 0
#define CONFIG_INTRA_EDGE 0
#define CONFIG_INTRABC 0 #define CONFIG_INTRABC 0
#define CONFIG_EXT_INTER 0 #define CONFIG_EXT_INTER 1
#define CONFIG_INTERINTRA 0 #define CONFIG_INTERINTRA 1
#define CONFIG_WEDGE 0 #define CONFIG_WEDGE 1
#define CONFIG_COMPOUND_SEGMENT 0 #define CONFIG_COMPOUND_SEGMENT 1
#define CONFIG_EXT_REFS 1 #define CONFIG_EXT_REFS 1
#define CONFIG_GLOBAL_MOTION 1 #define CONFIG_GLOBAL_MOTION 1
#define CONFIG_NEW_QUANT 0 #define CONFIG_NEW_QUANT 0
#define CONFIG_SUPERTX 0 #define CONFIG_SUPERTX 0
#define CONFIG_ANS 0 #define CONFIG_ANS 0
#define CONFIG_EC_MULTISYMBOL 1
#define CONFIG_NEW_TOKENSET 1
#define CONFIG_LOOP_RESTORATION 0 #define CONFIG_LOOP_RESTORATION 0
#define CONFIG_EXT_PARTITION 0 #define CONFIG_EXT_PARTITION 0
#define CONFIG_EXT_PARTITION_TYPES 0 #define CONFIG_EXT_PARTITION_TYPES 0
@ -118,20 +113,18 @@
#define CONFIG_NCOBMC 0 #define CONFIG_NCOBMC 0
#define CONFIG_WARPED_MOTION 1 #define CONFIG_WARPED_MOTION 1
#define CONFIG_Q_ADAPT_PROBS 0 #define CONFIG_Q_ADAPT_PROBS 0
#define CONFIG_SUBFRAME_PROB_UPDATE 0
#define CONFIG_BITSTREAM_DEBUG 0 #define CONFIG_BITSTREAM_DEBUG 0
#define CONFIG_ALT_INTRA 1 #define CONFIG_ALT_INTRA 1
#define CONFIG_PALETTE 1 #define CONFIG_PALETTE 1
#define CONFIG_PALETTE_DELTA_ENCODING 0 #define CONFIG_PALETTE_DELTA_ENCODING 0
#define CONFIG_DAALA_EC 1
#define CONFIG_RAWBITS 0 #define CONFIG_RAWBITS 0
#define CONFIG_EC_SMALLMUL 0 #define CONFIG_EC_SMALLMUL 1
#define CONFIG_PVQ 0 #define CONFIG_PVQ 0
#define CONFIG_CFL 0 #define CONFIG_CFL 0
#define CONFIG_XIPHRC 0 #define CONFIG_XIPHRC 0
#define CONFIG_CB4X4 1 #define CONFIG_CB4X4 1
#define CONFIG_CHROMA_2X2 0 #define CONFIG_CHROMA_2X2 0
#define CONFIG_CHROMA_SUB8X8 0 #define CONFIG_CHROMA_SUB8X8 1
#define CONFIG_FRAME_SIZE 0 #define CONFIG_FRAME_SIZE 0
#define CONFIG_DELTA_Q 1 #define CONFIG_DELTA_Q 1
#define CONFIG_EXT_DELTA_Q 1 #define CONFIG_EXT_DELTA_Q 1
@ -139,10 +132,10 @@
#define CONFIG_FILTER_7BIT 1 #define CONFIG_FILTER_7BIT 1
#define CONFIG_PARALLEL_DEBLOCKING 0 #define CONFIG_PARALLEL_DEBLOCKING 0
#define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 #define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
#define CONFIG_LOOPFILTERING_ACROSS_TILES 0 #define CONFIG_LOOPFILTERING_ACROSS_TILES 1
#define CONFIG_TILE_GROUPS 1 #define CONFIG_TILE_GROUPS 1
#define CONFIG_EC_ADAPT 1 #define CONFIG_EC_ADAPT 1
#define CONFIG_TEMPMV_SIGNALING 0 #define CONFIG_TEMPMV_SIGNALING 1
#define CONFIG_RD_DEBUG 0 #define CONFIG_RD_DEBUG 0
#define CONFIG_REFERENCE_BUFFER 1 #define CONFIG_REFERENCE_BUFFER 1
#define CONFIG_COEF_INTERLEAVE 0 #define CONFIG_COEF_INTERLEAVE 0
@ -160,7 +153,9 @@
#define CONFIG_NEW_MULTISYMBOL 0 #define CONFIG_NEW_MULTISYMBOL 0
#define CONFIG_COMPOUND_SINGLEREF 0 #define CONFIG_COMPOUND_SINGLEREF 0
#define CONFIG_AOM_QM 0 #define CONFIG_AOM_QM 0
#define CONFIG_LOWDELAY_COMPOUND 0 #define CONFIG_ONE_SIDED_COMPOUND 1
#define CONFIG_SMOOTH_HV 0
#define CONFIG_VAR_REFS 0
#define CONFIG_ANALYZER 0 #define CONFIG_ANALYZER 0
#define DECODE_WIDTH_LIMIT 8192 #define DECODE_WIDTH_LIMIT 8192
#define DECODE_HEIGHT_LIMIT 4608 #define DECODE_HEIGHT_LIMIT 4608

2207
media/libaom/config/mac/x64/aom_dsp_rtcd.h
View file

File diff suppressed because it is too large Load diff

267
media/libaom/config/mac/x64/av1_rtcd.h
View file

@ -60,13 +60,7 @@ void aom_clpf_hblock_hbd_sse4_1(uint16_t *dst, const uint16_t *src, int dstride,
RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd); RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd);
int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); #define av1_block_error av1_block_error_c
int64_t av1_block_error_avx2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
RTCD_EXTERN int64_t (*av1_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_fp_c(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
int64_t av1_block_error_fp_sse2(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
#define av1_block_error_fp av1_block_error_fp_sse2
void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
@ -80,9 +74,7 @@ int av1_diamond_search_sad_c(struct macroblock *x, const struct search_site_conf
#define av1_diamond_search_sad av1_diamond_search_sad_c #define av1_diamond_search_sad av1_diamond_search_sad_c
void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fdct8x8_quant_sse2(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_fdct8x8_quant av1_fdct8x8_quant_c
void av1_fdct8x8_quant_ssse3(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
RTCD_EXTERN void (*av1_fdct8x8_quant)(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type);
@ -149,9 +141,174 @@ RTCD_EXTERN int (*av1_full_search_sad)(const struct macroblock *x, const struct
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type); void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type);
#define av1_fwd_idtx av1_fwd_idtx_c #define av1_fwd_idtx av1_fwd_idtx_c
void av1_fwd_txfm2d_16x16_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_16x16_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_16x16)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_32x32)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_4x4)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_64x64)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_8x8)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_fwht4x4 av1_fwht4x4_c #define av1_fwht4x4 av1_fwht4x4_c
int64_t av1_highbd_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
int64_t av1_highbd_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
#define av1_highbd_block_error av1_highbd_block_error_sse2
void av1_highbd_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8 av1_highbd_convolve8_sse2
void av1_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg av1_highbd_convolve8_avg_sse2
void av1_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_horiz av1_highbd_convolve8_avg_horiz_sse2
void av1_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_vert av1_highbd_convolve8_avg_vert_sse2
void av1_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_horiz av1_highbd_convolve8_horiz_sse2
void av1_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_vert av1_highbd_convolve8_vert_sse2
void av1_highbd_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_avg av1_highbd_convolve_avg_c
void av1_highbd_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_copy av1_highbd_convolve_copy_c
void av1_highbd_convolve_horiz_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_horiz_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_horiz)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_init_c(void);
void av1_highbd_convolve_init_sse4_1(void);
RTCD_EXTERN void (*av1_highbd_convolve_init)(void);
void av1_highbd_convolve_vert_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_vert_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_vert)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x16 av1_highbd_fht16x16_c
void av1_highbd_fht16x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x32 av1_highbd_fht16x32_c
void av1_highbd_fht16x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x4 av1_highbd_fht16x4_c
void av1_highbd_fht16x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x8 av1_highbd_fht16x8_c
void av1_highbd_fht32x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x16 av1_highbd_fht32x16_c
void av1_highbd_fht32x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x32 av1_highbd_fht32x32_c
void av1_highbd_fht32x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x8 av1_highbd_fht32x8_c
void av1_highbd_fht4x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x16 av1_highbd_fht4x16_c
void av1_highbd_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x4_sse4_1(const int16_t *input, tran_low_t *output, int stride, int tx_type);
RTCD_EXTERN void (*av1_highbd_fht4x4)(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x8 av1_highbd_fht4x8_c
void av1_highbd_fht8x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x16 av1_highbd_fht8x16_c
void av1_highbd_fht8x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x32 av1_highbd_fht8x32_c
void av1_highbd_fht8x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x4 av1_highbd_fht8x4_c
void av1_highbd_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x8 av1_highbd_fht8x8_c
void av1_highbd_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_highbd_fwht4x4 av1_highbd_fwht4x4_c
void av1_highbd_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd);
#define av1_highbd_iht16x16_256_add av1_highbd_iht16x16_256_add_c
void av1_highbd_iht16x32_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x32_512_add av1_highbd_iht16x32_512_add_c
void av1_highbd_iht16x4_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x4_64_add av1_highbd_iht16x4_64_add_c
void av1_highbd_iht16x8_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x8_128_add av1_highbd_iht16x8_128_add_c
void av1_highbd_iht32x16_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x16_512_add av1_highbd_iht32x16_512_add_c
void av1_highbd_iht32x8_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x8_256_add av1_highbd_iht32x8_256_add_c
void av1_highbd_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x16_64_add av1_highbd_iht4x16_64_add_c
void av1_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x4_16_add av1_highbd_iht4x4_16_add_c
void av1_highbd_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x8_32_add av1_highbd_iht4x8_32_add_c
void av1_highbd_iht8x16_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x16_128_add av1_highbd_iht8x16_128_add_c
void av1_highbd_iht8x32_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x32_256_add av1_highbd_iht8x32_256_add_c
void av1_highbd_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x4_32_add av1_highbd_iht8x4_32_add_c
void av1_highbd_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x8_64_add av1_highbd_iht8x8_64_add_c
void av1_highbd_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_b av1_highbd_quantize_b_c
void av1_highbd_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_quantize_fp_sse4_1(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
RTCD_EXTERN void (*av1_highbd_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_highbd_temporal_filter_apply av1_highbd_temporal_filter_apply_c
void av1_highbd_warp_affine_c(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_highbd_warp_affine_ssse3(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_highbd_warp_affine)(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
@ -204,6 +361,25 @@ void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride
void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
#define av1_iht8x8_64_add av1_iht8x8_64_add_sse2 #define av1_iht8x8_64_add av1_iht8x8_64_add_sse2
void av1_inv_txfm2d_add_16x16_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_16x16_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_16x16)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_avx2(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_32x32)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_4x4)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_64x64_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_64x64 av1_inv_txfm2d_add_64x64_c
void av1_inv_txfm2d_add_8x8_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_8x8_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_8x8)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_lowbd_convolve_init_c(void); void av1_lowbd_convolve_init_c(void);
void av1_lowbd_convolve_init_ssse3(void); void av1_lowbd_convolve_init_ssse3(void);
RTCD_EXTERN void (*av1_lowbd_convolve_init)(void); RTCD_EXTERN void (*av1_lowbd_convolve_init)(void);
@ -213,20 +389,35 @@ void av1_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_b
void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_sse2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_sse2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_ssse3(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_quantize_fp av1_quantize_fp_sse2
RTCD_EXTERN void (*av1_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_32x32_ssse3(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_quantize_fp_32x32 av1_quantize_fp_32x32_c
RTCD_EXTERN void (*av1_quantize_fp_32x32)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_temporal_filter_apply av1_temporal_filter_apply_sse2 #define av1_temporal_filter_apply av1_temporal_filter_apply_sse2
void av1_warp_affine_c(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_c(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_warp_affine_sse2(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_sse2(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
#define av1_warp_affine av1_warp_affine_sse2 void av1_warp_affine_ssse3(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_warp_affine)(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_wedge_compute_delta_squares_c(int16_t *d, const int16_t *a, const int16_t *b, int N);
void av1_wedge_compute_delta_squares_sse2(int16_t *d, const int16_t *a, const int16_t *b, int N);
#define av1_wedge_compute_delta_squares av1_wedge_compute_delta_squares_sse2
int av1_wedge_sign_from_residuals_c(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
int av1_wedge_sign_from_residuals_sse2(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
#define av1_wedge_sign_from_residuals av1_wedge_sign_from_residuals_sse2
uint64_t av1_wedge_sse_from_residuals_c(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
uint64_t av1_wedge_sse_from_residuals_sse2(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
#define av1_wedge_sse_from_residuals av1_wedge_sse_from_residuals_sse2
double compute_cross_correlation_c(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
double compute_cross_correlation_sse4_1(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
RTCD_EXTERN double (*compute_cross_correlation)(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
@ -304,14 +495,10 @@ static void setup_rtcd_internal(void)
aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2; aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2;
if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3; if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3;
if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1; if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1;
av1_block_error = av1_block_error_sse2;
if (flags & HAS_AVX2) av1_block_error = av1_block_error_avx2;
av1_convolve_horiz = av1_convolve_horiz_c; av1_convolve_horiz = av1_convolve_horiz_c;
if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3; if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3;
av1_convolve_vert = av1_convolve_vert_c; av1_convolve_vert = av1_convolve_vert_c;
if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3; if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3;
av1_fdct8x8_quant = av1_fdct8x8_quant_sse2;
if (flags & HAS_SSSE3) av1_fdct8x8_quant = av1_fdct8x8_quant_ssse3;
av1_fht16x16 = av1_fht16x16_sse2; av1_fht16x16 = av1_fht16x16_sse2;
if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2; if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2;
av1_fht32x32 = av1_fht32x32_sse2; av1_fht32x32 = av1_fht32x32_sse2;
@ -319,14 +506,44 @@ static void setup_rtcd_internal(void)
av1_full_search_sad = av1_full_search_sad_c; av1_full_search_sad = av1_full_search_sad_c;
if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3; if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3;
if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8; if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8;
av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_sse4_1;
av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_sse4_1;
av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_sse4_1;
av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_sse4_1;
av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_sse4_1;
av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_sse4_1;
av1_highbd_convolve_init = av1_highbd_convolve_init_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_init = av1_highbd_convolve_init_sse4_1;
av1_highbd_convolve_vert = av1_highbd_convolve_vert_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_vert = av1_highbd_convolve_vert_sse4_1;
av1_highbd_fht4x4 = av1_highbd_fht4x4_c;
if (flags & HAS_SSE4_1) av1_highbd_fht4x4 = av1_highbd_fht4x4_sse4_1;
av1_highbd_quantize_fp = av1_highbd_quantize_fp_c;
if (flags & HAS_SSE4_1) av1_highbd_quantize_fp = av1_highbd_quantize_fp_sse4_1;
av1_highbd_warp_affine = av1_highbd_warp_affine_c;
if (flags & HAS_SSSE3) av1_highbd_warp_affine = av1_highbd_warp_affine_ssse3;
av1_iht16x16_256_add = av1_iht16x16_256_add_sse2; av1_iht16x16_256_add = av1_iht16x16_256_add_sse2;
if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2; if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2;
av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_sse4_1;
av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_c;
if (flags & HAS_AVX2) av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_avx2;
av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_sse4_1;
av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_sse4_1;
av1_lowbd_convolve_init = av1_lowbd_convolve_init_c; av1_lowbd_convolve_init = av1_lowbd_convolve_init_c;
if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3; if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3;
av1_quantize_fp = av1_quantize_fp_sse2; av1_warp_affine = av1_warp_affine_sse2;
if (flags & HAS_SSSE3) av1_quantize_fp = av1_quantize_fp_ssse3; if (flags & HAS_SSSE3) av1_warp_affine = av1_warp_affine_ssse3;
av1_quantize_fp_32x32 = av1_quantize_fp_32x32_c; compute_cross_correlation = compute_cross_correlation_c;
if (flags & HAS_SSSE3) av1_quantize_fp_32x32 = av1_quantize_fp_32x32_ssse3; if (flags & HAS_SSE4_1) compute_cross_correlation = compute_cross_correlation_sse4_1;
copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2; copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2;
if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3; if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3;
if (flags & HAS_SSE4_1) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse4_1; if (flags & HAS_SSE4_1) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse4_1;

39
media/libaom/config/win/ia32/aom_config.asm
View file

@ -2,8 +2,6 @@
%define ARCH_MIPS 0 %define ARCH_MIPS 0
%define ARCH_X86 1 %define ARCH_X86 1
%define ARCH_X86_64 0 %define ARCH_X86_64 0
%define HAVE_EDSP 0
%define HAVE_MEDIA 0
%define HAVE_NEON 0 %define HAVE_NEON 0
%define HAVE_NEON_ASM 0 %define HAVE_NEON_ASM 0
%define HAVE_MIPS32 0 %define HAVE_MIPS32 0
@ -45,8 +43,6 @@
%define CONFIG_AV1_ENCODER 1 %define CONFIG_AV1_ENCODER 1
%define CONFIG_AV1_DECODER 1 %define CONFIG_AV1_DECODER 1
%define CONFIG_AV1 1 %define CONFIG_AV1 1
%define CONFIG_ENCODERS 1
%define CONFIG_DECODERS 1
%define CONFIG_STATIC_MSVCRT 0 %define CONFIG_STATIC_MSVCRT 0
%define CONFIG_SPATIAL_RESAMPLING 1 %define CONFIG_SPATIAL_RESAMPLING 1
%define CONFIG_REALTIME_ONLY 0 %define CONFIG_REALTIME_ONLY 0
@ -66,36 +62,35 @@
%define CONFIG_ENCODE_PERF_TESTS 0 %define CONFIG_ENCODE_PERF_TESTS 0
%define CONFIG_COEFFICIENT_RANGE_CHECKING 0 %define CONFIG_COEFFICIENT_RANGE_CHECKING 0
%define CONFIG_LOWBITDEPTH 1 %define CONFIG_LOWBITDEPTH 1
%define CONFIG_HIGHBITDEPTH 0 %define CONFIG_HIGHBITDEPTH 1
%define CONFIG_EXPERIMENTAL 0 %define CONFIG_EXPERIMENTAL 0
%define CONFIG_SIZE_LIMIT 1 %define CONFIG_SIZE_LIMIT 1
%define CONFIG_FP_MB_STATS 0 %define CONFIG_FP_MB_STATS 0
%define CONFIG_CDEF 1 %define CONFIG_CDEF 1
%define CONFIG_VAR_TX 0 %define CONFIG_VAR_TX 1
%define CONFIG_RECT_TX 1 %define CONFIG_RECT_TX 1
%define CONFIG_REF_MV 1 %define CONFIG_RECT_TX_EXT 0
%define CONFIG_TPL_MV 0 %define CONFIG_TPL_MV 0
%define CONFIG_DUAL_FILTER 1 %define CONFIG_DUAL_FILTER 1
%define CONFIG_CONVOLVE_ROUND 0 %define CONFIG_CONVOLVE_ROUND 0
%define CONFIG_COMPOUND_ROUND 0 %define CONFIG_COMPOUND_ROUND 0
%define CONFIG_EXT_TX 0 %define CONFIG_EXT_TX 1
%define CONFIG_DPCM_INTRA 0
%define CONFIG_TX64X64 0 %define CONFIG_TX64X64 0
%define CONFIG_SUB8X8_MC 0
%define CONFIG_EXT_INTRA 1 %define CONFIG_EXT_INTRA 1
%define CONFIG_INTRA_INTERP 0 %define CONFIG_INTRA_INTERP 0
%define CONFIG_FILTER_INTRA 0 %define CONFIG_FILTER_INTRA 0
%define CONFIG_INTRA_EDGE 0
%define CONFIG_INTRABC 0 %define CONFIG_INTRABC 0
%define CONFIG_EXT_INTER 0 %define CONFIG_EXT_INTER 1
%define CONFIG_INTERINTRA 0 %define CONFIG_INTERINTRA 1
%define CONFIG_WEDGE 0 %define CONFIG_WEDGE 1
%define CONFIG_COMPOUND_SEGMENT 0 %define CONFIG_COMPOUND_SEGMENT 1
%define CONFIG_EXT_REFS 1 %define CONFIG_EXT_REFS 1
%define CONFIG_GLOBAL_MOTION 1 %define CONFIG_GLOBAL_MOTION 1
%define CONFIG_NEW_QUANT 0 %define CONFIG_NEW_QUANT 0
%define CONFIG_SUPERTX 0 %define CONFIG_SUPERTX 0
%define CONFIG_ANS 0 %define CONFIG_ANS 0
%define CONFIG_EC_MULTISYMBOL 1
%define CONFIG_NEW_TOKENSET 1
%define CONFIG_LOOP_RESTORATION 0 %define CONFIG_LOOP_RESTORATION 0
%define CONFIG_EXT_PARTITION 0 %define CONFIG_EXT_PARTITION 0
%define CONFIG_EXT_PARTITION_TYPES 0 %define CONFIG_EXT_PARTITION_TYPES 0
@ -105,20 +100,18 @@
%define CONFIG_NCOBMC 0 %define CONFIG_NCOBMC 0
%define CONFIG_WARPED_MOTION 1 %define CONFIG_WARPED_MOTION 1
%define CONFIG_Q_ADAPT_PROBS 0 %define CONFIG_Q_ADAPT_PROBS 0
%define CONFIG_SUBFRAME_PROB_UPDATE 0
%define CONFIG_BITSTREAM_DEBUG 0 %define CONFIG_BITSTREAM_DEBUG 0
%define CONFIG_ALT_INTRA 1 %define CONFIG_ALT_INTRA 1
%define CONFIG_PALETTE 1 %define CONFIG_PALETTE 1
%define CONFIG_PALETTE_DELTA_ENCODING 0 %define CONFIG_PALETTE_DELTA_ENCODING 0
%define CONFIG_DAALA_EC 1
%define CONFIG_RAWBITS 0 %define CONFIG_RAWBITS 0
%define CONFIG_EC_SMALLMUL 0 %define CONFIG_EC_SMALLMUL 1
%define CONFIG_PVQ 0 %define CONFIG_PVQ 0
%define CONFIG_CFL 0 %define CONFIG_CFL 0
%define CONFIG_XIPHRC 0 %define CONFIG_XIPHRC 0
%define CONFIG_CB4X4 1 %define CONFIG_CB4X4 1
%define CONFIG_CHROMA_2X2 0 %define CONFIG_CHROMA_2X2 0
%define CONFIG_CHROMA_SUB8X8 0 %define CONFIG_CHROMA_SUB8X8 1
%define CONFIG_FRAME_SIZE 0 %define CONFIG_FRAME_SIZE 0
%define CONFIG_DELTA_Q 1 %define CONFIG_DELTA_Q 1
%define CONFIG_EXT_DELTA_Q 1 %define CONFIG_EXT_DELTA_Q 1
@ -126,10 +119,10 @@
%define CONFIG_FILTER_7BIT 1 %define CONFIG_FILTER_7BIT 1
%define CONFIG_PARALLEL_DEBLOCKING 0 %define CONFIG_PARALLEL_DEBLOCKING 0
%define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 %define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
%define CONFIG_LOOPFILTERING_ACROSS_TILES 0 %define CONFIG_LOOPFILTERING_ACROSS_TILES 1
%define CONFIG_TILE_GROUPS 1 %define CONFIG_TILE_GROUPS 1
%define CONFIG_EC_ADAPT 1 %define CONFIG_EC_ADAPT 1
%define CONFIG_TEMPMV_SIGNALING 0 %define CONFIG_TEMPMV_SIGNALING 1
%define CONFIG_RD_DEBUG 0 %define CONFIG_RD_DEBUG 0
%define CONFIG_REFERENCE_BUFFER 1 %define CONFIG_REFERENCE_BUFFER 1
%define CONFIG_COEF_INTERLEAVE 0 %define CONFIG_COEF_INTERLEAVE 0
@ -147,5 +140,7 @@
%define CONFIG_NEW_MULTISYMBOL 0 %define CONFIG_NEW_MULTISYMBOL 0
%define CONFIG_COMPOUND_SINGLEREF 0 %define CONFIG_COMPOUND_SINGLEREF 0
%define CONFIG_AOM_QM 0 %define CONFIG_AOM_QM 0
%define CONFIG_LOWDELAY_COMPOUND 0 %define CONFIG_ONE_SIDED_COMPOUND 1
%define CONFIG_SMOOTH_HV 0
%define CONFIG_VAR_REFS 0
%define CONFIG_ANALYZER 0 %define CONFIG_ANALYZER 0

39
media/libaom/config/win/ia32/aom_config.h
View file

@ -15,8 +15,6 @@
#define ARCH_MIPS 0 #define ARCH_MIPS 0
#define ARCH_X86 1 #define ARCH_X86 1
#define ARCH_X86_64 0 #define ARCH_X86_64 0
#define HAVE_EDSP 0
#define HAVE_MEDIA 0
#define HAVE_NEON 0 #define HAVE_NEON 0
#define HAVE_NEON_ASM 0 #define HAVE_NEON_ASM 0
#define HAVE_MIPS32 0 #define HAVE_MIPS32 0
@ -58,8 +56,6 @@
#define CONFIG_AV1_ENCODER 1 #define CONFIG_AV1_ENCODER 1
#define CONFIG_AV1_DECODER 1 #define CONFIG_AV1_DECODER 1
#define CONFIG_AV1 1 #define CONFIG_AV1 1
#define CONFIG_ENCODERS 1
#define CONFIG_DECODERS 1
#define CONFIG_STATIC_MSVCRT 0 #define CONFIG_STATIC_MSVCRT 0
#define CONFIG_SPATIAL_RESAMPLING 1 #define CONFIG_SPATIAL_RESAMPLING 1
#define CONFIG_REALTIME_ONLY 0 #define CONFIG_REALTIME_ONLY 0
@ -79,36 +75,35 @@
#define CONFIG_ENCODE_PERF_TESTS 0 #define CONFIG_ENCODE_PERF_TESTS 0
#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 #define CONFIG_COEFFICIENT_RANGE_CHECKING 0
#define CONFIG_LOWBITDEPTH 1 #define CONFIG_LOWBITDEPTH 1
#define CONFIG_HIGHBITDEPTH 0 #define CONFIG_HIGHBITDEPTH 1
#define CONFIG_EXPERIMENTAL 0 #define CONFIG_EXPERIMENTAL 0
#define CONFIG_SIZE_LIMIT 1 #define CONFIG_SIZE_LIMIT 1
#define CONFIG_FP_MB_STATS 0 #define CONFIG_FP_MB_STATS 0
#define CONFIG_CDEF 1 #define CONFIG_CDEF 1
#define CONFIG_VAR_TX 0 #define CONFIG_VAR_TX 1
#define CONFIG_RECT_TX 1 #define CONFIG_RECT_TX 1
#define CONFIG_REF_MV 1 #define CONFIG_RECT_TX_EXT 0
#define CONFIG_TPL_MV 0 #define CONFIG_TPL_MV 0
#define CONFIG_DUAL_FILTER 1 #define CONFIG_DUAL_FILTER 1
#define CONFIG_CONVOLVE_ROUND 0 #define CONFIG_CONVOLVE_ROUND 0
#define CONFIG_COMPOUND_ROUND 0 #define CONFIG_COMPOUND_ROUND 0
#define CONFIG_EXT_TX 0 #define CONFIG_EXT_TX 1
#define CONFIG_DPCM_INTRA 0
#define CONFIG_TX64X64 0 #define CONFIG_TX64X64 0
#define CONFIG_SUB8X8_MC 0
#define CONFIG_EXT_INTRA 1 #define CONFIG_EXT_INTRA 1
#define CONFIG_INTRA_INTERP 0 #define CONFIG_INTRA_INTERP 0
#define CONFIG_FILTER_INTRA 0 #define CONFIG_FILTER_INTRA 0
#define CONFIG_INTRA_EDGE 0
#define CONFIG_INTRABC 0 #define CONFIG_INTRABC 0
#define CONFIG_EXT_INTER 0 #define CONFIG_EXT_INTER 1
#define CONFIG_INTERINTRA 0 #define CONFIG_INTERINTRA 1
#define CONFIG_WEDGE 0 #define CONFIG_WEDGE 1
#define CONFIG_COMPOUND_SEGMENT 0 #define CONFIG_COMPOUND_SEGMENT 1
#define CONFIG_EXT_REFS 1 #define CONFIG_EXT_REFS 1
#define CONFIG_GLOBAL_MOTION 1 #define CONFIG_GLOBAL_MOTION 1
#define CONFIG_NEW_QUANT 0 #define CONFIG_NEW_QUANT 0
#define CONFIG_SUPERTX 0 #define CONFIG_SUPERTX 0
#define CONFIG_ANS 0 #define CONFIG_ANS 0
#define CONFIG_EC_MULTISYMBOL 1
#define CONFIG_NEW_TOKENSET 1
#define CONFIG_LOOP_RESTORATION 0 #define CONFIG_LOOP_RESTORATION 0
#define CONFIG_EXT_PARTITION 0 #define CONFIG_EXT_PARTITION 0
#define CONFIG_EXT_PARTITION_TYPES 0 #define CONFIG_EXT_PARTITION_TYPES 0
@ -118,20 +113,18 @@
#define CONFIG_NCOBMC 0 #define CONFIG_NCOBMC 0
#define CONFIG_WARPED_MOTION 1 #define CONFIG_WARPED_MOTION 1
#define CONFIG_Q_ADAPT_PROBS 0 #define CONFIG_Q_ADAPT_PROBS 0
#define CONFIG_SUBFRAME_PROB_UPDATE 0
#define CONFIG_BITSTREAM_DEBUG 0 #define CONFIG_BITSTREAM_DEBUG 0
#define CONFIG_ALT_INTRA 1 #define CONFIG_ALT_INTRA 1
#define CONFIG_PALETTE 1 #define CONFIG_PALETTE 1
#define CONFIG_PALETTE_DELTA_ENCODING 0 #define CONFIG_PALETTE_DELTA_ENCODING 0
#define CONFIG_DAALA_EC 1
#define CONFIG_RAWBITS 0 #define CONFIG_RAWBITS 0
#define CONFIG_EC_SMALLMUL 0 #define CONFIG_EC_SMALLMUL 1
#define CONFIG_PVQ 0 #define CONFIG_PVQ 0
#define CONFIG_CFL 0 #define CONFIG_CFL 0
#define CONFIG_XIPHRC 0 #define CONFIG_XIPHRC 0
#define CONFIG_CB4X4 1 #define CONFIG_CB4X4 1
#define CONFIG_CHROMA_2X2 0 #define CONFIG_CHROMA_2X2 0
#define CONFIG_CHROMA_SUB8X8 0 #define CONFIG_CHROMA_SUB8X8 1
#define CONFIG_FRAME_SIZE 0 #define CONFIG_FRAME_SIZE 0
#define CONFIG_DELTA_Q 1 #define CONFIG_DELTA_Q 1
#define CONFIG_EXT_DELTA_Q 1 #define CONFIG_EXT_DELTA_Q 1
@ -139,10 +132,10 @@
#define CONFIG_FILTER_7BIT 1 #define CONFIG_FILTER_7BIT 1
#define CONFIG_PARALLEL_DEBLOCKING 0 #define CONFIG_PARALLEL_DEBLOCKING 0
#define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 #define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
#define CONFIG_LOOPFILTERING_ACROSS_TILES 0 #define CONFIG_LOOPFILTERING_ACROSS_TILES 1
#define CONFIG_TILE_GROUPS 1 #define CONFIG_TILE_GROUPS 1
#define CONFIG_EC_ADAPT 1 #define CONFIG_EC_ADAPT 1
#define CONFIG_TEMPMV_SIGNALING 0 #define CONFIG_TEMPMV_SIGNALING 1
#define CONFIG_RD_DEBUG 0 #define CONFIG_RD_DEBUG 0
#define CONFIG_REFERENCE_BUFFER 1 #define CONFIG_REFERENCE_BUFFER 1
#define CONFIG_COEF_INTERLEAVE 0 #define CONFIG_COEF_INTERLEAVE 0
@ -160,7 +153,9 @@
#define CONFIG_NEW_MULTISYMBOL 0 #define CONFIG_NEW_MULTISYMBOL 0
#define CONFIG_COMPOUND_SINGLEREF 0 #define CONFIG_COMPOUND_SINGLEREF 0
#define CONFIG_AOM_QM 0 #define CONFIG_AOM_QM 0
#define CONFIG_LOWDELAY_COMPOUND 0 #define CONFIG_ONE_SIDED_COMPOUND 1
#define CONFIG_SMOOTH_HV 0
#define CONFIG_VAR_REFS 0
#define CONFIG_ANALYZER 0 #define CONFIG_ANALYZER 0
#define DECODE_WIDTH_LIMIT 8192 #define DECODE_WIDTH_LIMIT 8192
#define DECODE_HEIGHT_LIMIT 4608 #define DECODE_HEIGHT_LIMIT 4608

2507
media/libaom/config/win/ia32/aom_dsp_rtcd.h
View file

File diff suppressed because it is too large Load diff

262
media/libaom/config/win/ia32/av1_rtcd.h
View file

@ -60,13 +60,7 @@ void aom_clpf_hblock_hbd_sse4_1(uint16_t *dst, const uint16_t *src, int dstride,
RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd); RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd);
int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); #define av1_block_error av1_block_error_c
int64_t av1_block_error_avx2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
RTCD_EXTERN int64_t (*av1_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_fp_c(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
int64_t av1_block_error_fp_sse2(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
RTCD_EXTERN int64_t (*av1_block_error_fp)(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
@ -80,9 +74,7 @@ int av1_diamond_search_sad_c(struct macroblock *x, const struct search_site_conf
#define av1_diamond_search_sad av1_diamond_search_sad_c #define av1_diamond_search_sad av1_diamond_search_sad_c
void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fdct8x8_quant_sse2(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_fdct8x8_quant av1_fdct8x8_quant_c
void av1_fdct8x8_quant_ssse3(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
RTCD_EXTERN void (*av1_fdct8x8_quant)(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type);
@ -149,9 +141,168 @@ RTCD_EXTERN int (*av1_full_search_sad)(const struct macroblock *x, const struct
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type); void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type);
#define av1_fwd_idtx av1_fwd_idtx_c #define av1_fwd_idtx av1_fwd_idtx_c
void av1_fwd_txfm2d_16x16_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_16x16_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_16x16)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_32x32)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_4x4)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_64x64)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_8x8)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_fwht4x4 av1_fwht4x4_c #define av1_fwht4x4 av1_fwht4x4_c
int64_t av1_highbd_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
int64_t av1_highbd_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
RTCD_EXTERN int64_t (*av1_highbd_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
void av1_highbd_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8 av1_highbd_convolve8_c
void av1_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg av1_highbd_convolve8_avg_c
void av1_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_horiz av1_highbd_convolve8_avg_horiz_c
void av1_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_vert av1_highbd_convolve8_avg_vert_c
void av1_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_horiz av1_highbd_convolve8_horiz_c
void av1_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_vert av1_highbd_convolve8_vert_c
void av1_highbd_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_avg av1_highbd_convolve_avg_c
void av1_highbd_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_copy av1_highbd_convolve_copy_c
void av1_highbd_convolve_horiz_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_horiz_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_horiz)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_init_c(void);
void av1_highbd_convolve_init_sse4_1(void);
RTCD_EXTERN void (*av1_highbd_convolve_init)(void);
void av1_highbd_convolve_vert_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_vert_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_vert)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x16 av1_highbd_fht16x16_c
void av1_highbd_fht16x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x32 av1_highbd_fht16x32_c
void av1_highbd_fht16x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x4 av1_highbd_fht16x4_c
void av1_highbd_fht16x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x8 av1_highbd_fht16x8_c
void av1_highbd_fht32x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x16 av1_highbd_fht32x16_c
void av1_highbd_fht32x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x32 av1_highbd_fht32x32_c
void av1_highbd_fht32x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x8 av1_highbd_fht32x8_c
void av1_highbd_fht4x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x16 av1_highbd_fht4x16_c
void av1_highbd_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x4_sse4_1(const int16_t *input, tran_low_t *output, int stride, int tx_type);
RTCD_EXTERN void (*av1_highbd_fht4x4)(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x8 av1_highbd_fht4x8_c
void av1_highbd_fht8x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x16 av1_highbd_fht8x16_c
void av1_highbd_fht8x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x32 av1_highbd_fht8x32_c
void av1_highbd_fht8x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x4 av1_highbd_fht8x4_c
void av1_highbd_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x8 av1_highbd_fht8x8_c
void av1_highbd_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_highbd_fwht4x4 av1_highbd_fwht4x4_c
void av1_highbd_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd);
#define av1_highbd_iht16x16_256_add av1_highbd_iht16x16_256_add_c
void av1_highbd_iht16x32_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x32_512_add av1_highbd_iht16x32_512_add_c
void av1_highbd_iht16x4_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x4_64_add av1_highbd_iht16x4_64_add_c
void av1_highbd_iht16x8_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x8_128_add av1_highbd_iht16x8_128_add_c
void av1_highbd_iht32x16_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x16_512_add av1_highbd_iht32x16_512_add_c
void av1_highbd_iht32x8_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x8_256_add av1_highbd_iht32x8_256_add_c
void av1_highbd_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x16_64_add av1_highbd_iht4x16_64_add_c
void av1_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x4_16_add av1_highbd_iht4x4_16_add_c
void av1_highbd_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x8_32_add av1_highbd_iht4x8_32_add_c
void av1_highbd_iht8x16_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x16_128_add av1_highbd_iht8x16_128_add_c
void av1_highbd_iht8x32_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x32_256_add av1_highbd_iht8x32_256_add_c
void av1_highbd_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x4_32_add av1_highbd_iht8x4_32_add_c
void av1_highbd_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x8_64_add av1_highbd_iht8x8_64_add_c
void av1_highbd_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_b av1_highbd_quantize_b_c
void av1_highbd_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_quantize_fp_sse4_1(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
RTCD_EXTERN void (*av1_highbd_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_highbd_temporal_filter_apply av1_highbd_temporal_filter_apply_c
void av1_highbd_warp_affine_c(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_highbd_warp_affine_ssse3(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_highbd_warp_affine)(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
@ -204,6 +355,25 @@ void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride
void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
RTCD_EXTERN void (*av1_iht8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); RTCD_EXTERN void (*av1_iht8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
void av1_inv_txfm2d_add_16x16_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_16x16_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_16x16)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_avx2(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_32x32)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_4x4)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_64x64_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_64x64 av1_inv_txfm2d_add_64x64_c
void av1_inv_txfm2d_add_8x8_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_8x8_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_8x8)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_lowbd_convolve_init_c(void); void av1_lowbd_convolve_init_c(void);
void av1_lowbd_convolve_init_ssse3(void); void av1_lowbd_convolve_init_ssse3(void);
RTCD_EXTERN void (*av1_lowbd_convolve_init)(void); RTCD_EXTERN void (*av1_lowbd_convolve_init)(void);
@ -222,9 +392,26 @@ void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *
void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
RTCD_EXTERN void (*av1_temporal_filter_apply)(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); RTCD_EXTERN void (*av1_temporal_filter_apply)(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
void av1_warp_affine_c(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_c(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_warp_affine_sse2(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_sse2(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_warp_affine)(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_ssse3(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_warp_affine)(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_wedge_compute_delta_squares_c(int16_t *d, const int16_t *a, const int16_t *b, int N);
void av1_wedge_compute_delta_squares_sse2(int16_t *d, const int16_t *a, const int16_t *b, int N);
RTCD_EXTERN void (*av1_wedge_compute_delta_squares)(int16_t *d, const int16_t *a, const int16_t *b, int N);
int av1_wedge_sign_from_residuals_c(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
int av1_wedge_sign_from_residuals_sse2(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
RTCD_EXTERN int (*av1_wedge_sign_from_residuals)(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
uint64_t av1_wedge_sse_from_residuals_c(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
uint64_t av1_wedge_sse_from_residuals_sse2(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
RTCD_EXTERN uint64_t (*av1_wedge_sse_from_residuals)(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
double compute_cross_correlation_c(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
double compute_cross_correlation_sse4_1(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
RTCD_EXTERN double (*compute_cross_correlation)(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
@ -306,18 +493,10 @@ static void setup_rtcd_internal(void)
if (flags & HAS_SSE2) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2; if (flags & HAS_SSE2) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2;
if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3; if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3;
if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1; if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1;
av1_block_error = av1_block_error_c;
if (flags & HAS_SSE2) av1_block_error = av1_block_error_sse2;
if (flags & HAS_AVX2) av1_block_error = av1_block_error_avx2;
av1_block_error_fp = av1_block_error_fp_c;
if (flags & HAS_SSE2) av1_block_error_fp = av1_block_error_fp_sse2;
av1_convolve_horiz = av1_convolve_horiz_c; av1_convolve_horiz = av1_convolve_horiz_c;
if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3; if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3;
av1_convolve_vert = av1_convolve_vert_c; av1_convolve_vert = av1_convolve_vert_c;
if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3; if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3;
av1_fdct8x8_quant = av1_fdct8x8_quant_c;
if (flags & HAS_SSE2) av1_fdct8x8_quant = av1_fdct8x8_quant_sse2;
if (flags & HAS_SSSE3) av1_fdct8x8_quant = av1_fdct8x8_quant_ssse3;
av1_fht16x16 = av1_fht16x16_c; av1_fht16x16 = av1_fht16x16_c;
if (flags & HAS_SSE2) av1_fht16x16 = av1_fht16x16_sse2; if (flags & HAS_SSE2) av1_fht16x16 = av1_fht16x16_sse2;
if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2; if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2;
@ -343,6 +522,30 @@ static void setup_rtcd_internal(void)
av1_full_search_sad = av1_full_search_sad_c; av1_full_search_sad = av1_full_search_sad_c;
if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3; if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3;
if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8; if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8;
av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_sse4_1;
av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_sse4_1;
av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_sse4_1;
av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_sse4_1;
av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_sse4_1;
av1_highbd_block_error = av1_highbd_block_error_c;
if (flags & HAS_SSE2) av1_highbd_block_error = av1_highbd_block_error_sse2;
av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_sse4_1;
av1_highbd_convolve_init = av1_highbd_convolve_init_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_init = av1_highbd_convolve_init_sse4_1;
av1_highbd_convolve_vert = av1_highbd_convolve_vert_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_vert = av1_highbd_convolve_vert_sse4_1;
av1_highbd_fht4x4 = av1_highbd_fht4x4_c;
if (flags & HAS_SSE4_1) av1_highbd_fht4x4 = av1_highbd_fht4x4_sse4_1;
av1_highbd_quantize_fp = av1_highbd_quantize_fp_c;
if (flags & HAS_SSE4_1) av1_highbd_quantize_fp = av1_highbd_quantize_fp_sse4_1;
av1_highbd_warp_affine = av1_highbd_warp_affine_c;
if (flags & HAS_SSSE3) av1_highbd_warp_affine = av1_highbd_warp_affine_ssse3;
av1_iht16x16_256_add = av1_iht16x16_256_add_c; av1_iht16x16_256_add = av1_iht16x16_256_add_c;
if (flags & HAS_SSE2) av1_iht16x16_256_add = av1_iht16x16_256_add_sse2; if (flags & HAS_SSE2) av1_iht16x16_256_add = av1_iht16x16_256_add_sse2;
if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2; if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2;
@ -362,6 +565,14 @@ static void setup_rtcd_internal(void)
if (flags & HAS_SSE2) av1_iht8x4_32_add = av1_iht8x4_32_add_sse2; if (flags & HAS_SSE2) av1_iht8x4_32_add = av1_iht8x4_32_add_sse2;
av1_iht8x8_64_add = av1_iht8x8_64_add_c; av1_iht8x8_64_add = av1_iht8x8_64_add_c;
if (flags & HAS_SSE2) av1_iht8x8_64_add = av1_iht8x8_64_add_sse2; if (flags & HAS_SSE2) av1_iht8x8_64_add = av1_iht8x8_64_add_sse2;
av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_sse4_1;
av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_c;
if (flags & HAS_AVX2) av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_avx2;
av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_sse4_1;
av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_sse4_1;
av1_lowbd_convolve_init = av1_lowbd_convolve_init_c; av1_lowbd_convolve_init = av1_lowbd_convolve_init_c;
if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3; if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3;
av1_quantize_fp = av1_quantize_fp_c; av1_quantize_fp = av1_quantize_fp_c;
@ -370,6 +581,15 @@ static void setup_rtcd_internal(void)
if (flags & HAS_SSE2) av1_temporal_filter_apply = av1_temporal_filter_apply_sse2; if (flags & HAS_SSE2) av1_temporal_filter_apply = av1_temporal_filter_apply_sse2;
av1_warp_affine = av1_warp_affine_c; av1_warp_affine = av1_warp_affine_c;
if (flags & HAS_SSE2) av1_warp_affine = av1_warp_affine_sse2; if (flags & HAS_SSE2) av1_warp_affine = av1_warp_affine_sse2;
if (flags & HAS_SSSE3) av1_warp_affine = av1_warp_affine_ssse3;
av1_wedge_compute_delta_squares = av1_wedge_compute_delta_squares_c;
if (flags & HAS_SSE2) av1_wedge_compute_delta_squares = av1_wedge_compute_delta_squares_sse2;
av1_wedge_sign_from_residuals = av1_wedge_sign_from_residuals_c;
if (flags & HAS_SSE2) av1_wedge_sign_from_residuals = av1_wedge_sign_from_residuals_sse2;
av1_wedge_sse_from_residuals = av1_wedge_sse_from_residuals_c;
if (flags & HAS_SSE2) av1_wedge_sse_from_residuals = av1_wedge_sse_from_residuals_sse2;
compute_cross_correlation = compute_cross_correlation_c;
if (flags & HAS_SSE4_1) compute_cross_correlation = compute_cross_correlation_sse4_1;
copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_c; copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_c;
if (flags & HAS_SSE2) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2; if (flags & HAS_SSE2) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2;
if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3; if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3;

39
media/libaom/config/win/x64/aom_config.asm
View file

@ -2,8 +2,6 @@
%define ARCH_MIPS 0 %define ARCH_MIPS 0
%define ARCH_X86 0 %define ARCH_X86 0
%define ARCH_X86_64 1 %define ARCH_X86_64 1
%define HAVE_EDSP 0
%define HAVE_MEDIA 0
%define HAVE_NEON 0 %define HAVE_NEON 0
%define HAVE_NEON_ASM 0 %define HAVE_NEON_ASM 0
%define HAVE_MIPS32 0 %define HAVE_MIPS32 0
@ -45,8 +43,6 @@
%define CONFIG_AV1_ENCODER 1 %define CONFIG_AV1_ENCODER 1
%define CONFIG_AV1_DECODER 1 %define CONFIG_AV1_DECODER 1
%define CONFIG_AV1 1 %define CONFIG_AV1 1
%define CONFIG_ENCODERS 1
%define CONFIG_DECODERS 1
%define CONFIG_STATIC_MSVCRT 0 %define CONFIG_STATIC_MSVCRT 0
%define CONFIG_SPATIAL_RESAMPLING 1 %define CONFIG_SPATIAL_RESAMPLING 1
%define CONFIG_REALTIME_ONLY 0 %define CONFIG_REALTIME_ONLY 0
@ -66,36 +62,35 @@
%define CONFIG_ENCODE_PERF_TESTS 0 %define CONFIG_ENCODE_PERF_TESTS 0
%define CONFIG_COEFFICIENT_RANGE_CHECKING 0 %define CONFIG_COEFFICIENT_RANGE_CHECKING 0
%define CONFIG_LOWBITDEPTH 1 %define CONFIG_LOWBITDEPTH 1
%define CONFIG_HIGHBITDEPTH 0 %define CONFIG_HIGHBITDEPTH 1
%define CONFIG_EXPERIMENTAL 0 %define CONFIG_EXPERIMENTAL 0
%define CONFIG_SIZE_LIMIT 1 %define CONFIG_SIZE_LIMIT 1
%define CONFIG_FP_MB_STATS 0 %define CONFIG_FP_MB_STATS 0
%define CONFIG_CDEF 1 %define CONFIG_CDEF 1
%define CONFIG_VAR_TX 0 %define CONFIG_VAR_TX 1
%define CONFIG_RECT_TX 1 %define CONFIG_RECT_TX 1
%define CONFIG_REF_MV 1 %define CONFIG_RECT_TX_EXT 0
%define CONFIG_TPL_MV 0 %define CONFIG_TPL_MV 0
%define CONFIG_DUAL_FILTER 1 %define CONFIG_DUAL_FILTER 1
%define CONFIG_CONVOLVE_ROUND 0 %define CONFIG_CONVOLVE_ROUND 0
%define CONFIG_COMPOUND_ROUND 0 %define CONFIG_COMPOUND_ROUND 0
%define CONFIG_EXT_TX 0 %define CONFIG_EXT_TX 1
%define CONFIG_DPCM_INTRA 0
%define CONFIG_TX64X64 0 %define CONFIG_TX64X64 0
%define CONFIG_SUB8X8_MC 0
%define CONFIG_EXT_INTRA 1 %define CONFIG_EXT_INTRA 1
%define CONFIG_INTRA_INTERP 0 %define CONFIG_INTRA_INTERP 0
%define CONFIG_FILTER_INTRA 0 %define CONFIG_FILTER_INTRA 0
%define CONFIG_INTRA_EDGE 0
%define CONFIG_INTRABC 0 %define CONFIG_INTRABC 0
%define CONFIG_EXT_INTER 0 %define CONFIG_EXT_INTER 1
%define CONFIG_INTERINTRA 0 %define CONFIG_INTERINTRA 1
%define CONFIG_WEDGE 0 %define CONFIG_WEDGE 1
%define CONFIG_COMPOUND_SEGMENT 0 %define CONFIG_COMPOUND_SEGMENT 1
%define CONFIG_EXT_REFS 1 %define CONFIG_EXT_REFS 1
%define CONFIG_GLOBAL_MOTION 1 %define CONFIG_GLOBAL_MOTION 1
%define CONFIG_NEW_QUANT 0 %define CONFIG_NEW_QUANT 0
%define CONFIG_SUPERTX 0 %define CONFIG_SUPERTX 0
%define CONFIG_ANS 0 %define CONFIG_ANS 0
%define CONFIG_EC_MULTISYMBOL 1
%define CONFIG_NEW_TOKENSET 1
%define CONFIG_LOOP_RESTORATION 0 %define CONFIG_LOOP_RESTORATION 0
%define CONFIG_EXT_PARTITION 0 %define CONFIG_EXT_PARTITION 0
%define CONFIG_EXT_PARTITION_TYPES 0 %define CONFIG_EXT_PARTITION_TYPES 0
@ -105,20 +100,18 @@
%define CONFIG_NCOBMC 0 %define CONFIG_NCOBMC 0
%define CONFIG_WARPED_MOTION 1 %define CONFIG_WARPED_MOTION 1
%define CONFIG_Q_ADAPT_PROBS 0 %define CONFIG_Q_ADAPT_PROBS 0
%define CONFIG_SUBFRAME_PROB_UPDATE 0
%define CONFIG_BITSTREAM_DEBUG 0 %define CONFIG_BITSTREAM_DEBUG 0
%define CONFIG_ALT_INTRA 1 %define CONFIG_ALT_INTRA 1
%define CONFIG_PALETTE 1 %define CONFIG_PALETTE 1
%define CONFIG_PALETTE_DELTA_ENCODING 0 %define CONFIG_PALETTE_DELTA_ENCODING 0
%define CONFIG_DAALA_EC 1
%define CONFIG_RAWBITS 0 %define CONFIG_RAWBITS 0
%define CONFIG_EC_SMALLMUL 0 %define CONFIG_EC_SMALLMUL 1
%define CONFIG_PVQ 0 %define CONFIG_PVQ 0
%define CONFIG_CFL 0 %define CONFIG_CFL 0
%define CONFIG_XIPHRC 0 %define CONFIG_XIPHRC 0
%define CONFIG_CB4X4 1 %define CONFIG_CB4X4 1
%define CONFIG_CHROMA_2X2 0 %define CONFIG_CHROMA_2X2 0
%define CONFIG_CHROMA_SUB8X8 0 %define CONFIG_CHROMA_SUB8X8 1
%define CONFIG_FRAME_SIZE 0 %define CONFIG_FRAME_SIZE 0
%define CONFIG_DELTA_Q 1 %define CONFIG_DELTA_Q 1
%define CONFIG_EXT_DELTA_Q 1 %define CONFIG_EXT_DELTA_Q 1
@ -126,10 +119,10 @@
%define CONFIG_FILTER_7BIT 1 %define CONFIG_FILTER_7BIT 1
%define CONFIG_PARALLEL_DEBLOCKING 0 %define CONFIG_PARALLEL_DEBLOCKING 0
%define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 %define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
%define CONFIG_LOOPFILTERING_ACROSS_TILES 0 %define CONFIG_LOOPFILTERING_ACROSS_TILES 1
%define CONFIG_TILE_GROUPS 1 %define CONFIG_TILE_GROUPS 1
%define CONFIG_EC_ADAPT 1 %define CONFIG_EC_ADAPT 1
%define CONFIG_TEMPMV_SIGNALING 0 %define CONFIG_TEMPMV_SIGNALING 1
%define CONFIG_RD_DEBUG 0 %define CONFIG_RD_DEBUG 0
%define CONFIG_REFERENCE_BUFFER 1 %define CONFIG_REFERENCE_BUFFER 1
%define CONFIG_COEF_INTERLEAVE 0 %define CONFIG_COEF_INTERLEAVE 0
@ -147,5 +140,7 @@
%define CONFIG_NEW_MULTISYMBOL 0 %define CONFIG_NEW_MULTISYMBOL 0
%define CONFIG_COMPOUND_SINGLEREF 0 %define CONFIG_COMPOUND_SINGLEREF 0
%define CONFIG_AOM_QM 0 %define CONFIG_AOM_QM 0
%define CONFIG_LOWDELAY_COMPOUND 0 %define CONFIG_ONE_SIDED_COMPOUND 1
%define CONFIG_SMOOTH_HV 0
%define CONFIG_VAR_REFS 0
%define CONFIG_ANALYZER 0 %define CONFIG_ANALYZER 0

39
media/libaom/config/win/x64/aom_config.h
View file

@ -15,8 +15,6 @@
#define ARCH_MIPS 0 #define ARCH_MIPS 0
#define ARCH_X86 0 #define ARCH_X86 0
#define ARCH_X86_64 1 #define ARCH_X86_64 1
#define HAVE_EDSP 0
#define HAVE_MEDIA 0
#define HAVE_NEON 0 #define HAVE_NEON 0
#define HAVE_NEON_ASM 0 #define HAVE_NEON_ASM 0
#define HAVE_MIPS32 0 #define HAVE_MIPS32 0
@ -58,8 +56,6 @@
#define CONFIG_AV1_ENCODER 1 #define CONFIG_AV1_ENCODER 1
#define CONFIG_AV1_DECODER 1 #define CONFIG_AV1_DECODER 1
#define CONFIG_AV1 1 #define CONFIG_AV1 1
#define CONFIG_ENCODERS 1
#define CONFIG_DECODERS 1
#define CONFIG_STATIC_MSVCRT 0 #define CONFIG_STATIC_MSVCRT 0
#define CONFIG_SPATIAL_RESAMPLING 1 #define CONFIG_SPATIAL_RESAMPLING 1
#define CONFIG_REALTIME_ONLY 0 #define CONFIG_REALTIME_ONLY 0
@ -79,36 +75,35 @@
#define CONFIG_ENCODE_PERF_TESTS 0 #define CONFIG_ENCODE_PERF_TESTS 0
#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 #define CONFIG_COEFFICIENT_RANGE_CHECKING 0
#define CONFIG_LOWBITDEPTH 1 #define CONFIG_LOWBITDEPTH 1
#define CONFIG_HIGHBITDEPTH 0 #define CONFIG_HIGHBITDEPTH 1
#define CONFIG_EXPERIMENTAL 0 #define CONFIG_EXPERIMENTAL 0
#define CONFIG_SIZE_LIMIT 1 #define CONFIG_SIZE_LIMIT 1
#define CONFIG_FP_MB_STATS 0 #define CONFIG_FP_MB_STATS 0
#define CONFIG_CDEF 1 #define CONFIG_CDEF 1
#define CONFIG_VAR_TX 0 #define CONFIG_VAR_TX 1
#define CONFIG_RECT_TX 1 #define CONFIG_RECT_TX 1
#define CONFIG_REF_MV 1 #define CONFIG_RECT_TX_EXT 0
#define CONFIG_TPL_MV 0 #define CONFIG_TPL_MV 0
#define CONFIG_DUAL_FILTER 1 #define CONFIG_DUAL_FILTER 1
#define CONFIG_CONVOLVE_ROUND 0 #define CONFIG_CONVOLVE_ROUND 0
#define CONFIG_COMPOUND_ROUND 0 #define CONFIG_COMPOUND_ROUND 0
#define CONFIG_EXT_TX 0 #define CONFIG_EXT_TX 1
#define CONFIG_DPCM_INTRA 0
#define CONFIG_TX64X64 0 #define CONFIG_TX64X64 0
#define CONFIG_SUB8X8_MC 0
#define CONFIG_EXT_INTRA 1 #define CONFIG_EXT_INTRA 1
#define CONFIG_INTRA_INTERP 0 #define CONFIG_INTRA_INTERP 0
#define CONFIG_FILTER_INTRA 0 #define CONFIG_FILTER_INTRA 0
#define CONFIG_INTRA_EDGE 0
#define CONFIG_INTRABC 0 #define CONFIG_INTRABC 0
#define CONFIG_EXT_INTER 0 #define CONFIG_EXT_INTER 1
#define CONFIG_INTERINTRA 0 #define CONFIG_INTERINTRA 1
#define CONFIG_WEDGE 0 #define CONFIG_WEDGE 1
#define CONFIG_COMPOUND_SEGMENT 0 #define CONFIG_COMPOUND_SEGMENT 1
#define CONFIG_EXT_REFS 1 #define CONFIG_EXT_REFS 1
#define CONFIG_GLOBAL_MOTION 1 #define CONFIG_GLOBAL_MOTION 1
#define CONFIG_NEW_QUANT 0 #define CONFIG_NEW_QUANT 0
#define CONFIG_SUPERTX 0 #define CONFIG_SUPERTX 0
#define CONFIG_ANS 0 #define CONFIG_ANS 0
#define CONFIG_EC_MULTISYMBOL 1
#define CONFIG_NEW_TOKENSET 1
#define CONFIG_LOOP_RESTORATION 0 #define CONFIG_LOOP_RESTORATION 0
#define CONFIG_EXT_PARTITION 0 #define CONFIG_EXT_PARTITION 0
#define CONFIG_EXT_PARTITION_TYPES 0 #define CONFIG_EXT_PARTITION_TYPES 0
@ -118,20 +113,18 @@
#define CONFIG_NCOBMC 0 #define CONFIG_NCOBMC 0
#define CONFIG_WARPED_MOTION 1 #define CONFIG_WARPED_MOTION 1
#define CONFIG_Q_ADAPT_PROBS 0 #define CONFIG_Q_ADAPT_PROBS 0
#define CONFIG_SUBFRAME_PROB_UPDATE 0
#define CONFIG_BITSTREAM_DEBUG 0 #define CONFIG_BITSTREAM_DEBUG 0
#define CONFIG_ALT_INTRA 1 #define CONFIG_ALT_INTRA 1
#define CONFIG_PALETTE 1 #define CONFIG_PALETTE 1
#define CONFIG_PALETTE_DELTA_ENCODING 0 #define CONFIG_PALETTE_DELTA_ENCODING 0
#define CONFIG_DAALA_EC 1
#define CONFIG_RAWBITS 0 #define CONFIG_RAWBITS 0
#define CONFIG_EC_SMALLMUL 0 #define CONFIG_EC_SMALLMUL 1
#define CONFIG_PVQ 0 #define CONFIG_PVQ 0
#define CONFIG_CFL 0 #define CONFIG_CFL 0
#define CONFIG_XIPHRC 0 #define CONFIG_XIPHRC 0
#define CONFIG_CB4X4 1 #define CONFIG_CB4X4 1
#define CONFIG_CHROMA_2X2 0 #define CONFIG_CHROMA_2X2 0
#define CONFIG_CHROMA_SUB8X8 0 #define CONFIG_CHROMA_SUB8X8 1
#define CONFIG_FRAME_SIZE 0 #define CONFIG_FRAME_SIZE 0
#define CONFIG_DELTA_Q 1 #define CONFIG_DELTA_Q 1
#define CONFIG_EXT_DELTA_Q 1 #define CONFIG_EXT_DELTA_Q 1
@ -139,10 +132,10 @@
#define CONFIG_FILTER_7BIT 1 #define CONFIG_FILTER_7BIT 1
#define CONFIG_PARALLEL_DEBLOCKING 0 #define CONFIG_PARALLEL_DEBLOCKING 0
#define CONFIG_PARALLEL_DEBLOCKING_15TAP 0 #define CONFIG_PARALLEL_DEBLOCKING_15TAP 0
#define CONFIG_LOOPFILTERING_ACROSS_TILES 0 #define CONFIG_LOOPFILTERING_ACROSS_TILES 1
#define CONFIG_TILE_GROUPS 1 #define CONFIG_TILE_GROUPS 1
#define CONFIG_EC_ADAPT 1 #define CONFIG_EC_ADAPT 1
#define CONFIG_TEMPMV_SIGNALING 0 #define CONFIG_TEMPMV_SIGNALING 1
#define CONFIG_RD_DEBUG 0 #define CONFIG_RD_DEBUG 0
#define CONFIG_REFERENCE_BUFFER 1 #define CONFIG_REFERENCE_BUFFER 1
#define CONFIG_COEF_INTERLEAVE 0 #define CONFIG_COEF_INTERLEAVE 0
@ -160,7 +153,9 @@
#define CONFIG_NEW_MULTISYMBOL 0 #define CONFIG_NEW_MULTISYMBOL 0
#define CONFIG_COMPOUND_SINGLEREF 0 #define CONFIG_COMPOUND_SINGLEREF 0
#define CONFIG_AOM_QM 0 #define CONFIG_AOM_QM 0
#define CONFIG_LOWDELAY_COMPOUND 0 #define CONFIG_ONE_SIDED_COMPOUND 1
#define CONFIG_SMOOTH_HV 0
#define CONFIG_VAR_REFS 0
#define CONFIG_ANALYZER 0 #define CONFIG_ANALYZER 0
#define DECODE_WIDTH_LIMIT 8192 #define DECODE_WIDTH_LIMIT 8192
#define DECODE_HEIGHT_LIMIT 4608 #define DECODE_HEIGHT_LIMIT 4608

2207
media/libaom/config/win/x64/aom_dsp_rtcd.h
View file

File diff suppressed because it is too large Load diff

267
media/libaom/config/win/x64/av1_rtcd.h
View file

@ -60,13 +60,7 @@ void aom_clpf_hblock_hbd_sse4_1(uint16_t *dst, const uint16_t *src, int dstride,
RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd); RTCD_EXTERN void (*aom_clpf_hblock_hbd)(uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd);
int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t av1_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); #define av1_block_error av1_block_error_c
int64_t av1_block_error_avx2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
RTCD_EXTERN int64_t (*av1_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz);
int64_t av1_block_error_fp_c(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
int64_t av1_block_error_fp_sse2(const int16_t *coeff, const int16_t *dqcoeff, int block_size);
#define av1_block_error_fp av1_block_error_fp_sse2
void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_c(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params); void av1_convolve_horiz_ssse3(const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params);
@ -80,9 +74,7 @@ int av1_diamond_search_sad_c(struct macroblock *x, const struct search_site_conf
#define av1_diamond_search_sad av1_diamond_search_sad_c #define av1_diamond_search_sad av1_diamond_search_sad_c
void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_fdct8x8_quant_c(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fdct8x8_quant_sse2(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_fdct8x8_quant av1_fdct8x8_quant_c
void av1_fdct8x8_quant_ssse3(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
RTCD_EXTERN void (*av1_fdct8x8_quant)(const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type); void av1_fht16x16_sse2(const int16_t *input, tran_low_t *output, int stride, int tx_type);
@ -149,9 +141,174 @@ RTCD_EXTERN int (*av1_full_search_sad)(const struct macroblock *x, const struct
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type); void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type);
#define av1_fwd_idtx av1_fwd_idtx_c #define av1_fwd_idtx av1_fwd_idtx_c
void av1_fwd_txfm2d_16x16_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_16x16_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_16x16)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_32x32_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_32x32)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_4x4_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_4x4)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_64x64_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_64x64)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_c(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwd_txfm2d_8x8_sse4_1(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_fwd_txfm2d_8x8)(const int16_t *input, int32_t *output, int stride, int tx_type, int bd);
void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); void av1_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_fwht4x4 av1_fwht4x4_c #define av1_fwht4x4 av1_fwht4x4_c
int64_t av1_highbd_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
int64_t av1_highbd_block_error_sse2(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd);
#define av1_highbd_block_error av1_highbd_block_error_sse2
void av1_highbd_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8 av1_highbd_convolve8_sse2
void av1_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg av1_highbd_convolve8_avg_sse2
void av1_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_horiz av1_highbd_convolve8_avg_horiz_sse2
void av1_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_avg_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_avg_vert av1_highbd_convolve8_avg_vert_sse2
void av1_highbd_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_horiz av1_highbd_convolve8_horiz_sse2
void av1_highbd_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
void av1_highbd_convolve8_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve8_vert av1_highbd_convolve8_vert_sse2
void av1_highbd_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_avg av1_highbd_convolve_avg_c
void av1_highbd_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps);
#define av1_highbd_convolve_copy av1_highbd_convolve_copy_c
void av1_highbd_convolve_horiz_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_horiz_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_horiz)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_init_c(void);
void av1_highbd_convolve_init_sse4_1(void);
RTCD_EXTERN void (*av1_highbd_convolve_init)(void);
void av1_highbd_convolve_vert_c(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_convolve_vert_sse4_1(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
RTCD_EXTERN void (*av1_highbd_convolve_vert)(const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd);
void av1_highbd_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x16 av1_highbd_fht16x16_c
void av1_highbd_fht16x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x32 av1_highbd_fht16x32_c
void av1_highbd_fht16x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x4 av1_highbd_fht16x4_c
void av1_highbd_fht16x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht16x8 av1_highbd_fht16x8_c
void av1_highbd_fht32x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x16 av1_highbd_fht32x16_c
void av1_highbd_fht32x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x32 av1_highbd_fht32x32_c
void av1_highbd_fht32x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht32x8 av1_highbd_fht32x8_c
void av1_highbd_fht4x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x16 av1_highbd_fht4x16_c
void av1_highbd_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x4_sse4_1(const int16_t *input, tran_low_t *output, int stride, int tx_type);
RTCD_EXTERN void (*av1_highbd_fht4x4)(const int16_t *input, tran_low_t *output, int stride, int tx_type);
void av1_highbd_fht4x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht4x8 av1_highbd_fht4x8_c
void av1_highbd_fht8x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x16 av1_highbd_fht8x16_c
void av1_highbd_fht8x32_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x32 av1_highbd_fht8x32_c
void av1_highbd_fht8x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x4 av1_highbd_fht8x4_c
void av1_highbd_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type);
#define av1_highbd_fht8x8 av1_highbd_fht8x8_c
void av1_highbd_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride);
#define av1_highbd_fwht4x4 av1_highbd_fwht4x4_c
void av1_highbd_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd);
#define av1_highbd_iht16x16_256_add av1_highbd_iht16x16_256_add_c
void av1_highbd_iht16x32_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x32_512_add av1_highbd_iht16x32_512_add_c
void av1_highbd_iht16x4_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x4_64_add av1_highbd_iht16x4_64_add_c
void av1_highbd_iht16x8_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht16x8_128_add av1_highbd_iht16x8_128_add_c
void av1_highbd_iht32x16_512_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x16_512_add av1_highbd_iht32x16_512_add_c
void av1_highbd_iht32x8_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht32x8_256_add av1_highbd_iht32x8_256_add_c
void av1_highbd_iht4x16_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x16_64_add av1_highbd_iht4x16_64_add_c
void av1_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x4_16_add av1_highbd_iht4x4_16_add_c
void av1_highbd_iht4x8_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht4x8_32_add av1_highbd_iht4x8_32_add_c
void av1_highbd_iht8x16_128_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x16_128_add av1_highbd_iht8x16_128_add_c
void av1_highbd_iht8x32_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x32_256_add av1_highbd_iht8x32_256_add_c
void av1_highbd_iht8x4_32_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x4_32_add av1_highbd_iht8x4_32_add_c
void av1_highbd_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd);
#define av1_highbd_iht8x8_64_add av1_highbd_iht8x8_64_add_c
void av1_highbd_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
#define av1_highbd_quantize_b av1_highbd_quantize_b_c
void av1_highbd_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_quantize_fp_sse4_1(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
RTCD_EXTERN void (*av1_highbd_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan, int log_scale);
void av1_highbd_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_highbd_temporal_filter_apply av1_highbd_temporal_filter_apply_c
void av1_highbd_warp_affine_c(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_highbd_warp_affine_ssse3(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_highbd_warp_affine)(const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type); void av1_iht16x16_256_add_avx2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
@ -204,6 +361,25 @@ void av1_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride
void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type); void av1_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
#define av1_iht8x8_64_add av1_iht8x8_64_add_sse2 #define av1_iht8x8_64_add av1_iht8x8_64_add_sse2
void av1_inv_txfm2d_add_16x16_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_16x16_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_16x16)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_32x32_avx2(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_32x32)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_4x4_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_4x4)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_64x64_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
#define av1_inv_txfm2d_add_64x64 av1_inv_txfm2d_add_64x64_c
void av1_inv_txfm2d_add_8x8_c(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_inv_txfm2d_add_8x8_sse4_1(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
RTCD_EXTERN void (*av1_inv_txfm2d_add_8x8)(const int32_t *input, uint16_t *output, int stride, int tx_type, int bd);
void av1_lowbd_convolve_init_c(void); void av1_lowbd_convolve_init_c(void);
void av1_lowbd_convolve_init_ssse3(void); void av1_lowbd_convolve_init_ssse3(void);
RTCD_EXTERN void (*av1_lowbd_convolve_init)(void); RTCD_EXTERN void (*av1_lowbd_convolve_init)(void);
@ -213,20 +389,35 @@ void av1_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_b
void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_sse2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_sse2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_ssse3(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_quantize_fp av1_quantize_fp_sse2
RTCD_EXTERN void (*av1_quantize_fp)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); void av1_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_quantize_fp_32x32_ssse3(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan); #define av1_quantize_fp_32x32 av1_quantize_fp_32x32_c
RTCD_EXTERN void (*av1_quantize_fp_32x32)(const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan);
void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_c(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count); void av1_temporal_filter_apply_sse2(uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count);
#define av1_temporal_filter_apply av1_temporal_filter_apply_sse2 #define av1_temporal_filter_apply av1_temporal_filter_apply_sse2
void av1_warp_affine_c(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_c(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_warp_affine_sse2(int32_t *mat, uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int ref_frm, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta); void av1_warp_affine_sse2(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
#define av1_warp_affine av1_warp_affine_sse2 void av1_warp_affine_ssse3(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
RTCD_EXTERN void (*av1_warp_affine)(const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int comp_avg, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
void av1_wedge_compute_delta_squares_c(int16_t *d, const int16_t *a, const int16_t *b, int N);
void av1_wedge_compute_delta_squares_sse2(int16_t *d, const int16_t *a, const int16_t *b, int N);
#define av1_wedge_compute_delta_squares av1_wedge_compute_delta_squares_sse2
int av1_wedge_sign_from_residuals_c(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
int av1_wedge_sign_from_residuals_sse2(const int16_t *ds, const uint8_t *m, int N, int64_t limit);
#define av1_wedge_sign_from_residuals av1_wedge_sign_from_residuals_sse2
uint64_t av1_wedge_sse_from_residuals_c(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
uint64_t av1_wedge_sse_from_residuals_sse2(const int16_t *r1, const int16_t *d, const uint8_t *m, int N);
#define av1_wedge_sse_from_residuals av1_wedge_sse_from_residuals_sse2
double compute_cross_correlation_c(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
double compute_cross_correlation_sse4_1(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
RTCD_EXTERN double (*compute_cross_correlation)(unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2);
void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_c(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride); void copy_4x4_16bit_to_16bit_sse2(uint16_t *dst, int dstride, const uint16_t *src, int sstride);
@ -304,14 +495,10 @@ static void setup_rtcd_internal(void)
aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2; aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse2;
if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3; if (flags & HAS_SSSE3) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_ssse3;
if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1; if (flags & HAS_SSE4_1) aom_clpf_hblock_hbd = aom_clpf_hblock_hbd_sse4_1;
av1_block_error = av1_block_error_sse2;
if (flags & HAS_AVX2) av1_block_error = av1_block_error_avx2;
av1_convolve_horiz = av1_convolve_horiz_c; av1_convolve_horiz = av1_convolve_horiz_c;
if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3; if (flags & HAS_SSSE3) av1_convolve_horiz = av1_convolve_horiz_ssse3;
av1_convolve_vert = av1_convolve_vert_c; av1_convolve_vert = av1_convolve_vert_c;
if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3; if (flags & HAS_SSSE3) av1_convolve_vert = av1_convolve_vert_ssse3;
av1_fdct8x8_quant = av1_fdct8x8_quant_sse2;
if (flags & HAS_SSSE3) av1_fdct8x8_quant = av1_fdct8x8_quant_ssse3;
av1_fht16x16 = av1_fht16x16_sse2; av1_fht16x16 = av1_fht16x16_sse2;
if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2; if (flags & HAS_AVX2) av1_fht16x16 = av1_fht16x16_avx2;
av1_fht32x32 = av1_fht32x32_sse2; av1_fht32x32 = av1_fht32x32_sse2;
@ -319,14 +506,44 @@ static void setup_rtcd_internal(void)
av1_full_search_sad = av1_full_search_sad_c; av1_full_search_sad = av1_full_search_sad_c;
if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3; if (flags & HAS_SSE3) av1_full_search_sad = av1_full_search_sadx3;
if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8; if (flags & HAS_SSE4_1) av1_full_search_sad = av1_full_search_sadx8;
av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_16x16 = av1_fwd_txfm2d_16x16_sse4_1;
av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_32x32 = av1_fwd_txfm2d_32x32_sse4_1;
av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_4x4 = av1_fwd_txfm2d_4x4_sse4_1;
av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_64x64 = av1_fwd_txfm2d_64x64_sse4_1;
av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_c;
if (flags & HAS_SSE4_1) av1_fwd_txfm2d_8x8 = av1_fwd_txfm2d_8x8_sse4_1;
av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_horiz = av1_highbd_convolve_horiz_sse4_1;
av1_highbd_convolve_init = av1_highbd_convolve_init_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_init = av1_highbd_convolve_init_sse4_1;
av1_highbd_convolve_vert = av1_highbd_convolve_vert_c;
if (flags & HAS_SSE4_1) av1_highbd_convolve_vert = av1_highbd_convolve_vert_sse4_1;
av1_highbd_fht4x4 = av1_highbd_fht4x4_c;
if (flags & HAS_SSE4_1) av1_highbd_fht4x4 = av1_highbd_fht4x4_sse4_1;
av1_highbd_quantize_fp = av1_highbd_quantize_fp_c;
if (flags & HAS_SSE4_1) av1_highbd_quantize_fp = av1_highbd_quantize_fp_sse4_1;
av1_highbd_warp_affine = av1_highbd_warp_affine_c;
if (flags & HAS_SSSE3) av1_highbd_warp_affine = av1_highbd_warp_affine_ssse3;
av1_iht16x16_256_add = av1_iht16x16_256_add_sse2; av1_iht16x16_256_add = av1_iht16x16_256_add_sse2;
if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2; if (flags & HAS_AVX2) av1_iht16x16_256_add = av1_iht16x16_256_add_avx2;
av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_16x16 = av1_inv_txfm2d_add_16x16_sse4_1;
av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_c;
if (flags & HAS_AVX2) av1_inv_txfm2d_add_32x32 = av1_inv_txfm2d_add_32x32_avx2;
av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_4x4 = av1_inv_txfm2d_add_4x4_sse4_1;
av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_c;
if (flags & HAS_SSE4_1) av1_inv_txfm2d_add_8x8 = av1_inv_txfm2d_add_8x8_sse4_1;
av1_lowbd_convolve_init = av1_lowbd_convolve_init_c; av1_lowbd_convolve_init = av1_lowbd_convolve_init_c;
if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3; if (flags & HAS_SSSE3) av1_lowbd_convolve_init = av1_lowbd_convolve_init_ssse3;
av1_quantize_fp = av1_quantize_fp_sse2; av1_warp_affine = av1_warp_affine_sse2;
if (flags & HAS_SSSE3) av1_quantize_fp = av1_quantize_fp_ssse3; if (flags & HAS_SSSE3) av1_warp_affine = av1_warp_affine_ssse3;
av1_quantize_fp_32x32 = av1_quantize_fp_32x32_c; compute_cross_correlation = compute_cross_correlation_c;
if (flags & HAS_SSSE3) av1_quantize_fp_32x32 = av1_quantize_fp_32x32_ssse3; if (flags & HAS_SSE4_1) compute_cross_correlation = compute_cross_correlation_sse4_1;
copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2; copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse2;
if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3; if (flags & HAS_SSSE3) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_ssse3;
if (flags & HAS_SSE4_1) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse4_1; if (flags & HAS_SSE4_1) copy_4x4_16bit_to_16bit = copy_4x4_16bit_to_16bit_sse4_1;

73
media/libaom/sources.mozbuild
View file

@ -54,6 +54,8 @@ files = {
'../../third_party/aom/aom_dsp/variance.c', '../../third_party/aom/aom_dsp/variance.c',
'../../third_party/aom/aom_dsp/x86/aom_asm_stubs.c', '../../third_party/aom/aom_dsp/x86/aom_asm_stubs.c',
'../../third_party/aom/aom_dsp/x86/aom_convolve_copy_sse2.asm', '../../third_party/aom/aom_dsp/x86/aom_convolve_copy_sse2.asm',
'../../third_party/aom/aom_dsp/x86/aom_high_subpixel_8t_sse2.asm',
'../../third_party/aom/aom_dsp/x86/aom_high_subpixel_bilinear_sse2.asm',
'../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_avx2.c', '../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_avx2.c',
'../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_ssse3.c', '../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_ssse3.c',
'../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm', '../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm',
@ -71,13 +73,27 @@ files = {
'../../third_party/aom/aom_dsp/x86/fwd_txfm_ssse3_x86_64.asm', '../../third_party/aom/aom_dsp/x86/fwd_txfm_ssse3_x86_64.asm',
'../../third_party/aom/aom_dsp/x86/halfpix_variance_impl_sse2.asm', '../../third_party/aom/aom_dsp/x86/halfpix_variance_impl_sse2.asm',
'../../third_party/aom/aom_dsp/x86/halfpix_variance_sse2.c', '../../third_party/aom/aom_dsp/x86/halfpix_variance_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_convolve_avx2.c',
'../../third_party/aom/aom_dsp/x86/highbd_intrapred_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_loopfilter_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_quantize_intrin_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_sad4d_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_sad_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_subpel_variance_impl_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_subtract_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_variance_impl_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_variance_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_variance_sse4.c',
'../../third_party/aom/aom_dsp/x86/intrapred_sse2.asm', '../../third_party/aom/aom_dsp/x86/intrapred_sse2.asm',
'../../third_party/aom/aom_dsp/x86/intrapred_ssse3.asm', '../../third_party/aom/aom_dsp/x86/intrapred_ssse3.asm',
'../../third_party/aom/aom_dsp/x86/inv_txfm_avx2.c',
'../../third_party/aom/aom_dsp/x86/inv_txfm_sse2.c', '../../third_party/aom/aom_dsp/x86/inv_txfm_sse2.c',
'../../third_party/aom/aom_dsp/x86/inv_txfm_ssse3.c', '../../third_party/aom/aom_dsp/x86/inv_txfm_ssse3.c',
'../../third_party/aom/aom_dsp/x86/inv_wht_sse2.asm', '../../third_party/aom/aom_dsp/x86/inv_wht_sse2.asm',
'../../third_party/aom/aom_dsp/x86/loopfilter_avx2.c', '../../third_party/aom/aom_dsp/x86/loopfilter_avx2.c',
'../../third_party/aom/aom_dsp/x86/loopfilter_sse2.c', '../../third_party/aom/aom_dsp/x86/loopfilter_sse2.c',
'../../third_party/aom/aom_dsp/x86/masked_sad_intrin_ssse3.c',
'../../third_party/aom/aom_dsp/x86/masked_variance_intrin_ssse3.c',
'../../third_party/aom/aom_dsp/x86/obmc_sad_sse4.c', '../../third_party/aom/aom_dsp/x86/obmc_sad_sse4.c',
'../../third_party/aom/aom_dsp/x86/obmc_variance_sse4.c', '../../third_party/aom/aom_dsp/x86/obmc_variance_sse4.c',
'../../third_party/aom/aom_dsp/x86/quantize_avx_x86_64.asm', '../../third_party/aom/aom_dsp/x86/quantize_avx_x86_64.asm',
@ -86,6 +102,7 @@ files = {
'../../third_party/aom/aom_dsp/x86/sad4d_avx2.c', '../../third_party/aom/aom_dsp/x86/sad4d_avx2.c',
'../../third_party/aom/aom_dsp/x86/sad4d_sse2.asm', '../../third_party/aom/aom_dsp/x86/sad4d_sse2.asm',
'../../third_party/aom/aom_dsp/x86/sad_avx2.c', '../../third_party/aom/aom_dsp/x86/sad_avx2.c',
'../../third_party/aom/aom_dsp/x86/sad_highbd_avx2.c',
'../../third_party/aom/aom_dsp/x86/sad_sse2.asm', '../../third_party/aom/aom_dsp/x86/sad_sse2.asm',
'../../third_party/aom/aom_dsp/x86/sad_sse3.asm', '../../third_party/aom/aom_dsp/x86/sad_sse3.asm',
'../../third_party/aom/aom_dsp/x86/sad_sse4.asm', '../../third_party/aom/aom_dsp/x86/sad_sse4.asm',
@ -152,9 +169,14 @@ files = {
'../../third_party/aom/av1/common/x86/av1_convolve_ssse3.c', '../../third_party/aom/av1/common/x86/av1_convolve_ssse3.c',
'../../third_party/aom/av1/common/x86/av1_fwd_txfm1d_sse4.c', '../../third_party/aom/av1/common/x86/av1_fwd_txfm1d_sse4.c',
'../../third_party/aom/av1/common/x86/av1_fwd_txfm2d_sse4.c', '../../third_party/aom/av1/common/x86/av1_fwd_txfm2d_sse4.c',
'../../third_party/aom/av1/common/x86/av1_highbd_convolve_sse4.c',
'../../third_party/aom/av1/common/x86/highbd_inv_txfm_avx2.c',
'../../third_party/aom/av1/common/x86/highbd_inv_txfm_sse4.c',
'../../third_party/aom/av1/common/x86/highbd_warp_plane_ssse3.c',
'../../third_party/aom/av1/common/x86/hybrid_inv_txfm_avx2.c', '../../third_party/aom/av1/common/x86/hybrid_inv_txfm_avx2.c',
'../../third_party/aom/av1/common/x86/idct_intrin_sse2.c', '../../third_party/aom/av1/common/x86/idct_intrin_sse2.c',
'../../third_party/aom/av1/common/x86/warp_plane_sse2.c', '../../third_party/aom/av1/common/x86/warp_plane_sse2.c',
'../../third_party/aom/av1/common/x86/warp_plane_ssse3.c',
'../../third_party/aom/av1/decoder/decodeframe.c', '../../third_party/aom/av1/decoder/decodeframe.c',
'../../third_party/aom/av1/decoder/decodemv.c', '../../third_party/aom/av1/decoder/decodemv.c',
'../../third_party/aom/av1/decoder/decoder.c', '../../third_party/aom/av1/decoder/decoder.c',
@ -196,16 +218,21 @@ files = {
'../../third_party/aom/av1/encoder/temporal_filter.c', '../../third_party/aom/av1/encoder/temporal_filter.c',
'../../third_party/aom/av1/encoder/tokenize.c', '../../third_party/aom/av1/encoder/tokenize.c',
'../../third_party/aom/av1/encoder/treewriter.c', '../../third_party/aom/av1/encoder/treewriter.c',
'../../third_party/aom/av1/encoder/variance_tree.c', '../../third_party/aom/av1/encoder/wedge_utils.c',
'../../third_party/aom/av1/encoder/x86/av1_highbd_quantize_sse4.c',
'../../third_party/aom/av1/encoder/x86/av1_quantize_sse2.c', '../../third_party/aom/av1/encoder/x86/av1_quantize_sse2.c',
'../../third_party/aom/av1/encoder/x86/av1_quantize_ssse3_x86_64.asm', '../../third_party/aom/av1/encoder/x86/av1_quantize_ssse3_x86_64.asm',
'../../third_party/aom/av1/encoder/x86/corner_match_sse4.c',
'../../third_party/aom/av1/encoder/x86/dct_intrin_sse2.c', '../../third_party/aom/av1/encoder/x86/dct_intrin_sse2.c',
'../../third_party/aom/av1/encoder/x86/dct_sse2.asm', '../../third_party/aom/av1/encoder/x86/dct_sse2.asm',
'../../third_party/aom/av1/encoder/x86/dct_ssse3.c', '../../third_party/aom/av1/encoder/x86/dct_ssse3.c',
'../../third_party/aom/av1/encoder/x86/error_intrin_avx2.c', '../../third_party/aom/av1/encoder/x86/error_intrin_avx2.c',
'../../third_party/aom/av1/encoder/x86/error_sse2.asm', '../../third_party/aom/av1/encoder/x86/error_sse2.asm',
'../../third_party/aom/av1/encoder/x86/highbd_block_error_intrin_sse2.c',
'../../third_party/aom/av1/encoder/x86/highbd_fwd_txfm_sse4.c',
'../../third_party/aom/av1/encoder/x86/hybrid_fwd_txfm_avx2.c', '../../third_party/aom/av1/encoder/x86/hybrid_fwd_txfm_avx2.c',
'../../third_party/aom/av1/encoder/x86/temporal_filter_apply_sse2.asm', '../../third_party/aom/av1/encoder/x86/temporal_filter_apply_sse2.asm',
'../../third_party/aom/av1/encoder/x86/wedge_utils_sse2.c',
], ],
'IA32_EXPORTS': [ 'IA32_EXPORTS': [
'../../third_party/aom/aom/aom.h', '../../third_party/aom/aom/aom.h',
@ -260,6 +287,8 @@ files = {
'../../third_party/aom/aom_dsp/variance.c', '../../third_party/aom/aom_dsp/variance.c',
'../../third_party/aom/aom_dsp/x86/aom_asm_stubs.c', '../../third_party/aom/aom_dsp/x86/aom_asm_stubs.c',
'../../third_party/aom/aom_dsp/x86/aom_convolve_copy_sse2.asm', '../../third_party/aom/aom_dsp/x86/aom_convolve_copy_sse2.asm',
'../../third_party/aom/aom_dsp/x86/aom_high_subpixel_8t_sse2.asm',
'../../third_party/aom/aom_dsp/x86/aom_high_subpixel_bilinear_sse2.asm',
'../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_avx2.c', '../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_avx2.c',
'../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_ssse3.c', '../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_intrin_ssse3.c',
'../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm', '../../third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm',
@ -275,19 +304,34 @@ files = {
'../../third_party/aom/aom_dsp/x86/fwd_txfm_sse2.c', '../../third_party/aom/aom_dsp/x86/fwd_txfm_sse2.c',
'../../third_party/aom/aom_dsp/x86/halfpix_variance_impl_sse2.asm', '../../third_party/aom/aom_dsp/x86/halfpix_variance_impl_sse2.asm',
'../../third_party/aom/aom_dsp/x86/halfpix_variance_sse2.c', '../../third_party/aom/aom_dsp/x86/halfpix_variance_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_convolve_avx2.c',
'../../third_party/aom/aom_dsp/x86/highbd_intrapred_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_loopfilter_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_quantize_intrin_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_sad4d_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_sad_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_subpel_variance_impl_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_subtract_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_variance_impl_sse2.asm',
'../../third_party/aom/aom_dsp/x86/highbd_variance_sse2.c',
'../../third_party/aom/aom_dsp/x86/highbd_variance_sse4.c',
'../../third_party/aom/aom_dsp/x86/intrapred_sse2.asm', '../../third_party/aom/aom_dsp/x86/intrapred_sse2.asm',
'../../third_party/aom/aom_dsp/x86/intrapred_ssse3.asm', '../../third_party/aom/aom_dsp/x86/intrapred_ssse3.asm',
'../../third_party/aom/aom_dsp/x86/inv_txfm_avx2.c',
'../../third_party/aom/aom_dsp/x86/inv_txfm_sse2.c', '../../third_party/aom/aom_dsp/x86/inv_txfm_sse2.c',
'../../third_party/aom/aom_dsp/x86/inv_txfm_ssse3.c', '../../third_party/aom/aom_dsp/x86/inv_txfm_ssse3.c',
'../../third_party/aom/aom_dsp/x86/inv_wht_sse2.asm', '../../third_party/aom/aom_dsp/x86/inv_wht_sse2.asm',
'../../third_party/aom/aom_dsp/x86/loopfilter_avx2.c', '../../third_party/aom/aom_dsp/x86/loopfilter_avx2.c',
'../../third_party/aom/aom_dsp/x86/loopfilter_sse2.c', '../../third_party/aom/aom_dsp/x86/loopfilter_sse2.c',
'../../third_party/aom/aom_dsp/x86/masked_sad_intrin_ssse3.c',
'../../third_party/aom/aom_dsp/x86/masked_variance_intrin_ssse3.c',
'../../third_party/aom/aom_dsp/x86/obmc_sad_sse4.c', '../../third_party/aom/aom_dsp/x86/obmc_sad_sse4.c',
'../../third_party/aom/aom_dsp/x86/obmc_variance_sse4.c', '../../third_party/aom/aom_dsp/x86/obmc_variance_sse4.c',
'../../third_party/aom/aom_dsp/x86/quantize_sse2.c', '../../third_party/aom/aom_dsp/x86/quantize_sse2.c',
'../../third_party/aom/aom_dsp/x86/sad4d_avx2.c', '../../third_party/aom/aom_dsp/x86/sad4d_avx2.c',
'../../third_party/aom/aom_dsp/x86/sad4d_sse2.asm', '../../third_party/aom/aom_dsp/x86/sad4d_sse2.asm',
'../../third_party/aom/aom_dsp/x86/sad_avx2.c', '../../third_party/aom/aom_dsp/x86/sad_avx2.c',
'../../third_party/aom/aom_dsp/x86/sad_highbd_avx2.c',
'../../third_party/aom/aom_dsp/x86/sad_sse2.asm', '../../third_party/aom/aom_dsp/x86/sad_sse2.asm',
'../../third_party/aom/aom_dsp/x86/sad_sse3.asm', '../../third_party/aom/aom_dsp/x86/sad_sse3.asm',
'../../third_party/aom/aom_dsp/x86/sad_sse4.asm', '../../third_party/aom/aom_dsp/x86/sad_sse4.asm',
@ -353,9 +397,14 @@ files = {
'../../third_party/aom/av1/common/x86/av1_convolve_ssse3.c', '../../third_party/aom/av1/common/x86/av1_convolve_ssse3.c',
'../../third_party/aom/av1/common/x86/av1_fwd_txfm1d_sse4.c', '../../third_party/aom/av1/common/x86/av1_fwd_txfm1d_sse4.c',
'../../third_party/aom/av1/common/x86/av1_fwd_txfm2d_sse4.c', '../../third_party/aom/av1/common/x86/av1_fwd_txfm2d_sse4.c',
'../../third_party/aom/av1/common/x86/av1_highbd_convolve_sse4.c',
'../../third_party/aom/av1/common/x86/highbd_inv_txfm_avx2.c',
'../../third_party/aom/av1/common/x86/highbd_inv_txfm_sse4.c',
'../../third_party/aom/av1/common/x86/highbd_warp_plane_ssse3.c',
'../../third_party/aom/av1/common/x86/hybrid_inv_txfm_avx2.c', '../../third_party/aom/av1/common/x86/hybrid_inv_txfm_avx2.c',
'../../third_party/aom/av1/common/x86/idct_intrin_sse2.c', '../../third_party/aom/av1/common/x86/idct_intrin_sse2.c',
'../../third_party/aom/av1/common/x86/warp_plane_sse2.c', '../../third_party/aom/av1/common/x86/warp_plane_sse2.c',
'../../third_party/aom/av1/common/x86/warp_plane_ssse3.c',
'../../third_party/aom/av1/decoder/decodeframe.c', '../../third_party/aom/av1/decoder/decodeframe.c',
'../../third_party/aom/av1/decoder/decodemv.c', '../../third_party/aom/av1/decoder/decodemv.c',
'../../third_party/aom/av1/decoder/decoder.c', '../../third_party/aom/av1/decoder/decoder.c',
@ -397,15 +446,20 @@ files = {
'../../third_party/aom/av1/encoder/temporal_filter.c', '../../third_party/aom/av1/encoder/temporal_filter.c',
'../../third_party/aom/av1/encoder/tokenize.c', '../../third_party/aom/av1/encoder/tokenize.c',
'../../third_party/aom/av1/encoder/treewriter.c', '../../third_party/aom/av1/encoder/treewriter.c',
'../../third_party/aom/av1/encoder/variance_tree.c', '../../third_party/aom/av1/encoder/wedge_utils.c',
'../../third_party/aom/av1/encoder/x86/av1_highbd_quantize_sse4.c',
'../../third_party/aom/av1/encoder/x86/av1_quantize_sse2.c', '../../third_party/aom/av1/encoder/x86/av1_quantize_sse2.c',
'../../third_party/aom/av1/encoder/x86/corner_match_sse4.c',
'../../third_party/aom/av1/encoder/x86/dct_intrin_sse2.c', '../../third_party/aom/av1/encoder/x86/dct_intrin_sse2.c',
'../../third_party/aom/av1/encoder/x86/dct_sse2.asm', '../../third_party/aom/av1/encoder/x86/dct_sse2.asm',
'../../third_party/aom/av1/encoder/x86/dct_ssse3.c', '../../third_party/aom/av1/encoder/x86/dct_ssse3.c',
'../../third_party/aom/av1/encoder/x86/error_intrin_avx2.c', '../../third_party/aom/av1/encoder/x86/error_intrin_avx2.c',
'../../third_party/aom/av1/encoder/x86/error_sse2.asm', '../../third_party/aom/av1/encoder/x86/error_sse2.asm',
'../../third_party/aom/av1/encoder/x86/highbd_block_error_intrin_sse2.c',
'../../third_party/aom/av1/encoder/x86/highbd_fwd_txfm_sse4.c',
'../../third_party/aom/av1/encoder/x86/hybrid_fwd_txfm_avx2.c', '../../third_party/aom/av1/encoder/x86/hybrid_fwd_txfm_avx2.c',
'../../third_party/aom/av1/encoder/x86/temporal_filter_apply_sse2.asm', '../../third_party/aom/av1/encoder/x86/temporal_filter_apply_sse2.asm',
'../../third_party/aom/av1/encoder/x86/wedge_utils_sse2.c',
], ],
'ARM_EXPORTS': [ 'ARM_EXPORTS': [
'../../third_party/aom/aom/aom.h', '../../third_party/aom/aom/aom.h',
@ -440,7 +494,6 @@ files = {
'../../third_party/aom/aom_dsp/arm/aom_convolve_copy_neon_asm.asm', '../../third_party/aom/aom_dsp/arm/aom_convolve_copy_neon_asm.asm',
'../../third_party/aom/aom_dsp/arm/aom_convolve_neon.c', '../../third_party/aom/aom_dsp/arm/aom_convolve_neon.c',
'../../third_party/aom/aom_dsp/arm/avg_neon.c', '../../third_party/aom/aom_dsp/arm/avg_neon.c',
'../../third_party/aom/aom_dsp/arm/bilinear_filter_media.asm',
'../../third_party/aom/aom_dsp/arm/fwd_txfm_neon.c', '../../third_party/aom/aom_dsp/arm/fwd_txfm_neon.c',
'../../third_party/aom/aom_dsp/arm/hadamard_neon.c', '../../third_party/aom/aom_dsp/arm/hadamard_neon.c',
'../../third_party/aom/aom_dsp/arm/idct16x16_1_add_neon.asm', '../../third_party/aom/aom_dsp/arm/idct16x16_1_add_neon.asm',
@ -460,16 +513,10 @@ files = {
'../../third_party/aom/aom_dsp/arm/loopfilter_mb_neon.asm', '../../third_party/aom/aom_dsp/arm/loopfilter_mb_neon.asm',
'../../third_party/aom/aom_dsp/arm/loopfilter_neon.c', '../../third_party/aom/aom_dsp/arm/loopfilter_neon.c',
'../../third_party/aom/aom_dsp/arm/sad4d_neon.c', '../../third_party/aom/aom_dsp/arm/sad4d_neon.c',
'../../third_party/aom/aom_dsp/arm/sad_media.asm',
'../../third_party/aom/aom_dsp/arm/sad_neon.c', '../../third_party/aom/aom_dsp/arm/sad_neon.c',
'../../third_party/aom/aom_dsp/arm/save_reg_neon.asm', '../../third_party/aom/aom_dsp/arm/save_reg_neon.asm',
'../../third_party/aom/aom_dsp/arm/subpel_variance_media.c',
'../../third_party/aom/aom_dsp/arm/subpel_variance_neon.c', '../../third_party/aom/aom_dsp/arm/subpel_variance_neon.c',
'../../third_party/aom/aom_dsp/arm/subtract_neon.c', '../../third_party/aom/aom_dsp/arm/subtract_neon.c',
'../../third_party/aom/aom_dsp/arm/variance_halfpixvar16x16_h_media.asm',
'../../third_party/aom/aom_dsp/arm/variance_halfpixvar16x16_hv_media.asm',
'../../third_party/aom/aom_dsp/arm/variance_halfpixvar16x16_v_media.asm',
'../../third_party/aom/aom_dsp/arm/variance_media.asm',
'../../third_party/aom/aom_dsp/arm/variance_neon.c', '../../third_party/aom/aom_dsp/arm/variance_neon.c',
'../../third_party/aom/aom_dsp/avg.c', '../../third_party/aom/aom_dsp/avg.c',
'../../third_party/aom/aom_dsp/binary_codes_reader.c', '../../third_party/aom/aom_dsp/binary_codes_reader.c',
@ -509,8 +556,6 @@ files = {
'../../third_party/aom/av1/av1_cx_iface.c', '../../third_party/aom/av1/av1_cx_iface.c',
'../../third_party/aom/av1/av1_dx_iface.c', '../../third_party/aom/av1/av1_dx_iface.c',
'../../third_party/aom/av1/common/alloccommon.c', '../../third_party/aom/av1/common/alloccommon.c',
'../../third_party/aom/av1/common/arm/neon/iht4x4_add_neon.c',
'../../third_party/aom/av1/common/arm/neon/iht8x8_add_neon.c',
'../../third_party/aom/av1/common/av1_fwd_txfm1d.c', '../../third_party/aom/av1/common/av1_fwd_txfm1d.c',
'../../third_party/aom/av1/common/av1_fwd_txfm2d.c', '../../third_party/aom/av1/common/av1_fwd_txfm2d.c',
'../../third_party/aom/av1/common/av1_inv_txfm1d.c', '../../third_party/aom/av1/common/av1_inv_txfm1d.c',
@ -553,8 +598,6 @@ files = {
'../../third_party/aom/av1/encoder/aq_complexity.c', '../../third_party/aom/av1/encoder/aq_complexity.c',
'../../third_party/aom/av1/encoder/aq_cyclicrefresh.c', '../../third_party/aom/av1/encoder/aq_cyclicrefresh.c',
'../../third_party/aom/av1/encoder/aq_variance.c', '../../third_party/aom/av1/encoder/aq_variance.c',
'../../third_party/aom/av1/encoder/arm/neon/dct_neon.c',
'../../third_party/aom/av1/encoder/arm/neon/error_neon.c',
'../../third_party/aom/av1/encoder/arm/neon/quantize_neon.c', '../../third_party/aom/av1/encoder/arm/neon/quantize_neon.c',
'../../third_party/aom/av1/encoder/av1_quantize.c', '../../third_party/aom/av1/encoder/av1_quantize.c',
'../../third_party/aom/av1/encoder/bitstream.c', '../../third_party/aom/av1/encoder/bitstream.c',
@ -588,7 +631,7 @@ files = {
'../../third_party/aom/av1/encoder/temporal_filter.c', '../../third_party/aom/av1/encoder/temporal_filter.c',
'../../third_party/aom/av1/encoder/tokenize.c', '../../third_party/aom/av1/encoder/tokenize.c',
'../../third_party/aom/av1/encoder/treewriter.c', '../../third_party/aom/av1/encoder/treewriter.c',
'../../third_party/aom/av1/encoder/variance_tree.c', '../../third_party/aom/av1/encoder/wedge_utils.c',
], ],
'GENERIC_EXPORTS': [ 'GENERIC_EXPORTS': [
'../../third_party/aom/aom/aom.h', '../../third_party/aom/aom/aom.h',
@ -725,6 +768,6 @@ files = {
'../../third_party/aom/av1/encoder/temporal_filter.c', '../../third_party/aom/av1/encoder/temporal_filter.c',
'../../third_party/aom/av1/encoder/tokenize.c', '../../third_party/aom/av1/encoder/tokenize.c',
'../../third_party/aom/av1/encoder/treewriter.c', '../../third_party/aom/av1/encoder/treewriter.c',
'../../third_party/aom/av1/encoder/variance_tree.c', '../../third_party/aom/av1/encoder/wedge_utils.c',
], ],
} }

1
third_party/aom/CHANGELOG vendored
View file

@ -1,6 +1,7 @@
Next Release Next Release
- Incompatible changes: - Incompatible changes:
The AV1 encoder's default keyframe interval changed to 128 from 9999. The AV1 encoder's default keyframe interval changed to 128 from 9999.
Support for armv6 was removed.
2016-04-07 v0.1.0 "AOMedia Codec 1" 2016-04-07 v0.1.0 "AOMedia Codec 1"
This release is the first Alliance for Open Media codec. This release is the first Alliance for Open Media codec.

259
third_party/aom/CMakeLists.txt vendored
View file

@ -9,6 +9,18 @@
## PATENTS file, you can obtain it at www.aomedia.org/license/patent. ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
## ##
cmake_minimum_required(VERSION 3.5) cmake_minimum_required(VERSION 3.5)
if (NOT EMSCRIPTEN)
if ("${CMAKE_BUILD_TYPE}" STREQUAL "")
set(CMAKE_BUILD_TYPE "RelWithDebInfo")
endif ()
endif ()
option(ENABLE_CCACHE "Enable ccache support." OFF)
option(ENABLE_DISTCC "Enable distcc support." OFF)
option(ENABLE_DOCS "Enable documentation generation (doxygen required)." ON)
option(ENABLE_NASM "Use nasm instead of yasm for x86 assembly." OFF)
project(AOM C CXX) project(AOM C CXX)
set(AOM_ROOT "${CMAKE_CURRENT_SOURCE_DIR}") set(AOM_ROOT "${CMAKE_CURRENT_SOURCE_DIR}")
@ -114,15 +126,15 @@ set(AOM_COMMON_APP_UTIL_SOURCES
"${AOM_ROOT}/md5_utils.h" "${AOM_ROOT}/md5_utils.h"
"${AOM_ROOT}/tools_common.c" "${AOM_ROOT}/tools_common.c"
"${AOM_ROOT}/tools_common.h" "${AOM_ROOT}/tools_common.h"
"${AOM_ROOT}/video_common.h") "${AOM_ROOT}/video_common.h"
"${AOM_ROOT}/y4menc.c"
"${AOM_ROOT}/y4menc.h")
set(AOM_DECODER_APP_UTIL_SOURCES set(AOM_DECODER_APP_UTIL_SOURCES
"${AOM_ROOT}/ivfdec.c" "${AOM_ROOT}/ivfdec.c"
"${AOM_ROOT}/ivfdec.h" "${AOM_ROOT}/ivfdec.h"
"${AOM_ROOT}/video_reader.c" "${AOM_ROOT}/video_reader.c"
"${AOM_ROOT}/video_reader.h" "${AOM_ROOT}/video_reader.h")
"${AOM_ROOT}/y4menc.c"
"${AOM_ROOT}/y4menc.h")
set(AOM_ENCODER_APP_UTIL_SOURCES set(AOM_ENCODER_APP_UTIL_SOURCES
"${AOM_ROOT}/ivfenc.c" "${AOM_ROOT}/ivfenc.c"
@ -190,88 +202,117 @@ foreach (aom_lib ${AOM_LIB_TARGETS})
add_dependencies(${aom_lib} aom_rtcd) add_dependencies(${aom_lib} aom_rtcd)
endif () endif ()
endforeach () endforeach ()
# #
# Application and application support targets. # Application and application support targets.
# #
add_library(aom_common_app_util OBJECT ${AOM_COMMON_APP_UTIL_SOURCES}) add_library(aom_common_app_util OBJECT ${AOM_COMMON_APP_UTIL_SOURCES})
add_library(aom_decoder_app_util OBJECT ${AOM_DECODER_APP_UTIL_SOURCES})
add_library(aom_encoder_app_util OBJECT ${AOM_ENCODER_APP_UTIL_SOURCES})
add_executable(aom_cx_set_ref
"${AOM_ROOT}/examples/aom_cx_set_ref.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>)
add_executable(aomdec if (CONFIG_AV1_DECODER)
add_library(aom_decoder_app_util OBJECT ${AOM_DECODER_APP_UTIL_SOURCES})
add_executable(aomdec
"${AOM_ROOT}/aomdec.c" "${AOM_ROOT}/aomdec.c"
$<TARGET_OBJECTS:aom_common_app_util> $<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>) $<TARGET_OBJECTS:aom_decoder_app_util>)
add_executable(aomenc if (CONFIG_ANALYZER)
"${AOM_ROOT}/aomenc.c" add_executable(analyzer
$<TARGET_OBJECTS:aom_common_app_util> "${AOM_ROOT}/examples/analyzer.cc"
$<TARGET_OBJECTS:aom_encoder_app_util> $<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_stats>) $<TARGET_OBJECTS:aom_decoder_app_util>)
target_link_libraries(analyzer ${AOM_LIB_LINK_TYPE} ${wxWidgets_LIBRARIES})
set(AOM_APP_TARGETS ${AOM_APP_TARGETS} analyzer)
endif ()
if (CONFIG_ANALYZER) add_executable(decode_to_md5
add_executable(analyzer "${AOM_ROOT}/examples/decode_to_md5.c"
"${AOM_ROOT}/examples/analyzer.cc"
$<TARGET_OBJECTS:aom_common_app_util> $<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>) $<TARGET_OBJECTS:aom_decoder_app_util>)
target_link_libraries(analyzer PUBLIC ${wxWidgets_LIBRARIES}) add_executable(decode_with_drops
set(AOM_APP_TARGETS ${AOM_APP_TARGETS} analyzer) "${AOM_ROOT}/examples/decode_with_drops.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>)
add_executable(simple_decoder
"${AOM_ROOT}/examples/simple_decoder.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>)
if (CONFIG_INSPECTION)
add_executable(inspect
"${AOM_ROOT}/examples/inspect.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>)
set(AOM_DECODER_EXAMPLE_TARGETS ${AOM_DECODER_EXAMPLE_TARGETS} inspect)
if (EMSCRIPTEN)
add_preproc_definition(_POSIX_SOURCE)
add_compiler_flag_if_supported("-s TOTAL_MEMORY=134217728")
add_compiler_flag_if_supported("-s MODULARIZE=1")
add_compiler_flag_if_supported("-s EXPORT_NAME=\"\'DecoderModule\'\"")
add_compiler_flag_if_supported("--memory-init-file 0")
if ("${CMAKE_BUILD_TYPE}" STREQUAL "")
# Default to -O3 when no build type specified.
add_compiler_flag_if_supported("-O3")
endif ()
em_link_post_js(inspect "${AOM_ROOT}/tools/inspect-post.js")
endif ()
endif ()
# Maintain lists of example and app targets.
set(AOM_DECODER_EXAMPLE_TARGETS ${AOM_DECODER_EXAMPLE_TARGETS}
decode_to_md5 decode_with_drops simple_decoder)
set(AOM_APP_TARGETS ${AOM_APP_TARGETS} aomdec ${AOM_DECODER_EXAMPLE_TARGETS})
endif () endif ()
add_executable(decode_to_md5
"${AOM_ROOT}/examples/decode_to_md5.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>)
add_executable(decode_with_drops if (CONFIG_AV1_ENCODER)
"${AOM_ROOT}/examples/decode_with_drops.c" add_library(aom_encoder_app_util OBJECT ${AOM_ENCODER_APP_UTIL_SOURCES})
$<TARGET_OBJECTS:aom_common_app_util> add_executable(aomenc
$<TARGET_OBJECTS:aom_decoder_app_util>) "${AOM_ROOT}/aomenc.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>
$<TARGET_OBJECTS:aom_encoder_stats>)
add_executable(lossless_encoder
"${AOM_ROOT}/examples/lossless_encoder.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>)
add_executable(set_maps
"${AOM_ROOT}/examples/set_maps.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>)
add_executable(simple_encoder
"${AOM_ROOT}/examples/simple_encoder.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>)
add_executable(twopass_encoder
"${AOM_ROOT}/examples/twopass_encoder.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>)
add_executable(lossless_encoder # Add encoder apps and examples to target lists.
"${AOM_ROOT}/examples/lossless_encoder.c" set(AOM_ENCODER_EXAMPLE_TARGETS
$<TARGET_OBJECTS:aom_common_app_util> lossless_encoder set_maps simple_encoder twopass_encoder)
$<TARGET_OBJECTS:aom_encoder_app_util>) set(AOM_APP_TARGETS ${AOM_APP_TARGETS} aomenc ${AOM_ENCODER_EXAMPLE_TARGETS})
add_executable(set_maps
"${AOM_ROOT}/examples/set_maps.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>)
add_executable(simple_decoder
"${AOM_ROOT}/examples/simple_decoder.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>)
if (CONFIG_INSPECTION)
add_executable(inspect
"${AOM_ROOT}/examples/inspect.c"
$<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_decoder_app_util>)
set(AOM_APP_TARGETS ${AOM_APP_TARGETS} inspect)
endif () endif ()
add_executable(simple_encoder # Maintain a separate variable listing only the examples to facilitate
"${AOM_ROOT}/examples/simple_encoder.c" # installation of example programs into an examples sub directory of
$<TARGET_OBJECTS:aom_common_app_util> # $AOM_DIST_DIR/bin when building the dist target.
$<TARGET_OBJECTS:aom_encoder_app_util>) set(AOM_EXAMPLE_TARGETS
${AOM_DECODER_EXAMPLE_TARGETS} ${AOM_ENCODER_EXAMPLE_TARGETS})
add_executable(twopass_encoder if (CONFIG_AV1_DECODER AND CONFIG_AV1_ENCODER)
"${AOM_ROOT}/examples/twopass_encoder.c" add_executable(aom_cx_set_ref
$<TARGET_OBJECTS:aom_common_app_util> "${AOM_ROOT}/examples/aom_cx_set_ref.c"
$<TARGET_OBJECTS:aom_encoder_app_util>) $<TARGET_OBJECTS:aom_common_app_util>
$<TARGET_OBJECTS:aom_encoder_app_util>)
# List of app targets. set(AOM_EXAMPLE_TARGETS ${AOM_EXAMPLE_TARGETS} aom_cx_set_ref)
set(AOM_APP_TARGETS ${AOM_APP_TARGETS} aom_cx_set_ref aomdec aomenc set(AOM_APP_TARGETS ${AOM_APP_TARGETS} aom_cx_set_ref)
decode_to_md5 decode_with_drops lossless_encoder set_maps simple_decoder endif ()
simple_encoder twopass_encoder)
foreach (aom_app ${AOM_APP_TARGETS}) foreach (aom_app ${AOM_APP_TARGETS})
target_link_libraries(${aom_app} PUBLIC aom) target_link_libraries(${aom_app} ${AOM_LIB_LINK_TYPE} aom)
endforeach () endforeach ()
if (CONFIG_LIBYUV) if (CONFIG_LIBYUV)
@ -297,8 +338,13 @@ if (CONFIG_WEBM_IO)
endif () endif ()
# Add to existing targets. # Add to existing targets.
target_sources(aom_decoder_app_util PUBLIC ${AOM_WEBM_DECODER_SOURCES}) if (CONFIG_AV1_DECODER)
target_sources(aom_encoder_app_util PUBLIC ${AOM_WEBM_ENCODER_SOURCES}) target_sources(aom_decoder_app_util PUBLIC ${AOM_WEBM_DECODER_SOURCES})
endif ()
if (CONFIG_AV1_ENCODER)
target_sources(aom_encoder_app_util PUBLIC ${AOM_WEBM_ENCODER_SOURCES})
endif ()
foreach (aom_app ${AOM_APP_TARGETS}) foreach (aom_app ${AOM_APP_TARGETS})
target_sources(${aom_app} PUBLIC $<TARGET_OBJECTS:webm>) target_sources(${aom_app} PUBLIC $<TARGET_OBJECTS:webm>)
@ -314,7 +360,7 @@ endif ()
if (HAVE_PTHREAD_H AND CONFIG_MULTITHREAD) if (HAVE_PTHREAD_H AND CONFIG_MULTITHREAD)
find_package(Threads) find_package(Threads)
foreach (app_target ${AOM_APP_TARGETS}) foreach (app_target ${AOM_APP_TARGETS})
target_link_libraries(${app_target} PUBLIC Threads::Threads) target_link_libraries(${app_target} ${AOM_LIB_LINK_TYPE} Threads::Threads)
endforeach () endforeach ()
endif () endif ()
@ -337,23 +383,90 @@ if (XCODE)
endif () endif ()
endif () endif ()
if ("${CMAKE_GENERATOR}" MATCHES "Makefiles$" )
# Users of the configure build expect the example targets to be built in the
# examples sub directory of the configured build directory after running make.
file(MAKE_DIRECTORY "${AOM_CONFIG_DIR}/examples")
set_target_properties(${AOM_EXAMPLE_TARGETS} PROPERTIES
RUNTIME_OUTPUT_DIRECTORY "${AOM_CONFIG_DIR}/examples")
endif ()
# Aomedia documentation rule.
if (ENABLE_DOCS)
include(FindDoxygen)
if (DOXYGEN_FOUND)
include("${AOM_ROOT}/docs.cmake")
setup_documentation_targets()
else ()
message("--- Cannot find doxygen, ENABLE_DOCS turned off.")
set(ENABLE_DOCS OFF)
endif ()
endif ()
# Aomedia install rule. # Aomedia install rule.
# TODO(tomfinegan): Add the missing pkg-config related bits.
set(AOM_INSTALL_INCS set(AOM_INSTALL_INCS
"${AOM_ROOT}/aom/aom.h" "${AOM_ROOT}/aom/aom.h"
"${AOM_ROOT}/aom/aomcx.h"
"${AOM_ROOT}/aom/aomdx.h"
"${AOM_ROOT}/aom/aom_codec.h" "${AOM_ROOT}/aom/aom_codec.h"
"${AOM_ROOT}/aom/aom_frame_buffer.h" "${AOM_ROOT}/aom/aom_frame_buffer.h"
"${AOM_ROOT}/aom/aom_image.h" "${AOM_ROOT}/aom/aom_image.h"
"${AOM_ROOT}/aom/aom_integer.h" "${AOM_ROOT}/aom/aom_integer.h"
"${AOM_ROOT}/aom/aom_decoder.h"
"${AOM_ROOT}/aom/aom_encoder.h"
"${AOM_ROOT}/aom/aom.h") "${AOM_ROOT}/aom/aom.h")
set(AOM_INSTALL_BINS aomdec aomenc)
if (CONFIG_AV1_DECODER)
set(AOM_INSTALL_BINS ${AOM_INSTALL_BINS} aomdec)
set(AOM_INSTALL_INCS
${AOM_INSTALL_INCS}
"${AOM_ROOT}/aom/aom_decoder.h"
"${AOM_ROOT}/aom/aomdx.h")
endif ()
if (CONFIG_AV1_ENCODER)
set(AOM_INSTALL_INCS
${AOM_INSTALL_INCS}
"${AOM_ROOT}/aom/aomcx.h"
"${AOM_ROOT}/aom/aom_encoder.h")
set(AOM_INSTALL_BINS ${AOM_INSTALL_BINS} aomenc)
endif ()
set(AOM_INSTALL_LIBS aom) set(AOM_INSTALL_LIBS aom)
install(FILES ${AOM_INSTALL_INCS} install(FILES ${AOM_INSTALL_INCS}
DESTINATION "${CMAKE_INSTALL_PREFIX}/include/aom") DESTINATION "${CMAKE_INSTALL_PREFIX}/include/aom")
install(FILES "${AOM_CONFIG_DIR}/aom.pc"
DESTINATION "${CMAKE_INSTALL_PREFIX}/lib/pkgconfig")
install(TARGETS ${AOM_INSTALL_LIBS} DESTINATION "${CMAKE_INSTALL_PREFIX}/lib") install(TARGETS ${AOM_INSTALL_LIBS} DESTINATION "${CMAKE_INSTALL_PREFIX}/lib")
install(TARGETS ${AOM_INSTALL_BINS} DESTINATION "${CMAKE_INSTALL_PREFIX}/bin") install(TARGETS ${AOM_INSTALL_BINS} DESTINATION "${CMAKE_INSTALL_PREFIX}/bin")
# Aomedia dist rule.
if (CONFIG_AV1_DECODER)
set(AOM_DIST_APPS ${AOM_DIST_APPS} $<TARGET_FILE:aomdec>)
endif ()
if (CONFIG_AV1_ENCODER)
set(AOM_DIST_APPS ${AOM_DIST_APPS} $<TARGET_FILE:aomenc>)
endif ()
foreach (example ${AOM_EXAMPLE_TARGETS})
list(APPEND AOM_DIST_EXAMPLES $<TARGET_FILE:${example}>)
endforeach ()
if (NOT AOM_DIST_DIR)
set(AOM_DIST_DIR "${AOM_CONFIG_DIR}/dist")
endif ()
add_custom_target(dist
COMMAND ${CMAKE_COMMAND}
-DAOM_ROOT=${AOM_ROOT}
-DAOM_CONFIG_DIR=${AOM_CONFIG_DIR}
-DAOM_DIST_DIR=${AOM_DIST_DIR}
-DAOM_DIST_APPS="${AOM_DIST_APPS}"
-DAOM_DIST_EXAMPLES="${AOM_DIST_EXAMPLES}"
-DAOM_DIST_INCLUDES="${AOM_INSTALL_INCS}"
-DAOM_DIST_LIBS=$<TARGET_FILE:aom>
-DENABLE_DOCS=${ENABLE_DOCS}
-P "${AOM_ROOT}/build/cmake/dist.cmake"
DEPENDS ${AOM_INSTALL_BINS} ${AOM_INSTALL_LIBS}
${AOM_INSTALL_INCS} ${AOM_EXAMPLE_TARGETS})
if (ENABLE_DOCS)
add_dependencies(dist docs)
endif ()

3
third_party/aom/README vendored
View file

@ -47,9 +47,6 @@ COMPILING THE APPLICATIONS/LIBRARIES:
--help output of the configure script. As of this writing, the list of --help output of the configure script. As of this writing, the list of
available targets is: available targets is:
armv6-linux-rvct
armv6-linux-gcc
armv6-none-rvct
arm64-darwin-gcc arm64-darwin-gcc
armv7-android-gcc armv7-android-gcc
armv7-darwin-gcc armv7-darwin-gcc

27
third_party/aom/aom/aom_decoder.h vendored
View file

@ -84,11 +84,9 @@ extern "C" {
/*!\brief Stream properties /*!\brief Stream properties
* *
* This structure is used to query or set properties of the decoded * This structure is used to query or set properties of the decoded
* stream. Algorithms may extend this structure with data specific * stream.
* to their bitstream by setting the sz member appropriately.
*/ */
typedef struct aom_codec_stream_info { typedef struct aom_codec_stream_info {
unsigned int sz; /**< Size of this structure */
unsigned int w; /**< Width (or 0 for unknown/default) */ unsigned int w; /**< Width (or 0 for unknown/default) */
unsigned int h; /**< Height (or 0 for unknown/default) */ unsigned int h; /**< Height (or 0 for unknown/default) */
unsigned int is_kf; /**< Current frame is a keyframe */ unsigned int is_kf; /**< Current frame is a keyframe */
@ -154,13 +152,15 @@ aom_codec_err_t aom_codec_dec_init_ver(aom_codec_ctx_t *ctx,
* \param[in] iface Pointer to the algorithm interface * \param[in] iface Pointer to the algorithm interface
* \param[in] data Pointer to a block of data to parse * \param[in] data Pointer to a block of data to parse
* \param[in] data_sz Size of the data buffer * \param[in] data_sz Size of the data buffer
* \param[in,out] si Pointer to stream info to update. The size member * \param[in,out] si Pointer to stream info to update.
* \ref MUST be properly initialized, but \ref MAY be
* clobbered by the algorithm. This parameter \ref MAY
* be NULL.
* *
* \retval #AOM_CODEC_OK * \retval #AOM_CODEC_OK
* Bitstream is parsable and stream information updated * Bitstream is parsable and stream information updated.
* \retval #AOM_CODEC_INVALID_PARAM
* One of the arguments is invalid, for example a NULL pointer.
* \retval #AOM_CODEC_UNSUP_BITSTREAM
* The decoder didn't recognize the coded data, or the
* buffer was too short.
*/ */
aom_codec_err_t aom_codec_peek_stream_info(aom_codec_iface_t *iface, aom_codec_err_t aom_codec_peek_stream_info(aom_codec_iface_t *iface,
const uint8_t *data, const uint8_t *data,
@ -172,13 +172,14 @@ aom_codec_err_t aom_codec_peek_stream_info(aom_codec_iface_t *iface,
* Returns information about the stream that has been parsed during decoding. * Returns information about the stream that has been parsed during decoding.
* *
* \param[in] ctx Pointer to this instance's context * \param[in] ctx Pointer to this instance's context
* \param[in,out] si Pointer to stream info to update. The size member * \param[in,out] si Pointer to stream info to update.
* \ref MUST be properly initialized, but \ref MAY be
* clobbered by the algorithm. This parameter \ref MAY
* be NULL.
* *
* \retval #AOM_CODEC_OK * \retval #AOM_CODEC_OK
* Bitstream is parsable and stream information updated * Bitstream is parsable and stream information updated.
* \retval #AOM_CODEC_INVALID_PARAM
* One of the arguments is invalid, for example a NULL pointer.
* \retval #AOM_CODEC_UNSUP_BITSTREAM
* The decoder couldn't parse the submitted data.
*/ */
aom_codec_err_t aom_codec_get_stream_info(aom_codec_ctx_t *ctx, aom_codec_err_t aom_codec_get_stream_info(aom_codec_ctx_t *ctx,
aom_codec_stream_info_t *si); aom_codec_stream_info_t *si);

4
third_party/aom/aom/aomdx.h vendored
View file

@ -37,13 +37,17 @@ extern aom_codec_iface_t aom_codec_av1_dx_algo;
extern aom_codec_iface_t *aom_codec_av1_dx(void); extern aom_codec_iface_t *aom_codec_av1_dx(void);
/*!@} - end algorithm interface member group*/ /*!@} - end algorithm interface member group*/
#ifndef AOM_ACCOUNTING_H_
/** Data structure that stores bit accounting for debug /** Data structure that stores bit accounting for debug
*/ */
typedef struct Accounting Accounting; typedef struct Accounting Accounting;
#endif
#ifndef AOM_INSPECTION_H_
/** Callback that inspects decoder frame data. /** Callback that inspects decoder frame data.
*/ */
typedef void (*aom_inspect_cb)(void *decoder, void *ctx); typedef void (*aom_inspect_cb)(void *decoder, void *ctx);
#endif
/*!\brief Structure to hold inspection callback and context. /*!\brief Structure to hold inspection callback and context.
* *

20
third_party/aom/aom/internal/aom_codec_internal.h vendored
View file

@ -43,7 +43,6 @@
*/ */
#ifndef AOM_INTERNAL_AOM_CODEC_INTERNAL_H_ #ifndef AOM_INTERNAL_AOM_CODEC_INTERNAL_H_
#define AOM_INTERNAL_AOM_CODEC_INTERNAL_H_ #define AOM_INTERNAL_AOM_CODEC_INTERNAL_H_
#include "./aom_config.h"
#include "../aom_decoder.h" #include "../aom_decoder.h"
#include "../aom_encoder.h" #include "../aom_encoder.h"
#include <stdarg.h> #include <stdarg.h>
@ -441,25 +440,6 @@ void aom_internal_error(struct aom_internal_error_info *info,
...) CLANG_ANALYZER_NORETURN; ...) CLANG_ANALYZER_NORETURN;
void aom_merge_corrupted_flag(int *corrupted, int value); void aom_merge_corrupted_flag(int *corrupted, int value);
#if CONFIG_DEBUG
#define AOM_CHECK_MEM_ERROR(error_info, lval, expr) \
do { \
lval = (expr); \
if (!lval) \
aom_internal_error(error_info, AOM_CODEC_MEM_ERROR, \
"Failed to allocate " #lval " at %s:%d", __FILE__, \
__LINE__); \
} while (0)
#else
#define AOM_CHECK_MEM_ERROR(error_info, lval, expr) \
do { \
lval = (expr); \
if (!lval) \
aom_internal_error(error_info, AOM_CODEC_MEM_ERROR, \
"Failed to allocate " #lval); \
} while (0)
#endif
#ifdef __cplusplus #ifdef __cplusplus
} // extern "C" } // extern "C"
#endif #endif

11
third_party/aom/aom/src/aom_decoder.c vendored
View file

@ -69,10 +69,9 @@ aom_codec_err_t aom_codec_peek_stream_info(aom_codec_iface_t *iface,
aom_codec_stream_info_t *si) { aom_codec_stream_info_t *si) {
aom_codec_err_t res; aom_codec_err_t res;
if (!iface || !data || !data_sz || !si || if (!iface || !data || !data_sz || !si) {
si->sz < sizeof(aom_codec_stream_info_t))
res = AOM_CODEC_INVALID_PARAM; res = AOM_CODEC_INVALID_PARAM;
else { } else {
/* Set default/unknown values */ /* Set default/unknown values */
si->w = 0; si->w = 0;
si->h = 0; si->h = 0;
@ -87,11 +86,11 @@ aom_codec_err_t aom_codec_get_stream_info(aom_codec_ctx_t *ctx,
aom_codec_stream_info_t *si) { aom_codec_stream_info_t *si) {
aom_codec_err_t res; aom_codec_err_t res;
if (!ctx || !si || si->sz < sizeof(aom_codec_stream_info_t)) if (!ctx || !si) {
res = AOM_CODEC_INVALID_PARAM; res = AOM_CODEC_INVALID_PARAM;
else if (!ctx->iface || !ctx->priv) } else if (!ctx->iface || !ctx->priv) {
res = AOM_CODEC_ERROR; res = AOM_CODEC_ERROR;
else { } else {
/* Set default/unknown values */ /* Set default/unknown values */
si->w = 0; si->w = 0;
si->h = 0; si->h = 0;

267
third_party/aom/aom_dsp/aom_convolve.c vendored
View file

@ -337,14 +337,14 @@ static void convolve_add_src_horiz(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride, uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *x_filters, int x0_q4, const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h) { int x_step_q4, int w, int h) {
int x, y; int x, y, k;
src -= SUBPEL_TAPS / 2 - 1; src -= SUBPEL_TAPS / 2 - 1;
for (y = 0; y < h; ++y) { for (y = 0; y < h; ++y) {
int x_q4 = x0_q4; int x_q4 = x0_q4;
for (x = 0; x < w; ++x) { for (x = 0; x < w; ++x) {
const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS]; const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK]; const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int k, sum = 0; int sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k]; for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS) + dst[x] = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS) +
src_x[SUBPEL_TAPS / 2 - 1]); src_x[SUBPEL_TAPS / 2 - 1]);
@ -359,7 +359,7 @@ static void convolve_add_src_vert(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride, uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *y_filters, int y0_q4, const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h) { int y_step_q4, int w, int h) {
int x, y; int x, y, k;
src -= src_stride * (SUBPEL_TAPS / 2 - 1); src -= src_stride * (SUBPEL_TAPS / 2 - 1);
for (x = 0; x < w; ++x) { for (x = 0; x < w; ++x) {
@ -367,7 +367,7 @@ static void convolve_add_src_vert(const uint8_t *src, ptrdiff_t src_stride,
for (y = 0; y < h; ++y) { for (y = 0; y < h; ++y) {
const unsigned char *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride]; const unsigned char *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK]; const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK];
int k, sum = 0; int sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k) for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_y[k * src_stride] * y_filter[k]; sum += src_y[k * src_stride] * y_filter[k];
dst[y * dst_stride] = dst[y * dst_stride] =
@ -446,6 +446,127 @@ void aom_convolve8_add_src_c(const uint8_t *src, ptrdiff_t src_stride,
convolve_add_src(src, src_stride, dst, dst_stride, filters_x, x0_q4, convolve_add_src(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, filters_y, y0_q4, y_step_q4, w, h); x_step_q4, filters_y, y0_q4, y_step_q4, w, h);
} }
static void convolve_add_src_horiz_hip(const uint8_t *src, ptrdiff_t src_stride,
uint16_t *dst, ptrdiff_t dst_stride,
const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h) {
const int bd = 8;
int x, y, k;
src -= SUBPEL_TAPS / 2 - 1;
for (y = 0; y < h; ++y) {
int x_q4 = x0_q4;
for (x = 0; x < w; ++x) {
const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int sum = ((int)src_x[SUBPEL_TAPS / 2 - 1] << FILTER_BITS) +
(1 << (bd + FILTER_BITS - 1));
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] =
(uint16_t)clamp(ROUND_POWER_OF_TWO(sum, FILTER_BITS - EXTRAPREC_BITS),
0, EXTRAPREC_CLAMP_LIMIT(bd) - 1);
x_q4 += x_step_q4;
}
src += src_stride;
dst += dst_stride;
}
}
static void convolve_add_src_vert_hip(const uint16_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h) {
const int bd = 8;
int x, y, k;
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
for (x = 0; x < w; ++x) {
int y_q4 = y0_q4;
for (y = 0; y < h; ++y) {
const uint16_t *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK];
int sum =
((int)src_y[(SUBPEL_TAPS / 2 - 1) * src_stride] << FILTER_BITS) -
(1 << (bd + FILTER_BITS + EXTRAPREC_BITS - 1));
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_y[k * src_stride] * y_filter[k];
dst[y * dst_stride] =
clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS + EXTRAPREC_BITS));
y_q4 += y_step_q4;
}
++src;
++dst;
}
}
static void convolve_add_src_hip(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *const x_filters, int x0_q4,
int x_step_q4,
const InterpKernel *const y_filters, int y0_q4,
int y_step_q4, int w, int h) {
uint16_t temp[MAX_EXT_SIZE * MAX_SB_SIZE];
int intermediate_height =
(((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
assert(w <= MAX_SB_SIZE);
assert(h <= MAX_SB_SIZE);
assert(y_step_q4 <= 32);
assert(x_step_q4 <= 32);
convolve_add_src_horiz_hip(src - src_stride * (SUBPEL_TAPS / 2 - 1),
src_stride, temp, MAX_SB_SIZE, x_filters, x0_q4,
x_step_q4, w, intermediate_height);
convolve_add_src_vert_hip(temp + MAX_SB_SIZE * (SUBPEL_TAPS / 2 - 1),
MAX_SB_SIZE, dst, dst_stride, y_filters, y0_q4,
y_step_q4, w, h);
}
void aom_convolve8_add_src_horiz_hip_c(const uint8_t *src, ptrdiff_t src_stride,
uint16_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4,
int w, int h) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
(void)filter_y;
(void)y_step_q4;
convolve_add_src_horiz_hip(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, w, h);
}
void aom_convolve8_add_src_vert_hip_c(const uint16_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4,
int w, int h) {
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
(void)filter_x;
(void)x_step_q4;
convolve_add_src_vert_hip(src, src_stride, dst, dst_stride, filters_y, y0_q4,
y_step_q4, w, h);
}
void aom_convolve8_add_src_hip_c(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w,
int h) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
convolve_add_src_hip(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, filters_y, y0_q4, y_step_q4, w, h);
}
#endif // CONFIG_LOOP_RESTORATION #endif // CONFIG_LOOP_RESTORATION
#if CONFIG_HIGHBITDEPTH #if CONFIG_HIGHBITDEPTH
@ -721,7 +842,7 @@ static void highbd_convolve_add_src_horiz(const uint8_t *src8,
const InterpKernel *x_filters, const InterpKernel *x_filters,
int x0_q4, int x_step_q4, int w, int x0_q4, int x_step_q4, int w,
int h, int bd) { int h, int bd) {
int x, y; int x, y, k;
uint16_t *src = CONVERT_TO_SHORTPTR(src8); uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8); uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
src -= SUBPEL_TAPS / 2 - 1; src -= SUBPEL_TAPS / 2 - 1;
@ -730,7 +851,7 @@ static void highbd_convolve_add_src_horiz(const uint8_t *src8,
for (x = 0; x < w; ++x) { for (x = 0; x < w; ++x) {
const uint16_t *const src_x = &src[x_q4 >> SUBPEL_BITS]; const uint16_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK]; const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int k, sum = 0; int sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k]; for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] = clip_pixel_highbd( dst[x] = clip_pixel_highbd(
ROUND_POWER_OF_TWO(sum, FILTER_BITS) + src_x[SUBPEL_TAPS / 2 - 1], ROUND_POWER_OF_TWO(sum, FILTER_BITS) + src_x[SUBPEL_TAPS / 2 - 1],
@ -748,7 +869,7 @@ static void highbd_convolve_add_src_vert(const uint8_t *src8,
const InterpKernel *y_filters, const InterpKernel *y_filters,
int y0_q4, int y_step_q4, int w, int h, int y0_q4, int y_step_q4, int w, int h,
int bd) { int bd) {
int x, y; int x, y, k;
uint16_t *src = CONVERT_TO_SHORTPTR(src8); uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8); uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
src -= src_stride * (SUBPEL_TAPS / 2 - 1); src -= src_stride * (SUBPEL_TAPS / 2 - 1);
@ -757,7 +878,7 @@ static void highbd_convolve_add_src_vert(const uint8_t *src8,
for (y = 0; y < h; ++y) { for (y = 0; y < h; ++y) {
const uint16_t *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride]; const uint16_t *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK]; const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK];
int k, sum = 0; int sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k) for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_y[k * src_stride] * y_filter[k]; sum += src_y[k * src_stride] * y_filter[k];
dst[y * dst_stride] = dst[y * dst_stride] =
@ -850,5 +971,135 @@ void aom_highbd_convolve8_add_src_c(const uint8_t *src, ptrdiff_t src_stride,
highbd_convolve_add_src(src, src_stride, dst, dst_stride, filters_x, x0_q4, highbd_convolve_add_src(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, filters_y, y0_q4, y_step_q4, w, h, bd); x_step_q4, filters_y, y0_q4, y_step_q4, w, h, bd);
} }
static void highbd_convolve_add_src_horiz_hip(
const uint8_t *src8, ptrdiff_t src_stride, uint16_t *dst,
ptrdiff_t dst_stride, const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h, int bd) {
const int extraprec_clamp_limit = EXTRAPREC_CLAMP_LIMIT(bd);
int x, y, k;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
src -= SUBPEL_TAPS / 2 - 1;
for (y = 0; y < h; ++y) {
int x_q4 = x0_q4;
for (x = 0; x < w; ++x) {
const uint16_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int sum = ((int)src_x[SUBPEL_TAPS / 2 - 1] << FILTER_BITS) +
(1 << (bd + FILTER_BITS - 1));
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] =
(uint16_t)clamp(ROUND_POWER_OF_TWO(sum, FILTER_BITS - EXTRAPREC_BITS),
0, extraprec_clamp_limit - 1);
x_q4 += x_step_q4;
}
src += src_stride;
dst += dst_stride;
}
}
static void highbd_convolve_add_src_vert_hip(
const uint16_t *src, ptrdiff_t src_stride, uint8_t *dst8,
ptrdiff_t dst_stride, const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h, int bd) {
int x, y, k;
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
for (x = 0; x < w; ++x) {
int y_q4 = y0_q4;
for (y = 0; y < h; ++y) {
const uint16_t *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
const int16_t *const y_filter = y_filters[y_q4 & SUBPEL_MASK];
int sum =
((int)src_y[(SUBPEL_TAPS / 2 - 1) * src_stride] << FILTER_BITS) -
(1 << (bd + FILTER_BITS + EXTRAPREC_BITS - 1));
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_y[k * src_stride] * y_filter[k];
dst[y * dst_stride] = clip_pixel_highbd(
ROUND_POWER_OF_TWO(sum, FILTER_BITS + EXTRAPREC_BITS), bd);
y_q4 += y_step_q4;
}
++src;
++dst;
}
}
static void highbd_convolve_add_src_hip(
const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride, const InterpKernel *const x_filters, int x0_q4,
int x_step_q4, const InterpKernel *const y_filters, int y0_q4,
int y_step_q4, int w, int h, int bd) {
// Note: Fixed size intermediate buffer, temp, places limits on parameters.
// 2d filtering proceeds in 2 steps:
// (1) Interpolate horizontally into an intermediate buffer, temp.
// (2) Interpolate temp vertically to derive the sub-pixel result.
// Deriving the maximum number of rows in the temp buffer (135):
// --Smallest scaling factor is x1/2 ==> y_step_q4 = 32 (Normative).
// --Largest block size is 64x64 pixels.
// --64 rows in the downscaled frame span a distance of (64 - 1) * 32 in the
// original frame (in 1/16th pixel units).
// --Must round-up because block may be located at sub-pixel position.
// --Require an additional SUBPEL_TAPS rows for the 8-tap filter tails.
// --((64 - 1) * 32 + 15) >> 4 + 8 = 135.
uint16_t temp[MAX_EXT_SIZE * MAX_SB_SIZE];
int intermediate_height =
(((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
assert(w <= MAX_SB_SIZE);
assert(h <= MAX_SB_SIZE);
assert(y_step_q4 <= 32);
assert(x_step_q4 <= 32);
highbd_convolve_add_src_horiz_hip(
src - src_stride * (SUBPEL_TAPS / 2 - 1), src_stride, temp, MAX_SB_SIZE,
x_filters, x0_q4, x_step_q4, w, intermediate_height, bd);
highbd_convolve_add_src_vert_hip(temp + MAX_SB_SIZE * (SUBPEL_TAPS / 2 - 1),
MAX_SB_SIZE, dst, dst_stride, y_filters,
y0_q4, y_step_q4, w, h, bd);
}
void aom_highbd_convolve8_add_src_horiz_hip_c(
const uint8_t *src, ptrdiff_t src_stride, uint16_t *dst,
ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w, int h, int bd) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
(void)filter_y;
(void)y_step_q4;
highbd_convolve_add_src_horiz_hip(src, src_stride, dst, dst_stride, filters_x,
x0_q4, x_step_q4, w, h, bd);
}
void aom_highbd_convolve8_add_src_vert_hip_c(
const uint16_t *src, ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w, int h, int bd) {
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
(void)filter_x;
(void)x_step_q4;
highbd_convolve_add_src_vert_hip(src, src_stride, dst, dst_stride, filters_y,
y0_q4, y_step_q4, w, h, bd);
}
void aom_highbd_convolve8_add_src_hip_c(const uint8_t *src,
ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4,
int w, int h, int bd) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
highbd_convolve_add_src_hip(src, src_stride, dst, dst_stride, filters_x,
x0_q4, x_step_q4, filters_y, y0_q4, y_step_q4, w,
h, bd);
}
#endif // CONFIG_LOOP_RESTORATION #endif // CONFIG_LOOP_RESTORATION
#endif // CONFIG_HIGHBITDEPTH #endif // CONFIG_HIGHBITDEPTH

5
third_party/aom/aom_dsp/aom_convolve.h vendored
View file

@ -36,6 +36,11 @@ extern "C" {
#define MAX_EXT_SIZE 135 #define MAX_EXT_SIZE 135
#endif // CONFIG_AV1 && CONFIG_EXT_PARTITION #endif // CONFIG_AV1 && CONFIG_EXT_PARTITION
#if CONFIG_AV1 && CONFIG_LOOP_RESTORATION
#define EXTRAPREC_BITS 2
#define EXTRAPREC_CLAMP_LIMIT(bd) (1 << ((bd) + 1 + EXTRAPREC_BITS))
#endif
typedef void (*convolve_fn_t)(const uint8_t *src, ptrdiff_t src_stride, typedef void (*convolve_fn_t)(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride, uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4, const int16_t *filter_x, int x_step_q4,

84
third_party/aom/aom_dsp/aom_dsp.cmake vendored
View file

@ -8,6 +8,9 @@
## Media Patent License 1.0 was not distributed with this source code in the ## Media Patent License 1.0 was not distributed with this source code in the
## PATENTS file, you can obtain it at www.aomedia.org/license/patent. ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
## ##
if (NOT AOM_AOM_DSP_AOM_DSP_CMAKE_)
set(AOM_AOM_DSP_AOM_DSP_CMAKE_ 1)
set(AOM_DSP_COMMON_SOURCES set(AOM_DSP_COMMON_SOURCES
"${AOM_ROOT}/aom_dsp/aom_convolve.c" "${AOM_ROOT}/aom_dsp/aom_convolve.c"
"${AOM_ROOT}/aom_dsp/aom_convolve.h" "${AOM_ROOT}/aom_dsp/aom_convolve.h"
@ -23,7 +26,6 @@ set(AOM_DSP_COMMON_SOURCES
"${AOM_ROOT}/aom_dsp/loopfilter.c" "${AOM_ROOT}/aom_dsp/loopfilter.c"
"${AOM_ROOT}/aom_dsp/prob.c" "${AOM_ROOT}/aom_dsp/prob.c"
"${AOM_ROOT}/aom_dsp/prob.h" "${AOM_ROOT}/aom_dsp/prob.h"
"${AOM_ROOT}/aom_dsp/sad.c"
"${AOM_ROOT}/aom_dsp/simd/v128_intrinsics.h" "${AOM_ROOT}/aom_dsp/simd/v128_intrinsics.h"
"${AOM_ROOT}/aom_dsp/simd/v128_intrinsics_c.h" "${AOM_ROOT}/aom_dsp/simd/v128_intrinsics_c.h"
"${AOM_ROOT}/aom_dsp/simd/v256_intrinsics.h" "${AOM_ROOT}/aom_dsp/simd/v256_intrinsics.h"
@ -62,8 +64,10 @@ set(AOM_DSP_COMMON_INTRIN_SSE4_1
set(AOM_DSP_COMMON_INTRIN_AVX2 set(AOM_DSP_COMMON_INTRIN_AVX2
"${AOM_ROOT}/aom_dsp/x86/aom_subpixel_8t_intrin_avx2.c" "${AOM_ROOT}/aom_dsp/x86/aom_subpixel_8t_intrin_avx2.c"
"${AOM_ROOT}/aom_dsp/x86/fwd_txfm_avx2.c" "${AOM_ROOT}/aom_dsp/x86/loopfilter_avx2.c"
"${AOM_ROOT}/aom_dsp/x86/loopfilter_avx2.c") "${AOM_ROOT}/aom_dsp/x86/inv_txfm_avx2.c"
"${AOM_ROOT}/aom_dsp/x86/inv_txfm_common_avx2.h"
"${AOM_ROOT}/aom_dsp/x86/txfm_common_avx2.h")
set(AOM_DSP_COMMON_ASM_NEON set(AOM_DSP_COMMON_ASM_NEON
"${AOM_ROOT}/aom_dsp/arm/aom_convolve8_avg_neon_asm.asm" "${AOM_ROOT}/aom_dsp/arm/aom_convolve8_avg_neon_asm.asm"
@ -175,6 +179,8 @@ set(AOM_DSP_COMMON_INTRIN_MSA
if (CONFIG_HIGHBITDEPTH) if (CONFIG_HIGHBITDEPTH)
set(AOM_DSP_COMMON_ASM_SSE2 set(AOM_DSP_COMMON_ASM_SSE2
${AOM_DSP_COMMON_ASM_SSE2} ${AOM_DSP_COMMON_ASM_SSE2}
"${AOM_ROOT}/aom_dsp/x86/aom_high_subpixel_8t_sse2.asm"
"${AOM_ROOT}/aom_dsp/x86/aom_high_subpixel_bilinear_sse2.asm"
"${AOM_ROOT}/aom_dsp/x86/highbd_intrapred_sse2.asm") "${AOM_ROOT}/aom_dsp/x86/highbd_intrapred_sse2.asm")
set(AOM_DSP_COMMON_INTRIN_SSE2 set(AOM_DSP_COMMON_INTRIN_SSE2
@ -198,7 +204,7 @@ if (CONFIG_ANS)
set(AOM_DSP_COMMON_SOURCES set(AOM_DSP_COMMON_SOURCES
${AOM_DSP_COMMON_SOURCES} ${AOM_DSP_COMMON_SOURCES}
"${AOM_ROOT}/aom_dsp/ans.h") "${AOM_ROOT}/aom_dsp/ans.h")
elseif (CONFIG_DAALA_EC) else ()
set(AOM_DSP_COMMON_SOURCES set(AOM_DSP_COMMON_SOURCES
${AOM_DSP_COMMON_SOURCES} ${AOM_DSP_COMMON_SOURCES}
"${AOM_ROOT}/aom_dsp/entcode.c" "${AOM_ROOT}/aom_dsp/entcode.c"
@ -221,7 +227,7 @@ if (CONFIG_AV1)
"${AOM_ROOT}/aom_dsp/x86/inv_txfm_sse2.h") "${AOM_ROOT}/aom_dsp/x86/inv_txfm_sse2.h")
endif () endif ()
if (CONFIG_DECODERS) if (CONFIG_AV1_DECODER)
set(AOM_DSP_DECODER_SOURCES set(AOM_DSP_DECODER_SOURCES
"${AOM_ROOT}/aom_dsp/binary_codes_reader.c" "${AOM_ROOT}/aom_dsp/binary_codes_reader.c"
"${AOM_ROOT}/aom_dsp/binary_codes_reader.h" "${AOM_ROOT}/aom_dsp/binary_codes_reader.h"
@ -233,22 +239,17 @@ if (CONFIG_DECODERS)
set(AOM_DSP_DECODER_SOURCES set(AOM_DSP_DECODER_SOURCES
${AOM_DSP_DECODER_SOURCES} ${AOM_DSP_DECODER_SOURCES}
"${AOM_ROOT}/aom_dsp/ansreader.h") "${AOM_ROOT}/aom_dsp/ansreader.h")
elseif (CONFIG_DAALA_EC) else ()
set(AOM_DSP_DECODER_SOURCES set(AOM_DSP_DECODER_SOURCES
${AOM_DSP_DECODER_SOURCES} ${AOM_DSP_DECODER_SOURCES}
"${AOM_ROOT}/aom_dsp/daalaboolreader.c" "${AOM_ROOT}/aom_dsp/daalaboolreader.c"
"${AOM_ROOT}/aom_dsp/daalaboolreader.h" "${AOM_ROOT}/aom_dsp/daalaboolreader.h"
"${AOM_ROOT}/aom_dsp/entdec.c" "${AOM_ROOT}/aom_dsp/entdec.c"
"${AOM_ROOT}/aom_dsp/entdec.h") "${AOM_ROOT}/aom_dsp/entdec.h")
else ()
set(AOM_DSP_DECODER_SOURCES
${AOM_DSP_DECODER_SOURCES}
"${AOM_ROOT}/aom_dsp/dkboolreader.c"
"${AOM_ROOT}/aom_dsp/dkboolreader.h")
endif () endif ()
endif () endif ()
if (CONFIG_ENCODERS) if (CONFIG_AV1_ENCODER)
set(AOM_DSP_ENCODER_SOURCES set(AOM_DSP_ENCODER_SOURCES
"${AOM_ROOT}/aom_dsp/binary_codes_writer.c" "${AOM_ROOT}/aom_dsp/binary_codes_writer.c"
"${AOM_ROOT}/aom_dsp/binary_codes_writer.h" "${AOM_ROOT}/aom_dsp/binary_codes_writer.h"
@ -257,6 +258,7 @@ if (CONFIG_ENCODERS)
"${AOM_ROOT}/aom_dsp/bitwriter_buffer.h" "${AOM_ROOT}/aom_dsp/bitwriter_buffer.h"
"${AOM_ROOT}/aom_dsp/psnr.c" "${AOM_ROOT}/aom_dsp/psnr.c"
"${AOM_ROOT}/aom_dsp/psnr.h" "${AOM_ROOT}/aom_dsp/psnr.h"
"${AOM_ROOT}/aom_dsp/sad.c"
"${AOM_ROOT}/aom_dsp/variance.c" "${AOM_ROOT}/aom_dsp/variance.c"
"${AOM_ROOT}/aom_dsp/variance.h") "${AOM_ROOT}/aom_dsp/variance.h")
@ -282,6 +284,9 @@ if (CONFIG_ENCODERS)
set(AOM_DSP_ENCODER_ASM_SSE4_1 "${AOM_ROOT}/aom_dsp/x86/sad_sse4.asm") set(AOM_DSP_ENCODER_ASM_SSE4_1 "${AOM_ROOT}/aom_dsp/x86/sad_sse4.asm")
set(AOM_DSP_ENCODER_INTRIN_AVX2 set(AOM_DSP_ENCODER_INTRIN_AVX2
"${AOM_ROOT}/aom_dsp/x86/fwd_dct32x32_impl_avx2.h"
"${AOM_ROOT}/aom_dsp/x86/fwd_txfm_avx2.c"
"${AOM_ROOT}/aom_dsp/x86/fwd_txfm_avx2.h"
"${AOM_ROOT}/aom_dsp/x86/sad4d_avx2.c" "${AOM_ROOT}/aom_dsp/x86/sad4d_avx2.c"
"${AOM_ROOT}/aom_dsp/x86/sad_avx2.c" "${AOM_ROOT}/aom_dsp/x86/sad_avx2.c"
"${AOM_ROOT}/aom_dsp/x86/sad_impl_avx2.c" "${AOM_ROOT}/aom_dsp/x86/sad_impl_avx2.c"
@ -310,11 +315,6 @@ if (CONFIG_ENCODERS)
"${AOM_ROOT}/aom_dsp/x86/variance_sse2.c" "${AOM_ROOT}/aom_dsp/x86/variance_sse2.c"
"${AOM_ROOT}/aom_dsp/x86/sum_squares_sse2.c") "${AOM_ROOT}/aom_dsp/x86/sum_squares_sse2.c")
set(AOM_DSP_ENCODER_INTRIN_SSSE3
${AOM_DSP_ENCODER_INTRIN_SSSE3}
"${AOM_ROOT}/aom_dsp/x86/masked_sad_intrin_ssse3.c"
"${AOM_ROOT}/aom_dsp/x86/masked_variance_intrin_ssse3.c")
set(AOM_DSP_ENCODER_ASM_SSSE3_X86_64 set(AOM_DSP_ENCODER_ASM_SSSE3_X86_64
${AOM_DSP_ENCODER_ASM_SSSE3_X86_64} ${AOM_DSP_ENCODER_ASM_SSSE3_X86_64}
"${AOM_ROOT}/aom_dsp/x86/avg_ssse3_x86_64.asm" "${AOM_ROOT}/aom_dsp/x86/avg_ssse3_x86_64.asm"
@ -325,7 +325,6 @@ if (CONFIG_ENCODERS)
"${AOM_ROOT}/aom_dsp/x86/quantize_avx_x86_64.asm") "${AOM_ROOT}/aom_dsp/x86/quantize_avx_x86_64.asm")
set(AOM_DSP_ENCODER_INTRIN_MSA set(AOM_DSP_ENCODER_INTRIN_MSA
"${AOM_ROOT}/aom_dsp/mips/avg_msa.c"
"${AOM_ROOT}/aom_dsp/mips/sad_msa.c" "${AOM_ROOT}/aom_dsp/mips/sad_msa.c"
"${AOM_ROOT}/aom_dsp/mips/subtract_msa.c" "${AOM_ROOT}/aom_dsp/mips/subtract_msa.c"
"${AOM_ROOT}/aom_dsp/mips/variance_msa.c" "${AOM_ROOT}/aom_dsp/mips/variance_msa.c"
@ -345,9 +344,7 @@ if (CONFIG_ENCODERS)
"${AOM_ROOT}/aom_dsp/x86/highbd_sad4d_sse2.asm" "${AOM_ROOT}/aom_dsp/x86/highbd_sad4d_sse2.asm"
"${AOM_ROOT}/aom_dsp/x86/highbd_sad_sse2.asm" "${AOM_ROOT}/aom_dsp/x86/highbd_sad_sse2.asm"
"${AOM_ROOT}/aom_dsp/x86/highbd_subpel_variance_impl_sse2.asm" "${AOM_ROOT}/aom_dsp/x86/highbd_subpel_variance_impl_sse2.asm"
"${AOM_ROOT}/aom_dsp/x86/highbd_variance_impl_sse2.asm" "${AOM_ROOT}/aom_dsp/x86/highbd_variance_impl_sse2.asm")
"${AOM_ROOT}/aom_dsp/x86/aom_high_subpixel_8t_sse2.asm"
"${AOM_ROOT}/aom_dsp/x86/aom_high_subpixel_bilinear_sse2.asm")
set(AOM_DSP_ENCODER_INTRIN_SSE2 set(AOM_DSP_ENCODER_INTRIN_SSE2
${AOM_DSP_ENCODER_INTRIN_SSE2} ${AOM_DSP_ENCODER_INTRIN_SSE2}
@ -368,18 +365,13 @@ if (CONFIG_ENCODERS)
"${AOM_ROOT}/aom_dsp/answriter.h" "${AOM_ROOT}/aom_dsp/answriter.h"
"${AOM_ROOT}/aom_dsp/buf_ans.c" "${AOM_ROOT}/aom_dsp/buf_ans.c"
"${AOM_ROOT}/aom_dsp/buf_ans.h") "${AOM_ROOT}/aom_dsp/buf_ans.h")
elseif (CONFIG_DAALA_EC) else ()
set(AOM_DSP_ENCODER_SOURCES set(AOM_DSP_ENCODER_SOURCES
${AOM_DSP_ENCODER_SOURCES} ${AOM_DSP_ENCODER_SOURCES}
"${AOM_ROOT}/aom_dsp/daalaboolwriter.c" "${AOM_ROOT}/aom_dsp/daalaboolwriter.c"
"${AOM_ROOT}/aom_dsp/daalaboolwriter.h" "${AOM_ROOT}/aom_dsp/daalaboolwriter.h"
"${AOM_ROOT}/aom_dsp/entenc.c" "${AOM_ROOT}/aom_dsp/entenc.c"
"${AOM_ROOT}/aom_dsp/entenc.h") "${AOM_ROOT}/aom_dsp/entenc.h")
else ()
set(AOM_DSP_ENCODER_SOURCES
${AOM_DSP_ENCODER_SOURCES}
"${AOM_ROOT}/aom_dsp/dkboolwriter.c"
"${AOM_ROOT}/aom_dsp/dkboolwriter.h")
endif () endif ()
if (CONFIG_INTERNAL_STATS) if (CONFIG_INTERNAL_STATS)
@ -392,6 +384,18 @@ if (CONFIG_ENCODERS)
endif () endif ()
endif () endif ()
if (CONFIG_LOOP_RESTORATION)
set(AOM_DSP_COMMON_INTRIN_SSE2
${AOM_DSP_COMMON_INTRIN_SSE2}
"${AOM_ROOT}/aom_dsp/x86/aom_convolve_hip_sse2.c")
if (CONFIG_HIGHBITDEPTH)
set(AOM_DSP_COMMON_INTRIN_SSSE3
${AOM_DSP_COMMON_INTRIN_SSSE3}
"${AOM_ROOT}/aom_dsp/x86/aom_highbd_convolve_hip_ssse3.c")
endif ()
endif ()
if (CONFIG_MOTION_VAR) if (CONFIG_MOTION_VAR)
set(AOM_DSP_ENCODER_INTRIN_SSE4_1 set(AOM_DSP_ENCODER_INTRIN_SSE4_1
${AOM_DSP_ENCODER_INTRIN_SSE4_1} ${AOM_DSP_ENCODER_INTRIN_SSE4_1}
@ -406,13 +410,13 @@ function (setup_aom_dsp_targets)
set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_dsp_common) set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_dsp_common)
target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_dsp_common>) target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_dsp_common>)
if (CONFIG_DECODERS) if (CONFIG_AV1_DECODER)
add_library(aom_dsp_decoder OBJECT ${AOM_DSP_DECODER_SOURCES}) add_library(aom_dsp_decoder OBJECT ${AOM_DSP_DECODER_SOURCES})
set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_dsp_decoder) set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_dsp_decoder)
target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_dsp_decoder>) target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_dsp_decoder>)
endif () endif ()
if (CONFIG_ENCODERS) if (CONFIG_AV1_ENCODER)
add_library(aom_dsp_encoder OBJECT ${AOM_DSP_ENCODER_SOURCES}) add_library(aom_dsp_encoder OBJECT ${AOM_DSP_ENCODER_SOURCES})
set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_dsp_encoder) set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_dsp_encoder)
target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_dsp_encoder>) target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_dsp_encoder>)
@ -422,14 +426,14 @@ function (setup_aom_dsp_targets)
add_asm_library("aom_dsp_common_sse2" "AOM_DSP_COMMON_ASM_SSE2" "aom") add_asm_library("aom_dsp_common_sse2" "AOM_DSP_COMMON_ASM_SSE2" "aom")
add_intrinsics_object_library("-msse2" "sse2" "aom_dsp_common" add_intrinsics_object_library("-msse2" "sse2" "aom_dsp_common"
"AOM_DSP_COMMON_INTRIN_SSE2") "AOM_DSP_COMMON_INTRIN_SSE2")
if (CONFIG_ENCODERS) if (CONFIG_AV1_ENCODER)
add_asm_library("aom_dsp_encoder_sse2" "AOM_DSP_ENCODER_ASM_SSE2" "aom") add_asm_library("aom_dsp_encoder_sse2" "AOM_DSP_ENCODER_ASM_SSE2" "aom")
add_intrinsics_object_library("-msse2" "sse2" "aom_dsp_encoder" add_intrinsics_object_library("-msse2" "sse2" "aom_dsp_encoder"
"AOM_DSP_ENCODER_INTRIN_SSE2") "AOM_DSP_ENCODER_INTRIN_SSE2")
endif() endif()
endif () endif ()
if (HAVE_SSE3 AND CONFIG_ENCODERS) if (HAVE_SSE3 AND CONFIG_AV1_ENCODER)
add_asm_library("aom_dsp_encoder_sse3" "AOM_DSP_ENCODER_INTRIN_SSE3" "aom") add_asm_library("aom_dsp_encoder_sse3" "AOM_DSP_ENCODER_INTRIN_SSE3" "aom")
endif () endif ()
@ -438,21 +442,19 @@ function (setup_aom_dsp_targets)
add_intrinsics_object_library("-mssse3" "ssse3" "aom_dsp_common" add_intrinsics_object_library("-mssse3" "ssse3" "aom_dsp_common"
"AOM_DSP_COMMON_INTRIN_SSSE3") "AOM_DSP_COMMON_INTRIN_SSSE3")
if (CONFIG_ENCODERS) if (CONFIG_AV1_ENCODER)
if ("${AOM_TARGET_CPU}" STREQUAL "x86_64") if ("${AOM_TARGET_CPU}" STREQUAL "x86_64")
list(APPEND AOM_DSP_ENCODER_ASM_SSSE3 list(APPEND AOM_DSP_ENCODER_ASM_SSSE3
${AOM_DSP_ENCODER_ASM_SSSE3_X86_64}) ${AOM_DSP_ENCODER_ASM_SSSE3_X86_64})
endif () endif ()
add_asm_library("aom_dsp_encoder_ssse3" "AOM_DSP_ENCODER_ASM_SSSE3" "aom") add_asm_library("aom_dsp_encoder_ssse3" "AOM_DSP_ENCODER_ASM_SSSE3" "aom")
add_intrinsics_object_library("-mssse3" "ssse3" "aom_dsp_encoder"
"AOM_DSP_ENCODER_INTRIN_SSSE3")
endif () endif ()
endif () endif ()
if (HAVE_SSE4_1) if (HAVE_SSE4_1)
add_intrinsics_object_library("-msse4.1" "sse4_1" "aom_dsp_common" add_intrinsics_object_library("-msse4.1" "sse4_1" "aom_dsp_common"
"AOM_DSP_COMMON_INTRIN_SSE4_1") "AOM_DSP_COMMON_INTRIN_SSE4_1")
if (CONFIG_ENCODERS) if (CONFIG_AV1_ENCODER)
if (AOM_DSP_ENCODER_INTRIN_SSE4_1) if (AOM_DSP_ENCODER_INTRIN_SSE4_1)
add_intrinsics_object_library("-msse4.1" "sse4_1" "aom_dsp_encoder" add_intrinsics_object_library("-msse4.1" "sse4_1" "aom_dsp_encoder"
"AOM_DSP_ENCODER_INTRIN_SSE4_1") "AOM_DSP_ENCODER_INTRIN_SSE4_1")
@ -463,14 +465,16 @@ function (setup_aom_dsp_targets)
endif () endif ()
if (HAVE_AVX AND "${AOM_TARGET_CPU}" STREQUAL "x86_64") if (HAVE_AVX AND "${AOM_TARGET_CPU}" STREQUAL "x86_64")
add_asm_library("aom_dsp_encoder_avx" "AOM_DSP_ENCODER_AVX_ASM_X86_64" if (CONFIG_AV1_ENCODER)
"aom") add_asm_library("aom_dsp_encoder_avx" "AOM_DSP_ENCODER_AVX_ASM_X86_64"
"aom")
endif ()
endif () endif ()
if (HAVE_AVX2) if (HAVE_AVX2)
add_intrinsics_object_library("-mavx2" "avx2" "aom_dsp_common" add_intrinsics_object_library("-mavx2" "avx2" "aom_dsp_common"
"AOM_DSP_COMMON_INTRIN_AVX2") "AOM_DSP_COMMON_INTRIN_AVX2")
if (CONFIG_ENCODERS) if (CONFIG_AV1_ENCODER)
add_intrinsics_object_library("-mavx2" "avx2" "aom_dsp_encoder" add_intrinsics_object_library("-mavx2" "avx2" "aom_dsp_encoder"
"AOM_DSP_ENCODER_INTRIN_AVX2") "AOM_DSP_ENCODER_INTRIN_AVX2")
endif () endif ()
@ -497,7 +501,7 @@ function (setup_aom_dsp_targets)
if (HAVE_MSA) if (HAVE_MSA)
add_intrinsics_object_library("" "msa" "aom_dsp_common" add_intrinsics_object_library("" "msa" "aom_dsp_common"
"AOM_DSP_COMMON_INTRIN_MSA") "AOM_DSP_COMMON_INTRIN_MSA")
if (CONFIG_ENCODERS) if (CONFIG_AV1_ENCODER)
add_intrinsics_object_library("" "msa" "aom_dsp_encoder" add_intrinsics_object_library("" "msa" "aom_dsp_encoder"
"AOM_DSP_ENCODER_INTRIN_MSA") "AOM_DSP_ENCODER_INTRIN_MSA")
endif () endif ()
@ -507,3 +511,5 @@ function (setup_aom_dsp_targets)
# $AOM_LIB_TARGETS. # $AOM_LIB_TARGETS.
set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} PARENT_SCOPE) set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} PARENT_SCOPE)
endfunction () endfunction ()
endif () # AOM_AOM_DSP_AOM_DSP_CMAKE_

42
third_party/aom/aom_dsp/aom_dsp.mk vendored
View file

@ -22,19 +22,16 @@ DSP_SRCS-yes += prob.h
DSP_SRCS-yes += prob.c DSP_SRCS-yes += prob.c
DSP_SRCS-$(CONFIG_ANS) += ans.h DSP_SRCS-$(CONFIG_ANS) += ans.h
ifeq ($(CONFIG_ENCODERS),yes) ifeq ($(CONFIG_AV1_ENCODER),yes)
ifeq ($(CONFIG_ANS),yes) ifeq ($(CONFIG_ANS),yes)
DSP_SRCS-yes += answriter.h DSP_SRCS-yes += answriter.h
DSP_SRCS-yes += buf_ans.h DSP_SRCS-yes += buf_ans.h
DSP_SRCS-yes += buf_ans.c DSP_SRCS-yes += buf_ans.c
else ifeq ($(CONFIG_DAALA_EC),yes) else
DSP_SRCS-yes += entenc.c DSP_SRCS-yes += entenc.c
DSP_SRCS-yes += entenc.h DSP_SRCS-yes += entenc.h
DSP_SRCS-yes += daalaboolwriter.c DSP_SRCS-yes += daalaboolwriter.c
DSP_SRCS-yes += daalaboolwriter.h DSP_SRCS-yes += daalaboolwriter.h
else
DSP_SRCS-yes += dkboolwriter.h
DSP_SRCS-yes += dkboolwriter.c
endif endif
DSP_SRCS-yes += bitwriter.h DSP_SRCS-yes += bitwriter.h
DSP_SRCS-yes += bitwriter_buffer.c DSP_SRCS-yes += bitwriter_buffer.c
@ -49,17 +46,14 @@ DSP_SRCS-$(CONFIG_INTERNAL_STATS) += psnrhvs.c
DSP_SRCS-$(CONFIG_INTERNAL_STATS) += fastssim.c DSP_SRCS-$(CONFIG_INTERNAL_STATS) += fastssim.c
endif endif
ifeq ($(CONFIG_DECODERS),yes) ifeq ($(CONFIG_AV1_DECODER),yes)
ifeq ($(CONFIG_ANS),yes) ifeq ($(CONFIG_ANS),yes)
DSP_SRCS-yes += ansreader.h DSP_SRCS-yes += ansreader.h
else ifeq ($(CONFIG_DAALA_EC),yes) else
DSP_SRCS-yes += entdec.c DSP_SRCS-yes += entdec.c
DSP_SRCS-yes += entdec.h DSP_SRCS-yes += entdec.h
DSP_SRCS-yes += daalaboolreader.c DSP_SRCS-yes += daalaboolreader.c
DSP_SRCS-yes += daalaboolreader.h DSP_SRCS-yes += daalaboolreader.h
else
DSP_SRCS-yes += dkboolreader.h
DSP_SRCS-yes += dkboolreader.c
endif endif
DSP_SRCS-yes += bitreader.h DSP_SRCS-yes += bitreader.h
DSP_SRCS-yes += bitreader_buffer.c DSP_SRCS-yes += bitreader_buffer.c
@ -71,7 +65,7 @@ endif
# intra predictions # intra predictions
DSP_SRCS-yes += intrapred.c DSP_SRCS-yes += intrapred.c
ifeq ($(CONFIG_DAALA_EC),yes) ifneq ($(CONFIG_ANS),yes)
DSP_SRCS-yes += entcode.c DSP_SRCS-yes += entcode.c
DSP_SRCS-yes += entcode.h DSP_SRCS-yes += entcode.h
endif endif
@ -205,6 +199,7 @@ endif # CONFIG_HIGHBITDEPTH
DSP_SRCS-yes += txfm_common.h DSP_SRCS-yes += txfm_common.h
DSP_SRCS-yes += x86/txfm_common_intrin.h DSP_SRCS-yes += x86/txfm_common_intrin.h
DSP_SRCS-$(HAVE_SSE2) += x86/txfm_common_sse2.h DSP_SRCS-$(HAVE_SSE2) += x86/txfm_common_sse2.h
DSP_SRCS-$(HAVE_SSSE3) += x86/obmc_intrinsic_ssse3.h
DSP_SRCS-$(HAVE_MSA) += mips/txfm_macros_msa.h DSP_SRCS-$(HAVE_MSA) += mips/txfm_macros_msa.h
# forward transform # forward transform
@ -239,6 +234,8 @@ DSP_SRCS-$(HAVE_SSE2) += x86/inv_txfm_sse2.h
DSP_SRCS-$(HAVE_SSE2) += x86/inv_txfm_sse2.c DSP_SRCS-$(HAVE_SSE2) += x86/inv_txfm_sse2.c
DSP_SRCS-$(HAVE_SSE2) += x86/inv_wht_sse2.asm DSP_SRCS-$(HAVE_SSE2) += x86/inv_wht_sse2.asm
DSP_SRCS-$(HAVE_SSSE3) += x86/inv_txfm_ssse3.c DSP_SRCS-$(HAVE_SSSE3) += x86/inv_txfm_ssse3.c
DSP_SRCS-$(HAVE_AVX2) += x86/inv_txfm_common_avx2.h
DSP_SRCS-$(HAVE_AVX2) += x86/inv_txfm_avx2.c
ifeq ($(HAVE_NEON_ASM),yes) ifeq ($(HAVE_NEON_ASM),yes)
DSP_SRCS-yes += arm/save_reg_neon$(ASM) DSP_SRCS-yes += arm/save_reg_neon$(ASM)
@ -278,6 +275,13 @@ DSP_SRCS-$(HAVE_DSPR2) += mips/itrans16_dspr2.c
DSP_SRCS-$(HAVE_DSPR2) += mips/itrans32_dspr2.c DSP_SRCS-$(HAVE_DSPR2) += mips/itrans32_dspr2.c
DSP_SRCS-$(HAVE_DSPR2) += mips/itrans32_cols_dspr2.c DSP_SRCS-$(HAVE_DSPR2) += mips/itrans32_cols_dspr2.c
endif # CONFIG_HIGHBITDEPTH endif # CONFIG_HIGHBITDEPTH
ifeq ($(CONFIG_LOOP_RESTORATION),yes)
DSP_SRCS-$(HAVE_SSE2) += x86/aom_convolve_hip_sse2.c
ifeq ($(CONFIG_HIGHBITDEPTH),yes)
DSP_SRCS-$(HAVE_SSSE3) += x86/aom_highbd_convolve_hip_ssse3.c
endif
endif # CONFIG_LOOP_RESTORATION
endif # CONFIG_AV1 endif # CONFIG_AV1
# quantization # quantization
@ -298,7 +302,6 @@ endif
DSP_SRCS-yes += avg.c DSP_SRCS-yes += avg.c
DSP_SRCS-$(HAVE_SSE2) += x86/avg_intrin_sse2.c DSP_SRCS-$(HAVE_SSE2) += x86/avg_intrin_sse2.c
DSP_SRCS-$(HAVE_NEON) += arm/avg_neon.c DSP_SRCS-$(HAVE_NEON) += arm/avg_neon.c
DSP_SRCS-$(HAVE_MSA) += mips/avg_msa.c
DSP_SRCS-$(HAVE_NEON) += arm/hadamard_neon.c DSP_SRCS-$(HAVE_NEON) += arm/hadamard_neon.c
ifeq ($(ARCH_X86_64),yes) ifeq ($(ARCH_X86_64),yes)
DSP_SRCS-$(HAVE_SSSE3) += x86/avg_ssse3_x86_64.asm DSP_SRCS-$(HAVE_SSSE3) += x86/avg_ssse3_x86_64.asm
@ -317,11 +320,10 @@ DSP_SRCS-yes += sum_squares.c
DSP_SRCS-$(HAVE_SSE2) += x86/sum_squares_sse2.c DSP_SRCS-$(HAVE_SSE2) += x86/sum_squares_sse2.c
endif # CONFIG_AV1_ENCODER endif # CONFIG_AV1_ENCODER
ifeq ($(CONFIG_ENCODERS),yes) ifeq ($(CONFIG_AV1_ENCODER),yes)
DSP_SRCS-yes += sad.c DSP_SRCS-yes += sad.c
DSP_SRCS-yes += subtract.c DSP_SRCS-yes += subtract.c
DSP_SRCS-$(HAVE_MEDIA) += arm/sad_media$(ASM)
DSP_SRCS-$(HAVE_NEON) += arm/sad4d_neon.c DSP_SRCS-$(HAVE_NEON) += arm/sad4d_neon.c
DSP_SRCS-$(HAVE_NEON) += arm/sad_neon.c DSP_SRCS-$(HAVE_NEON) += arm/sad_neon.c
DSP_SRCS-$(HAVE_NEON) += arm/subtract_neon.c DSP_SRCS-$(HAVE_NEON) += arm/subtract_neon.c
@ -364,18 +366,12 @@ DSP_SRCS-$(HAVE_SSE2) += x86/highbd_sad4d_sse2.asm
DSP_SRCS-$(HAVE_SSE2) += x86/highbd_sad_sse2.asm DSP_SRCS-$(HAVE_SSE2) += x86/highbd_sad_sse2.asm
endif # CONFIG_HIGHBITDEPTH endif # CONFIG_HIGHBITDEPTH
endif # CONFIG_ENCODERS endif # CONFIG_AV1_ENCODER
ifneq ($(filter yes,$(CONFIG_ENCODERS)),) ifneq ($(filter yes,$(CONFIG_AV1_ENCODER)),)
DSP_SRCS-yes += variance.c DSP_SRCS-yes += variance.c
DSP_SRCS-yes += variance.h DSP_SRCS-yes += variance.h
DSP_SRCS-$(HAVE_MEDIA) += arm/bilinear_filter_media$(ASM)
DSP_SRCS-$(HAVE_MEDIA) += arm/subpel_variance_media.c
DSP_SRCS-$(HAVE_MEDIA) += arm/variance_halfpixvar16x16_h_media$(ASM)
DSP_SRCS-$(HAVE_MEDIA) += arm/variance_halfpixvar16x16_hv_media$(ASM)
DSP_SRCS-$(HAVE_MEDIA) += arm/variance_halfpixvar16x16_v_media$(ASM)
DSP_SRCS-$(HAVE_MEDIA) += arm/variance_media$(ASM)
DSP_SRCS-$(HAVE_NEON) += arm/subpel_variance_neon.c DSP_SRCS-$(HAVE_NEON) += arm/subpel_variance_neon.c
DSP_SRCS-$(HAVE_NEON) += arm/variance_neon.c DSP_SRCS-$(HAVE_NEON) += arm/variance_neon.c
@ -402,7 +398,7 @@ DSP_SRCS-$(HAVE_SSE4_1) += x86/highbd_variance_sse4.c
DSP_SRCS-$(HAVE_SSE2) += x86/highbd_variance_impl_sse2.asm DSP_SRCS-$(HAVE_SSE2) += x86/highbd_variance_impl_sse2.asm
DSP_SRCS-$(HAVE_SSE2) += x86/highbd_subpel_variance_impl_sse2.asm DSP_SRCS-$(HAVE_SSE2) += x86/highbd_subpel_variance_impl_sse2.asm
endif # CONFIG_HIGHBITDEPTH endif # CONFIG_HIGHBITDEPTH
endif # CONFIG_ENCODERS endif # CONFIG_AV1_ENCODER
DSP_SRCS-no += $(DSP_SRCS_REMOVE-yes) DSP_SRCS-no += $(DSP_SRCS_REMOVE-yes)

2
third_party/aom/aom_dsp/aom_dsp_common.h vendored
View file

@ -31,6 +31,8 @@ extern "C" {
#define AOMMIN(x, y) (((x) < (y)) ? (x) : (y)) #define AOMMIN(x, y) (((x) < (y)) ? (x) : (y))
#define AOMMAX(x, y) (((x) > (y)) ? (x) : (y)) #define AOMMAX(x, y) (((x) > (y)) ? (x) : (y))
#define NELEMENTS(x) (sizeof((x)) / sizeof((x)[0]))
#define IMPLIES(a, b) (!(a) || (b)) // Logical 'a implies b' (or 'a -> b') #define IMPLIES(a, b) (!(a) || (b)) // Logical 'a implies b' (or 'a -> b')
#define IS_POWER_OF_TWO(x) (((x) & ((x)-1)) == 0) #define IS_POWER_OF_TWO(x) (((x) & ((x)-1)) == 0)

175
third_party/aom/aom_dsp/aom_dsp_rtcd_defs.pl vendored
View file

@ -49,6 +49,9 @@ if (aom_config("CONFIG_TX64X64") eq "yes") {
@pred_names = qw/dc dc_top dc_left dc_128 v h d207e d63e d45e d117 d135 d153/; @pred_names = qw/dc dc_top dc_left dc_128 v h d207e d63e d45e d117 d135 d153/;
if (aom_config("CONFIG_ALT_INTRA") eq "yes") { if (aom_config("CONFIG_ALT_INTRA") eq "yes") {
push @pred_names, qw/paeth smooth/; push @pred_names, qw/paeth smooth/;
if (aom_config("CONFIG_SMOOTH_HV") eq "yes") {
push @pred_names, qw/smooth_v smooth_h/;
}
} else { } else {
push @pred_names, 'tm'; push @pred_names, 'tm';
} }
@ -168,10 +171,14 @@ if (aom_config("CONFIG_LOOP_RESTORATION") eq "yes") {
add_proto qw/void aom_convolve8_add_src/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; add_proto qw/void aom_convolve8_add_src/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
add_proto qw/void aom_convolve8_add_src_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; add_proto qw/void aom_convolve8_add_src_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
add_proto qw/void aom_convolve8_add_src_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; add_proto qw/void aom_convolve8_add_src_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
add_proto qw/void aom_convolve8_add_src_hip/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
add_proto qw/void aom_convolve8_add_src_horiz_hip/, "const uint8_t *src, ptrdiff_t src_stride, uint16_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
add_proto qw/void aom_convolve8_add_src_vert_hip/, "const uint16_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
specialize qw/aom_convolve8_add_src ssse3/; specialize qw/aom_convolve8_add_src ssse3/;
specialize qw/aom_convolve8_add_src_horiz ssse3/; specialize qw/aom_convolve8_add_src_horiz ssse3/;
specialize qw/aom_convolve8_add_src_vert ssse3/; specialize qw/aom_convolve8_add_src_vert ssse3/;
specialize qw/aom_convolve8_add_src_hip sse2/;
} # CONFIG_LOOP_RESTORATION } # CONFIG_LOOP_RESTORATION
# TODO(any): These need to be extended to up to 128x128 block sizes # TODO(any): These need to be extended to up to 128x128 block sizes
@ -215,8 +222,12 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
add_proto qw/void aom_highbd_convolve8_add_src/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps"; add_proto qw/void aom_highbd_convolve8_add_src/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
add_proto qw/void aom_highbd_convolve8_add_src_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps"; add_proto qw/void aom_highbd_convolve8_add_src_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
add_proto qw/void aom_highbd_convolve8_add_src_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps"; add_proto qw/void aom_highbd_convolve8_add_src_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
add_proto qw/void aom_highbd_convolve8_add_src_hip/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
add_proto qw/void aom_highbd_convolve8_add_src_horiz_hip/, "const uint8_t *src, ptrdiff_t src_stride, uint16_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
add_proto qw/void aom_highbd_convolve8_add_src_vert_hip/, "const uint16_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
specialize qw/aom_highbd_convolve8_add_src/, "$sse2_x86_64"; specialize qw/aom_highbd_convolve8_add_src/, "$sse2_x86_64";
specialize qw/aom_highbd_convolve8_add_src_hip ssse3/;
# The _horiz/_vert functions are currently unused, so we don't bother # The _horiz/_vert functions are currently unused, so we don't bother
# specialising them. # specialising them.
} # CONFIG_LOOP_RESTORATION } # CONFIG_LOOP_RESTORATION
@ -434,29 +445,30 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
specialize qw/aom_idct8x8_1_add sse2/; specialize qw/aom_idct8x8_1_add sse2/;
add_proto qw/void aom_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_256_add sse2/; specialize qw/aom_idct16x16_256_add sse2 avx2/;
add_proto qw/void aom_idct16x16_38_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_38_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_38_add avx2/;
add_proto qw/void aom_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_10_add sse2/; specialize qw/aom_idct16x16_10_add sse2 avx2/;
add_proto qw/void aom_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_1_add sse2/; specialize qw/aom_idct16x16_1_add sse2 avx2/;
add_proto qw/void aom_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_1024_add sse2 ssse3/; specialize qw/aom_idct32x32_1024_add sse2 ssse3 avx2/;
add_proto qw/void aom_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_135_add sse2 ssse3/; specialize qw/aom_idct32x32_135_add sse2 ssse3 avx2/;
# Need to add 135 eob idct32x32 implementations. # Need to add 135 eob idct32x32 implementations.
$aom_idct32x32_135_add_sse2=aom_idct32x32_1024_add_sse2; $aom_idct32x32_135_add_sse2=aom_idct32x32_1024_add_sse2;
add_proto qw/void aom_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_34_add sse2 ssse3/; specialize qw/aom_idct32x32_34_add sse2 ssse3 avx2/;
add_proto qw/void aom_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_1_add sse2/; specialize qw/aom_idct32x32_1_add sse2 avx2/;
add_proto qw/void aom_highbd_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd"; add_proto qw/void aom_highbd_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
specialize qw/aom_highbd_idct4x4_16_add sse2/; specialize qw/aom_highbd_idct4x4_16_add sse2/;
@ -479,21 +491,22 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
specialize qw/aom_idct8x8_12_add sse2 ssse3 neon dspr2 msa/; specialize qw/aom_idct8x8_12_add sse2 ssse3 neon dspr2 msa/;
add_proto qw/void aom_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_1_add sse2 neon dspr2 msa/; specialize qw/aom_idct16x16_1_add sse2 avx2 neon dspr2 msa/;
add_proto qw/void aom_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_256_add sse2 neon dspr2 msa/; specialize qw/aom_idct16x16_256_add sse2 avx2 neon dspr2 msa/;
add_proto qw/void aom_idct16x16_38_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_38_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_38_add avx2/;
add_proto qw/void aom_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct16x16_10_add sse2 neon dspr2 msa/; specialize qw/aom_idct16x16_10_add sse2 avx2 neon dspr2 msa/;
add_proto qw/void aom_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_1024_add sse2 ssse3 neon dspr2 msa/; specialize qw/aom_idct32x32_1024_add sse2 ssse3 avx2 neon dspr2 msa/;
add_proto qw/void aom_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_135_add sse2 ssse3 neon dspr2 msa/; specialize qw/aom_idct32x32_135_add sse2 ssse3 avx2 neon dspr2 msa/;
# Need to add 135 eob idct32x32 implementations. # Need to add 135 eob idct32x32 implementations.
$aom_idct32x32_135_add_sse2=aom_idct32x32_1024_add_sse2; $aom_idct32x32_135_add_sse2=aom_idct32x32_1024_add_sse2;
$aom_idct32x32_135_add_neon=aom_idct32x32_1024_add_neon; $aom_idct32x32_135_add_neon=aom_idct32x32_1024_add_neon;
@ -501,12 +514,12 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
$aom_idct32x32_135_add_msa=aom_idct32x32_1024_add_msa; $aom_idct32x32_135_add_msa=aom_idct32x32_1024_add_msa;
add_proto qw/void aom_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_34_add sse2 ssse3 neon dspr2 msa/; specialize qw/aom_idct32x32_34_add sse2 ssse3 avx2 neon dspr2 msa/;
# Need to add 34 eob idct32x32 neon implementation. # Need to add 34 eob idct32x32 neon implementation.
$aom_idct32x32_34_add_neon=aom_idct32x32_1024_add_neon; $aom_idct32x32_34_add_neon=aom_idct32x32_1024_add_neon;
add_proto qw/void aom_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_idct32x32_1_add sse2 neon dspr2 msa/; specialize qw/aom_idct32x32_1_add sse2 avx2 neon dspr2 msa/;
add_proto qw/void aom_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride"; add_proto qw/void aom_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
specialize qw/aom_iwht4x4_1_add msa/; specialize qw/aom_iwht4x4_1_add msa/;
@ -578,7 +591,7 @@ if (aom_config("CONFIG_AV1") eq "yes") {
} }
} # CONFIG_AV1 } # CONFIG_AV1
if (aom_config("CONFIG_ENCODERS") eq "yes") { if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
# #
# Block subtraction # Block subtraction
# #
@ -604,13 +617,8 @@ if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
# #
# Avg # Avg
# #
add_proto qw/unsigned int aom_avg_8x8/, "const uint8_t *, int p";
specialize qw/aom_avg_8x8 sse2 neon msa/; specialize qw/aom_avg_8x8 sse2 neon msa/;
add_proto qw/unsigned int aom_avg_4x4/, "const uint8_t *, int p";
specialize qw/aom_avg_4x4 sse2 neon msa/;
if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") { if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
add_proto qw/unsigned int aom_highbd_avg_8x8/, "const uint8_t *, int p";
add_proto qw/unsigned int aom_highbd_avg_4x4/, "const uint8_t *, int p";
add_proto qw/void aom_highbd_subtract_block/, "int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride, int bd"; add_proto qw/void aom_highbd_subtract_block/, "int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride, int bd";
specialize qw/aom_highbd_subtract_block sse2/; specialize qw/aom_highbd_subtract_block sse2/;
} }
@ -652,22 +660,22 @@ foreach (@block_sizes) {
add_proto qw/unsigned int/, "aom_sad${w}x${h}_avg", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred"; add_proto qw/unsigned int/, "aom_sad${w}x${h}_avg", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
} }
specialize qw/aom_sad128x128 avx2 sse2/; specialize qw/aom_sad128x128 avx2 sse2/;
specialize qw/aom_sad128x64 avx2 sse2/; specialize qw/aom_sad128x64 avx2 sse2/;
specialize qw/aom_sad64x128 avx2 sse2/; specialize qw/aom_sad64x128 avx2 sse2/;
specialize qw/aom_sad64x64 avx2 neon msa sse2/; specialize qw/aom_sad64x64 avx2 neon msa sse2/;
specialize qw/aom_sad64x32 avx2 msa sse2/; specialize qw/aom_sad64x32 avx2 msa sse2/;
specialize qw/aom_sad32x64 avx2 msa sse2/; specialize qw/aom_sad32x64 avx2 msa sse2/;
specialize qw/aom_sad32x32 avx2 neon msa sse2/; specialize qw/aom_sad32x32 avx2 neon msa sse2/;
specialize qw/aom_sad32x16 avx2 msa sse2/; specialize qw/aom_sad32x16 avx2 msa sse2/;
specialize qw/aom_sad16x32 msa sse2/; specialize qw/aom_sad16x32 msa sse2/;
specialize qw/aom_sad16x16 media neon msa sse2/; specialize qw/aom_sad16x16 neon msa sse2/;
specialize qw/aom_sad16x8 neon msa sse2/; specialize qw/aom_sad16x8 neon msa sse2/;
specialize qw/aom_sad8x16 neon msa sse2/; specialize qw/aom_sad8x16 neon msa sse2/;
specialize qw/aom_sad8x8 neon msa sse2/; specialize qw/aom_sad8x8 neon msa sse2/;
specialize qw/aom_sad8x4 msa sse2/; specialize qw/aom_sad8x4 msa sse2/;
specialize qw/aom_sad4x8 msa sse2/; specialize qw/aom_sad4x8 msa sse2/;
specialize qw/aom_sad4x4 neon msa sse2/; specialize qw/aom_sad4x4 neon msa sse2/;
specialize qw/aom_sad128x128_avg avx2 sse2/; specialize qw/aom_sad128x128_avg avx2 sse2/;
specialize qw/aom_sad128x64_avg avx2 sse2/; specialize qw/aom_sad128x64_avg avx2 sse2/;
@ -727,14 +735,14 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
if (aom_config("CONFIG_EXT_INTER") eq "yes") { if (aom_config("CONFIG_EXT_INTER") eq "yes") {
foreach (@block_sizes) { foreach (@block_sizes) {
($w, $h) = @$_; ($w, $h) = @$_;
add_proto qw/unsigned int/, "aom_masked_sad${w}x${h}", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride"; add_proto qw/unsigned int/, "aom_masked_sad${w}x${h}", "const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask";
specialize "aom_masked_sad${w}x${h}", qw/ssse3/; specialize "aom_masked_sad${w}x${h}", qw/ssse3/;
} }
if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") { if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
foreach (@block_sizes) { foreach (@block_sizes) {
($w, $h) = @$_; ($w, $h) = @$_;
add_proto qw/unsigned int/, "aom_highbd_masked_sad${w}x${h}", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride"; add_proto qw/unsigned int/, "aom_highbd_masked_sad${w}x${h}", "const uint8_t *src8, int src_stride, const uint8_t *ref8, int ref_stride, const uint8_t *second_pred8, const uint8_t *msk, int msk_stride, int invert_mask";
specialize "aom_highbd_masked_sad${w}x${h}", qw/ssse3/; specialize "aom_highbd_masked_sad${w}x${h}", qw/ssse3/;
} }
} }
@ -876,9 +884,9 @@ if (aom_config("CONFIG_INTERNAL_STATS") eq "yes") {
add_proto qw/void aom_highbd_ssim_parms_8x8/, "const uint16_t *s, int sp, const uint16_t *r, int rp, uint32_t *sum_s, uint32_t *sum_r, uint32_t *sum_sq_s, uint32_t *sum_sq_r, uint32_t *sum_sxr"; add_proto qw/void aom_highbd_ssim_parms_8x8/, "const uint16_t *s, int sp, const uint16_t *r, int rp, uint32_t *sum_s, uint32_t *sum_r, uint32_t *sum_sq_s, uint32_t *sum_sq_r, uint32_t *sum_sxr";
} }
} }
} # CONFIG_ENCODERS } # CONFIG_AV1_ENCODER
if (aom_config("CONFIG_ENCODERS") eq "yes") { if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
# #
# Specialty Variance # Specialty Variance
@ -896,10 +904,10 @@ add_proto qw/unsigned int aom_mse16x8/, "const uint8_t *src_ptr, int source_str
add_proto qw/unsigned int aom_mse8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse"; add_proto qw/unsigned int aom_mse8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse";
add_proto qw/unsigned int aom_mse8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse"; add_proto qw/unsigned int aom_mse8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse";
specialize qw/aom_mse16x16 sse2 avx2 media neon msa/; specialize qw/aom_mse16x16 sse2 avx2 neon msa/;
specialize qw/aom_mse16x8 sse2 msa/; specialize qw/aom_mse16x8 sse2 msa/;
specialize qw/aom_mse8x16 sse2 msa/; specialize qw/aom_mse8x16 sse2 msa/;
specialize qw/aom_mse8x8 sse2 msa/; specialize qw/aom_mse8x8 sse2 msa/;
if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") { if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
foreach $bd (8, 10, 12) { foreach $bd (8, 10, 12) {
@ -956,33 +964,33 @@ foreach (@block_sizes) {
add_proto qw/uint32_t/, "aom_sub_pixel_avg_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred"; add_proto qw/uint32_t/, "aom_sub_pixel_avg_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
} }
specialize qw/aom_variance64x64 sse2 avx2 neon msa/; specialize qw/aom_variance64x64 sse2 avx2 neon msa/;
specialize qw/aom_variance64x32 sse2 avx2 neon msa/; specialize qw/aom_variance64x32 sse2 avx2 neon msa/;
specialize qw/aom_variance32x64 sse2 neon msa/; specialize qw/aom_variance32x64 sse2 neon msa/;
specialize qw/aom_variance32x32 sse2 avx2 neon msa/; specialize qw/aom_variance32x32 sse2 avx2 neon msa/;
specialize qw/aom_variance32x16 sse2 avx2 msa/; specialize qw/aom_variance32x16 sse2 avx2 msa/;
specialize qw/aom_variance16x32 sse2 msa/; specialize qw/aom_variance16x32 sse2 msa/;
specialize qw/aom_variance16x16 sse2 avx2 media neon msa/; specialize qw/aom_variance16x16 sse2 avx2 neon msa/;
specialize qw/aom_variance16x8 sse2 neon msa/; specialize qw/aom_variance16x8 sse2 neon msa/;
specialize qw/aom_variance8x16 sse2 neon msa/; specialize qw/aom_variance8x16 sse2 neon msa/;
specialize qw/aom_variance8x8 sse2 media neon msa/; specialize qw/aom_variance8x8 sse2 neon msa/;
specialize qw/aom_variance8x4 sse2 msa/; specialize qw/aom_variance8x4 sse2 msa/;
specialize qw/aom_variance4x8 sse2 msa/; specialize qw/aom_variance4x8 sse2 msa/;
specialize qw/aom_variance4x4 sse2 msa/; specialize qw/aom_variance4x4 sse2 msa/;
specialize qw/aom_sub_pixel_variance64x64 avx2 neon msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance64x64 avx2 neon msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance64x32 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance64x32 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance32x64 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance32x64 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance32x32 avx2 neon msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance32x32 avx2 neon msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance32x16 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance32x16 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance16x32 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance16x32 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance16x16 media neon msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance16x16 neon msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance16x8 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance16x8 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance8x16 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance8x16 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance8x8 media neon msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance8x8 neon msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance8x4 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance8x4 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance4x8 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance4x8 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_variance4x4 msa sse2 ssse3/; specialize qw/aom_sub_pixel_variance4x4 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_avg_variance64x64 avx2 msa sse2 ssse3/; specialize qw/aom_sub_pixel_avg_variance64x64 avx2 msa sse2 ssse3/;
specialize qw/aom_sub_pixel_avg_variance64x32 msa sse2 ssse3/; specialize qw/aom_sub_pixel_avg_variance64x32 msa sse2 ssse3/;
@ -1034,19 +1042,15 @@ if (aom_config("CONFIG_EXT_INTER") eq "yes") {
# #
foreach (@block_sizes) { foreach (@block_sizes) {
($w, $h) = @$_; ($w, $h) = @$_;
add_proto qw/unsigned int/, "aom_masked_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse"; add_proto qw/unsigned int/, "aom_masked_sub_pixel_variance${w}x${h}", "const uint8_t *src, int src_stride, int xoffset, int yoffset, const uint8_t *ref, int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask, unsigned int *sse";
add_proto qw/unsigned int/, "aom_masked_sub_pixel_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize "aom_masked_variance${w}x${h}", qw/ssse3/;
specialize "aom_masked_sub_pixel_variance${w}x${h}", qw/ssse3/; specialize "aom_masked_sub_pixel_variance${w}x${h}", qw/ssse3/;
} }
if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") { if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
foreach $bd ("_", "_10_", "_12_") { foreach $bd ("_8_", "_10_", "_12_") {
foreach (@block_sizes) { foreach (@block_sizes) {
($w, $h) = @$_; ($w, $h) = @$_;
add_proto qw/unsigned int/, "aom_highbd${bd}masked_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse"; add_proto qw/unsigned int/, "aom_highbd${bd}masked_sub_pixel_variance${w}x${h}", "const uint8_t *src, int src_stride, int xoffset, int yoffset, const uint8_t *ref, int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask, unsigned int *sse";
add_proto qw/unsigned int/, "aom_highbd${bd}masked_sub_pixel_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize "aom_highbd${bd}masked_variance${w}x${h}", qw/ssse3/;
specialize "aom_highbd${bd}masked_sub_pixel_variance${w}x${h}", qw/ssse3/; specialize "aom_highbd${bd}masked_sub_pixel_variance${w}x${h}", qw/ssse3/;
} }
} }
@ -1119,13 +1123,13 @@ add_proto qw/uint32_t aom_sub_pixel_avg_variance4x4/, "const uint8_t *src_ptr, i
# Specialty Subpixel # Specialty Subpixel
# #
add_proto qw/uint32_t aom_variance_halfpixvar16x16_h/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, uint32_t *sse"; add_proto qw/uint32_t aom_variance_halfpixvar16x16_h/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, uint32_t *sse";
specialize qw/aom_variance_halfpixvar16x16_h sse2 media/; specialize qw/aom_variance_halfpixvar16x16_h sse2/;
add_proto qw/uint32_t aom_variance_halfpixvar16x16_v/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, uint32_t *sse"; add_proto qw/uint32_t aom_variance_halfpixvar16x16_v/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, uint32_t *sse";
specialize qw/aom_variance_halfpixvar16x16_v sse2 media/; specialize qw/aom_variance_halfpixvar16x16_v sse2/;
add_proto qw/uint32_t aom_variance_halfpixvar16x16_hv/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, uint32_t *sse"; add_proto qw/uint32_t aom_variance_halfpixvar16x16_hv/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, uint32_t *sse";
specialize qw/aom_variance_halfpixvar16x16_hv sse2 media/; specialize qw/aom_variance_halfpixvar16x16_hv sse2/;
# #
# Comp Avg # Comp Avg
@ -1490,6 +1494,15 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
} # CONFIG_HIGHBITDEPTH } # CONFIG_HIGHBITDEPTH
} # CONFIG_ENCODERS if (aom_config("CONFIG_EXT_INTER") eq "yes") {
add_proto qw/void aom_comp_mask_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride, const uint8_t *mask, int mask_stride, int invert_mask";
add_proto qw/void aom_comp_mask_upsampled_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride, const uint8_t *mask, int mask_stride, int invert_mask";
if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
add_proto qw/void aom_highbd_comp_mask_pred/, "uint16_t *comp_pred, const uint8_t *pred8, int width, int height, const uint8_t *ref8, int ref_stride, const uint8_t *mask, int mask_stride, int invert_mask";
add_proto qw/void aom_highbd_comp_mask_upsampled_pred/, "uint16_t *comp_pred, const uint8_t *pred8, int width, int height, const uint8_t *ref8, int ref_stride, const uint8_t *mask, int mask_stride, int invert_mask";
}
}
} # CONFIG_AV1_ENCODER
1; 1;

38
third_party/aom/aom_dsp/arm/avg_neon.c vendored
View file

@ -25,44 +25,6 @@ static INLINE unsigned int horizontal_add_u16x8(const uint16x8_t v_16x8) {
return vget_lane_u32(c, 0); return vget_lane_u32(c, 0);
} }
unsigned int aom_avg_4x4_neon(const uint8_t *s, int p) {
uint16x8_t v_sum;
uint32x2_t v_s0 = vdup_n_u32(0);
uint32x2_t v_s1 = vdup_n_u32(0);
v_s0 = vld1_lane_u32((const uint32_t *)s, v_s0, 0);
v_s0 = vld1_lane_u32((const uint32_t *)(s + p), v_s0, 1);
v_s1 = vld1_lane_u32((const uint32_t *)(s + 2 * p), v_s1, 0);
v_s1 = vld1_lane_u32((const uint32_t *)(s + 3 * p), v_s1, 1);
v_sum = vaddl_u8(vreinterpret_u8_u32(v_s0), vreinterpret_u8_u32(v_s1));
return (horizontal_add_u16x8(v_sum) + 8) >> 4;
}
unsigned int aom_avg_8x8_neon(const uint8_t *s, int p) {
uint8x8_t v_s0 = vld1_u8(s);
const uint8x8_t v_s1 = vld1_u8(s + p);
uint16x8_t v_sum = vaddl_u8(v_s0, v_s1);
v_s0 = vld1_u8(s + 2 * p);
v_sum = vaddw_u8(v_sum, v_s0);
v_s0 = vld1_u8(s + 3 * p);
v_sum = vaddw_u8(v_sum, v_s0);
v_s0 = vld1_u8(s + 4 * p);
v_sum = vaddw_u8(v_sum, v_s0);
v_s0 = vld1_u8(s + 5 * p);
v_sum = vaddw_u8(v_sum, v_s0);
v_s0 = vld1_u8(s + 6 * p);
v_sum = vaddw_u8(v_sum, v_s0);
v_s0 = vld1_u8(s + 7 * p);
v_sum = vaddw_u8(v_sum, v_s0);
return (horizontal_add_u16x8(v_sum) + 32) >> 6;
}
// coeff: 16 bits, dynamic range [-32640, 32640]. // coeff: 16 bits, dynamic range [-32640, 32640].
// length: value range {16, 64, 256, 1024}. // length: value range {16, 64, 256, 1024}.
int aom_satd_neon(const int16_t *coeff, int length) { int aom_satd_neon(const int16_t *coeff, int length) {

240
third_party/aom/aom_dsp/arm/bilinear_filter_media.asm vendored
View file

@ -1,240 +0,0 @@
;
; Copyright (c) 2016, Alliance for Open Media. All rights reserved
;
; This source code is subject to the terms of the BSD 2 Clause License and
; the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
; was not distributed with this source code in the LICENSE file, you can
; obtain it at www.aomedia.org/license/software. If the Alliance for Open
; Media Patent License 1.0 was not distributed with this source code in the
; PATENTS file, you can obtain it at www.aomedia.org/license/patent.
;
;
EXPORT |aom_filter_block2d_bil_first_pass_media|
EXPORT |aom_filter_block2d_bil_second_pass_media|
AREA |.text|, CODE, READONLY ; name this block of code
;-------------------------------------
; r0 unsigned char *src_ptr,
; r1 unsigned short *dst_ptr,
; r2 unsigned int src_pitch,
; r3 unsigned int height,
; stack unsigned int width,
; stack const short *aom_filter
;-------------------------------------
; The output is transposed stroed in output array to make it easy for second pass filtering.
|aom_filter_block2d_bil_first_pass_media| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #40] ; aom_filter address
ldr r4, [sp, #36] ; width
mov r12, r3 ; outer-loop counter
add r7, r2, r4 ; preload next row
pld [r0, r7]
sub r2, r2, r4 ; src increment for height loop
ldr r5, [r11] ; load up filter coefficients
mov r3, r3, lsl #1 ; height*2
add r3, r3, #2 ; plus 2 to make output buffer 4-bit aligned since height is actually (height+1)
mov r11, r1 ; save dst_ptr for each row
cmp r5, #128 ; if filter coef = 128, then skip the filter
beq bil_null_1st_filter
|bil_height_loop_1st_v6|
ldrb r6, [r0] ; load source data
ldrb r7, [r0, #1]
ldrb r8, [r0, #2]
mov lr, r4, lsr #2 ; 4-in-parellel loop counter
|bil_width_loop_1st_v6|
ldrb r9, [r0, #3]
ldrb r10, [r0, #4]
pkhbt r6, r6, r7, lsl #16 ; src[1] | src[0]
pkhbt r7, r7, r8, lsl #16 ; src[2] | src[1]
smuad r6, r6, r5 ; apply the filter
pkhbt r8, r8, r9, lsl #16 ; src[3] | src[2]
smuad r7, r7, r5
pkhbt r9, r9, r10, lsl #16 ; src[4] | src[3]
smuad r8, r8, r5
smuad r9, r9, r5
add r0, r0, #4
subs lr, lr, #1
add r6, r6, #0x40 ; round_shift_and_clamp
add r7, r7, #0x40
usat r6, #16, r6, asr #7
usat r7, #16, r7, asr #7
strh r6, [r1], r3 ; result is transposed and stored
add r8, r8, #0x40 ; round_shift_and_clamp
strh r7, [r1], r3
add r9, r9, #0x40
usat r8, #16, r8, asr #7
usat r9, #16, r9, asr #7
strh r8, [r1], r3 ; result is transposed and stored
ldrneb r6, [r0] ; load source data
strh r9, [r1], r3
ldrneb r7, [r0, #1]
ldrneb r8, [r0, #2]
bne bil_width_loop_1st_v6
add r0, r0, r2 ; move to next input row
subs r12, r12, #1
add r9, r2, r4, lsl #1 ; adding back block width
pld [r0, r9] ; preload next row
add r11, r11, #2 ; move over to next column
mov r1, r11
bne bil_height_loop_1st_v6
ldmia sp!, {r4 - r11, pc}
|bil_null_1st_filter|
|bil_height_loop_null_1st|
mov lr, r4, lsr #2 ; loop counter
|bil_width_loop_null_1st|
ldrb r6, [r0] ; load data
ldrb r7, [r0, #1]
ldrb r8, [r0, #2]
ldrb r9, [r0, #3]
strh r6, [r1], r3 ; store it to immediate buffer
add r0, r0, #4
strh r7, [r1], r3
subs lr, lr, #1
strh r8, [r1], r3
strh r9, [r1], r3
bne bil_width_loop_null_1st
subs r12, r12, #1
add r0, r0, r2 ; move to next input line
add r11, r11, #2 ; move over to next column
mov r1, r11
bne bil_height_loop_null_1st
ldmia sp!, {r4 - r11, pc}
ENDP ; |aom_filter_block2d_bil_first_pass_media|
;---------------------------------
; r0 unsigned short *src_ptr,
; r1 unsigned char *dst_ptr,
; r2 int dst_pitch,
; r3 unsigned int height,
; stack unsigned int width,
; stack const short *aom_filter
;---------------------------------
|aom_filter_block2d_bil_second_pass_media| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #40] ; aom_filter address
ldr r4, [sp, #36] ; width
ldr r5, [r11] ; load up filter coefficients
mov r12, r4 ; outer-loop counter = width, since we work on transposed data matrix
mov r11, r1
cmp r5, #128 ; if filter coef = 128, then skip the filter
beq bil_null_2nd_filter
|bil_height_loop_2nd|
ldr r6, [r0] ; load the data
ldr r8, [r0, #4]
ldrh r10, [r0, #8]
mov lr, r3, lsr #2 ; loop counter
|bil_width_loop_2nd|
pkhtb r7, r6, r8 ; src[1] | src[2]
pkhtb r9, r8, r10 ; src[3] | src[4]
smuad r6, r6, r5 ; apply filter
smuad r8, r8, r5 ; apply filter
subs lr, lr, #1
smuadx r7, r7, r5 ; apply filter
smuadx r9, r9, r5 ; apply filter
add r0, r0, #8
add r6, r6, #0x40 ; round_shift_and_clamp
add r7, r7, #0x40
usat r6, #8, r6, asr #7
usat r7, #8, r7, asr #7
strb r6, [r1], r2 ; the result is transposed back and stored
add r8, r8, #0x40 ; round_shift_and_clamp
strb r7, [r1], r2
add r9, r9, #0x40
usat r8, #8, r8, asr #7
usat r9, #8, r9, asr #7
strb r8, [r1], r2 ; the result is transposed back and stored
ldrne r6, [r0] ; load data
strb r9, [r1], r2
ldrne r8, [r0, #4]
ldrneh r10, [r0, #8]
bne bil_width_loop_2nd
subs r12, r12, #1
add r0, r0, #4 ; update src for next row
add r11, r11, #1
mov r1, r11
bne bil_height_loop_2nd
ldmia sp!, {r4 - r11, pc}
|bil_null_2nd_filter|
|bil_height_loop_null_2nd|
mov lr, r3, lsr #2
|bil_width_loop_null_2nd|
ldr r6, [r0], #4 ; load data
subs lr, lr, #1
ldr r8, [r0], #4
strb r6, [r1], r2 ; store data
mov r7, r6, lsr #16
strb r7, [r1], r2
mov r9, r8, lsr #16
strb r8, [r1], r2
strb r9, [r1], r2
bne bil_width_loop_null_2nd
subs r12, r12, #1
add r0, r0, #4
add r11, r11, #1
mov r1, r11
bne bil_height_loop_null_2nd
ldmia sp!, {r4 - r11, pc}
ENDP ; |aom_filter_block2d_second_pass_media|
END

98
third_party/aom/aom_dsp/arm/sad_media.asm vendored
View file

@ -1,98 +0,0 @@
;
; Copyright (c) 2016, Alliance for Open Media. All rights reserved
;
; This source code is subject to the terms of the BSD 2 Clause License and
; the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
; was not distributed with this source code in the LICENSE file, you can
; obtain it at www.aomedia.org/license/software. If the Alliance for Open
; Media Patent License 1.0 was not distributed with this source code in the
; PATENTS file, you can obtain it at www.aomedia.org/license/patent.
;
;
EXPORT |aom_sad16x16_media|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; r0 const unsigned char *src_ptr
; r1 int src_stride
; r2 const unsigned char *ref_ptr
; r3 int ref_stride
|aom_sad16x16_media| PROC
stmfd sp!, {r4-r12, lr}
pld [r0, r1, lsl #0]
pld [r2, r3, lsl #0]
pld [r0, r1, lsl #1]
pld [r2, r3, lsl #1]
mov r4, #0 ; sad = 0;
mov r5, #8 ; loop count
loop
; 1st row
ldr r6, [r0, #0x0] ; load 4 src pixels (1A)
ldr r8, [r2, #0x0] ; load 4 ref pixels (1A)
ldr r7, [r0, #0x4] ; load 4 src pixels (1A)
ldr r9, [r2, #0x4] ; load 4 ref pixels (1A)
ldr r10, [r0, #0x8] ; load 4 src pixels (1B)
ldr r11, [r0, #0xC] ; load 4 src pixels (1B)
usada8 r4, r8, r6, r4 ; calculate sad for 4 pixels
usad8 r8, r7, r9 ; calculate sad for 4 pixels
ldr r12, [r2, #0x8] ; load 4 ref pixels (1B)
ldr lr, [r2, #0xC] ; load 4 ref pixels (1B)
add r0, r0, r1 ; set src pointer to next row
add r2, r2, r3 ; set dst pointer to next row
pld [r0, r1, lsl #1]
pld [r2, r3, lsl #1]
usada8 r4, r10, r12, r4 ; calculate sad for 4 pixels
usada8 r8, r11, lr, r8 ; calculate sad for 4 pixels
ldr r6, [r0, #0x0] ; load 4 src pixels (2A)
ldr r7, [r0, #0x4] ; load 4 src pixels (2A)
add r4, r4, r8 ; add partial sad values
; 2nd row
ldr r8, [r2, #0x0] ; load 4 ref pixels (2A)
ldr r9, [r2, #0x4] ; load 4 ref pixels (2A)
ldr r10, [r0, #0x8] ; load 4 src pixels (2B)
ldr r11, [r0, #0xC] ; load 4 src pixels (2B)
usada8 r4, r6, r8, r4 ; calculate sad for 4 pixels
usad8 r8, r7, r9 ; calculate sad for 4 pixels
ldr r12, [r2, #0x8] ; load 4 ref pixels (2B)
ldr lr, [r2, #0xC] ; load 4 ref pixels (2B)
add r0, r0, r1 ; set src pointer to next row
add r2, r2, r3 ; set dst pointer to next row
usada8 r4, r10, r12, r4 ; calculate sad for 4 pixels
usada8 r8, r11, lr, r8 ; calculate sad for 4 pixels
pld [r0, r1, lsl #1]
pld [r2, r3, lsl #1]
subs r5, r5, #1 ; decrement loop counter
add r4, r4, r8 ; add partial sad values
bne loop
mov r0, r4 ; return sad
ldmfd sp!, {r4-r12, pc}
ENDP
END

81
third_party/aom/aom_dsp/arm/subpel_variance_media.c vendored
View file

@ -1,81 +0,0 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include "./aom_config.h"
#include "./aom_dsp_rtcd.h"
#include "aom/aom_integer.h"
#include "aom_ports/mem.h"
#if HAVE_MEDIA
static const int16_t bilinear_filters_media[8][2] = { { 128, 0 }, { 112, 16 },
{ 96, 32 }, { 80, 48 },
{ 64, 64 }, { 48, 80 },
{ 32, 96 }, { 16, 112 } };
extern void aom_filter_block2d_bil_first_pass_media(
const uint8_t *src_ptr, uint16_t *dst_ptr, uint32_t src_pitch,
uint32_t height, uint32_t width, const int16_t *filter);
extern void aom_filter_block2d_bil_second_pass_media(
const uint16_t *src_ptr, uint8_t *dst_ptr, int32_t src_pitch,
uint32_t height, uint32_t width, const int16_t *filter);
unsigned int aom_sub_pixel_variance8x8_media(
const uint8_t *src_ptr, int src_pixels_per_line, int xoffset, int yoffset,
const uint8_t *dst_ptr, int dst_pixels_per_line, unsigned int *sse) {
uint16_t first_pass[10 * 8];
uint8_t second_pass[8 * 8];
const int16_t *HFilter, *VFilter;
HFilter = bilinear_filters_media[xoffset];
VFilter = bilinear_filters_media[yoffset];
aom_filter_block2d_bil_first_pass_media(src_ptr, first_pass,
src_pixels_per_line, 9, 8, HFilter);
aom_filter_block2d_bil_second_pass_media(first_pass, second_pass, 8, 8, 8,
VFilter);
return aom_variance8x8_media(second_pass, 8, dst_ptr, dst_pixels_per_line,
sse);
}
unsigned int aom_sub_pixel_variance16x16_media(
const uint8_t *src_ptr, int src_pixels_per_line, int xoffset, int yoffset,
const uint8_t *dst_ptr, int dst_pixels_per_line, unsigned int *sse) {
uint16_t first_pass[36 * 16];
uint8_t second_pass[20 * 16];
const int16_t *HFilter, *VFilter;
unsigned int var;
if (xoffset == 4 && yoffset == 0) {
var = aom_variance_halfpixvar16x16_h_media(
src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
} else if (xoffset == 0 && yoffset == 4) {
var = aom_variance_halfpixvar16x16_v_media(
src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
} else if (xoffset == 4 && yoffset == 4) {
var = aom_variance_halfpixvar16x16_hv_media(
src_ptr, src_pixels_per_line, dst_ptr, dst_pixels_per_line, sse);
} else {
HFilter = bilinear_filters_media[xoffset];
VFilter = bilinear_filters_media[yoffset];
aom_filter_block2d_bil_first_pass_media(
src_ptr, first_pass, src_pixels_per_line, 17, 16, HFilter);
aom_filter_block2d_bil_second_pass_media(first_pass, second_pass, 16, 16,
16, VFilter);
var = aom_variance16x16_media(second_pass, 16, dst_ptr, dst_pixels_per_line,
sse);
}
return var;
}
#endif // HAVE_MEDIA

185
third_party/aom/aom_dsp/arm/variance_halfpixvar16x16_h_media.asm vendored
View file

@ -1,185 +0,0 @@
;
; Copyright (c) 2016, Alliance for Open Media. All rights reserved
;
; This source code is subject to the terms of the BSD 2 Clause License and
; the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
; was not distributed with this source code in the LICENSE file, you can
; obtain it at www.aomedia.org/license/software. If the Alliance for Open
; Media Patent License 1.0 was not distributed with this source code in the
; PATENTS file, you can obtain it at www.aomedia.org/license/patent.
;
;
EXPORT |aom_variance_halfpixvar16x16_h_media|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; r0 unsigned char *src_ptr
; r1 int source_stride
; r2 unsigned char *ref_ptr
; r3 int recon_stride
; stack unsigned int *sse
|aom_variance_halfpixvar16x16_h_media| PROC
stmfd sp!, {r4-r12, lr}
pld [r0, r1, lsl #0]
pld [r2, r3, lsl #0]
mov r8, #0 ; initialize sum = 0
ldr r10, c80808080
mov r11, #0 ; initialize sse = 0
mov r12, #16 ; set loop counter to 16 (=block height)
mov lr, #0 ; constant zero
loop
; 1st 4 pixels
ldr r4, [r0, #0] ; load 4 src pixels
ldr r6, [r0, #1] ; load 4 src pixels with 1 byte offset
ldr r5, [r2, #0] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
usub8 r6, r4, r5 ; calculate difference
pld [r0, r1, lsl #1]
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
pld [r2, r3, lsl #1]
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
adds r8, r8, r4 ; add positive differences to sum
subs r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 2nd 4 pixels
ldr r4, [r0, #4] ; load 4 src pixels
ldr r6, [r0, #5] ; load 4 src pixels with 1 byte offset
ldr r5, [r2, #4] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 3rd 4 pixels
ldr r4, [r0, #8] ; load 4 src pixels
ldr r6, [r0, #9] ; load 4 src pixels with 1 byte offset
ldr r5, [r2, #8] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 4th 4 pixels
ldr r4, [r0, #12] ; load 4 src pixels
ldr r6, [r0, #13] ; load 4 src pixels with 1 byte offset
ldr r5, [r2, #12] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
add r0, r0, r1 ; set src_ptr to next row
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
add r2, r2, r3 ; set dst_ptr to next row
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
subs r12, r12, #1
bne loop
; return stuff
ldr r6, [sp, #40] ; get address of sse
mul r0, r8, r8 ; sum * sum
str r11, [r6] ; store sse
sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
ldmfd sp!, {r4-r12, pc}
ENDP
c80808080
DCD 0x80808080
END

225
third_party/aom/aom_dsp/arm/variance_halfpixvar16x16_hv_media.asm vendored
View file

@ -1,225 +0,0 @@
;
; Copyright (c) 2016, Alliance for Open Media. All rights reserved
;
; This source code is subject to the terms of the BSD 2 Clause License and
; the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
; was not distributed with this source code in the LICENSE file, you can
; obtain it at www.aomedia.org/license/software. If the Alliance for Open
; Media Patent License 1.0 was not distributed with this source code in the
; PATENTS file, you can obtain it at www.aomedia.org/license/patent.
;
;
EXPORT |aom_variance_halfpixvar16x16_hv_media|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; r0 unsigned char *src_ptr
; r1 int source_stride
; r2 unsigned char *ref_ptr
; r3 int recon_stride
; stack unsigned int *sse
|aom_variance_halfpixvar16x16_hv_media| PROC
stmfd sp!, {r4-r12, lr}
pld [r0, r1, lsl #0]
pld [r2, r3, lsl #0]
mov r8, #0 ; initialize sum = 0
ldr r10, c80808080
mov r11, #0 ; initialize sse = 0
mov r12, #16 ; set loop counter to 16 (=block height)
mov lr, #0 ; constant zero
loop
add r9, r0, r1 ; pointer to pixels on the next row
; 1st 4 pixels
ldr r4, [r0, #0] ; load source pixels a, row N
ldr r6, [r0, #1] ; load source pixels b, row N
ldr r5, [r9, #0] ; load source pixels c, row N+1
ldr r7, [r9, #1] ; load source pixels d, row N+1
; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
mvn r7, r7
uhsub8 r5, r5, r7
eor r5, r5, r10
; z = (x + y + 1) >> 1, interpolate half pixel values vertically
mvn r5, r5
uhsub8 r4, r4, r5
ldr r5, [r2, #0] ; load 4 ref pixels
eor r4, r4, r10
usub8 r6, r4, r5 ; calculate difference
pld [r0, r1, lsl #1]
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
pld [r2, r3, lsl #1]
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
adds r8, r8, r4 ; add positive differences to sum
subs r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 2nd 4 pixels
ldr r4, [r0, #4] ; load source pixels a, row N
ldr r6, [r0, #5] ; load source pixels b, row N
ldr r5, [r9, #4] ; load source pixels c, row N+1
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
ldr r7, [r9, #5] ; load source pixels d, row N+1
; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
mvn r7, r7
uhsub8 r5, r5, r7
eor r5, r5, r10
; z = (x + y + 1) >> 1, interpolate half pixel values vertically
mvn r5, r5
uhsub8 r4, r4, r5
ldr r5, [r2, #4] ; load 4 ref pixels
eor r4, r4, r10
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 3rd 4 pixels
ldr r4, [r0, #8] ; load source pixels a, row N
ldr r6, [r0, #9] ; load source pixels b, row N
ldr r5, [r9, #8] ; load source pixels c, row N+1
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
ldr r7, [r9, #9] ; load source pixels d, row N+1
; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
mvn r7, r7
uhsub8 r5, r5, r7
eor r5, r5, r10
; z = (x + y + 1) >> 1, interpolate half pixel values vertically
mvn r5, r5
uhsub8 r4, r4, r5
ldr r5, [r2, #8] ; load 4 ref pixels
eor r4, r4, r10
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 4th 4 pixels
ldr r4, [r0, #12] ; load source pixels a, row N
ldr r6, [r0, #13] ; load source pixels b, row N
ldr r5, [r9, #12] ; load source pixels c, row N+1
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
ldr r7, [r9, #13] ; load source pixels d, row N+1
; x = (a + b + 1) >> 1, interpolate pixels horizontally on row N
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
; y = (c + d + 1) >> 1, interpolate pixels horizontally on row N+1
mvn r7, r7
uhsub8 r5, r5, r7
eor r5, r5, r10
; z = (x + y + 1) >> 1, interpolate half pixel values vertically
mvn r5, r5
uhsub8 r4, r4, r5
ldr r5, [r2, #12] ; load 4 ref pixels
eor r4, r4, r10
usub8 r6, r4, r5 ; calculate difference
add r0, r0, r1 ; set src_ptr to next row
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
add r2, r2, r3 ; set dst_ptr to next row
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
subs r12, r12, #1
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
bne loop
; return stuff
ldr r6, [sp, #40] ; get address of sse
mul r0, r8, r8 ; sum * sum
str r11, [r6] ; store sse
sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
ldmfd sp!, {r4-r12, pc}
ENDP
c80808080
DCD 0x80808080
END

187
third_party/aom/aom_dsp/arm/variance_halfpixvar16x16_v_media.asm vendored
View file

@ -1,187 +0,0 @@
;
; Copyright (c) 2016, Alliance for Open Media. All rights reserved
;
; This source code is subject to the terms of the BSD 2 Clause License and
; the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
; was not distributed with this source code in the LICENSE file, you can
; obtain it at www.aomedia.org/license/software. If the Alliance for Open
; Media Patent License 1.0 was not distributed with this source code in the
; PATENTS file, you can obtain it at www.aomedia.org/license/patent.
;
;
EXPORT |aom_variance_halfpixvar16x16_v_media|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; r0 unsigned char *src_ptr
; r1 int source_stride
; r2 unsigned char *ref_ptr
; r3 int recon_stride
; stack unsigned int *sse
|aom_variance_halfpixvar16x16_v_media| PROC
stmfd sp!, {r4-r12, lr}
pld [r0, r1, lsl #0]
pld [r2, r3, lsl #0]
mov r8, #0 ; initialize sum = 0
ldr r10, c80808080
mov r11, #0 ; initialize sse = 0
mov r12, #16 ; set loop counter to 16 (=block height)
mov lr, #0 ; constant zero
loop
add r9, r0, r1 ; set src pointer to next row
; 1st 4 pixels
ldr r4, [r0, #0] ; load 4 src pixels
ldr r6, [r9, #0] ; load 4 src pixels from next row
ldr r5, [r2, #0] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
usub8 r6, r4, r5 ; calculate difference
pld [r0, r1, lsl #1]
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
pld [r2, r3, lsl #1]
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
adds r8, r8, r4 ; add positive differences to sum
subs r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 2nd 4 pixels
ldr r4, [r0, #4] ; load 4 src pixels
ldr r6, [r9, #4] ; load 4 src pixels from next row
ldr r5, [r2, #4] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 3rd 4 pixels
ldr r4, [r0, #8] ; load 4 src pixels
ldr r6, [r9, #8] ; load 4 src pixels from next row
ldr r5, [r2, #8] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 4th 4 pixels
ldr r4, [r0, #12] ; load 4 src pixels
ldr r6, [r9, #12] ; load 4 src pixels from next row
ldr r5, [r2, #12] ; load 4 ref pixels
; bilinear interpolation
mvn r6, r6
uhsub8 r4, r4, r6
eor r4, r4, r10
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
add r0, r0, r1 ; set src_ptr to next row
sel r7, r6, lr ; select bytes with positive difference
usub8 r6, r5, r4 ; calculate difference with reversed operands
add r2, r2, r3 ; set dst_ptr to next row
sel r6, r6, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r7, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
smlad r11, r7, r7, r11 ; dual signed multiply, add and accumulate (2)
subs r12, r12, #1
bne loop
; return stuff
ldr r6, [sp, #40] ; get address of sse
mul r0, r8, r8 ; sum * sum
str r11, [r6] ; store sse
sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
ldmfd sp!, {r4-r12, pc}
ENDP
c80808080
DCD 0x80808080
END

361
third_party/aom/aom_dsp/arm/variance_media.asm vendored
View file

@ -1,361 +0,0 @@
;
; Copyright (c) 2016, Alliance for Open Media. All rights reserved
;
; This source code is subject to the terms of the BSD 2 Clause License and
; the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
; was not distributed with this source code in the LICENSE file, you can
; obtain it at www.aomedia.org/license/software. If the Alliance for Open
; Media Patent License 1.0 was not distributed with this source code in the
; PATENTS file, you can obtain it at www.aomedia.org/license/patent.
;
;
EXPORT |aom_variance16x16_media|
EXPORT |aom_variance8x8_media|
EXPORT |aom_mse16x16_media|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; r0 unsigned char *src_ptr
; r1 int source_stride
; r2 unsigned char *ref_ptr
; r3 int recon_stride
; stack unsigned int *sse
|aom_variance16x16_media| PROC
stmfd sp!, {r4-r12, lr}
pld [r0, r1, lsl #0]
pld [r2, r3, lsl #0]
mov r8, #0 ; initialize sum = 0
mov r11, #0 ; initialize sse = 0
mov r12, #16 ; set loop counter to 16 (=block height)
loop16x16
; 1st 4 pixels
ldr r4, [r0, #0] ; load 4 src pixels
ldr r5, [r2, #0] ; load 4 ref pixels
mov lr, #0 ; constant zero
usub8 r6, r4, r5 ; calculate difference
pld [r0, r1, lsl #1]
sel r7, r6, lr ; select bytes with positive difference
usub8 r9, r5, r4 ; calculate difference with reversed operands
pld [r2, r3, lsl #1]
sel r6, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
adds r8, r8, r4 ; add positive differences to sum
subs r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 2nd 4 pixels
ldr r4, [r0, #4] ; load 4 src pixels
ldr r5, [r2, #4] ; load 4 ref pixels
smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r9, r5, r4 ; calculate difference with reversed operands
sel r6, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 3rd 4 pixels
ldr r4, [r0, #8] ; load 4 src pixels
ldr r5, [r2, #8] ; load 4 ref pixels
smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
sel r7, r6, lr ; select bytes with positive difference
usub8 r9, r5, r4 ; calculate difference with reversed operands
sel r6, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
; 4th 4 pixels
ldr r4, [r0, #12] ; load 4 src pixels
ldr r5, [r2, #12] ; load 4 ref pixels
smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
usub8 r6, r4, r5 ; calculate difference
add r0, r0, r1 ; set src_ptr to next row
sel r7, r6, lr ; select bytes with positive difference
usub8 r9, r5, r4 ; calculate difference with reversed operands
add r2, r2, r3 ; set dst_ptr to next row
sel r6, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r4, r7, lr ; calculate sum of positive differences
usad8 r5, r6, lr ; calculate sum of negative differences
orr r6, r6, r7 ; differences of all 4 pixels
; calculate total sum
add r8, r8, r4 ; add positive differences to sum
sub r8, r8, r5 ; subtract negative differences from sum
; calculate sse
uxtb16 r5, r6 ; byte (two pixels) to halfwords
uxtb16 r10, r6, ror #8 ; another two pixels to halfwords
smlad r11, r5, r5, r11 ; dual signed multiply, add and accumulate (1)
smlad r11, r10, r10, r11 ; dual signed multiply, add and accumulate (2)
subs r12, r12, #1
bne loop16x16
; return stuff
ldr r6, [sp, #40] ; get address of sse
mul r0, r8, r8 ; sum * sum
str r11, [r6] ; store sse
sub r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
ldmfd sp!, {r4-r12, pc}
ENDP
; r0 unsigned char *src_ptr
; r1 int source_stride
; r2 unsigned char *ref_ptr
; r3 int recon_stride
; stack unsigned int *sse
|aom_variance8x8_media| PROC
push {r4-r10, lr}
pld [r0, r1, lsl #0]
pld [r2, r3, lsl #0]
mov r12, #8 ; set loop counter to 8 (=block height)
mov r4, #0 ; initialize sum = 0
mov r5, #0 ; initialize sse = 0
loop8x8
; 1st 4 pixels
ldr r6, [r0, #0x0] ; load 4 src pixels
ldr r7, [r2, #0x0] ; load 4 ref pixels
mov lr, #0 ; constant zero
usub8 r8, r6, r7 ; calculate difference
pld [r0, r1, lsl #1]
sel r10, r8, lr ; select bytes with positive difference
usub8 r9, r7, r6 ; calculate difference with reversed operands
pld [r2, r3, lsl #1]
sel r8, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r6, r10, lr ; calculate sum of positive differences
usad8 r7, r8, lr ; calculate sum of negative differences
orr r8, r8, r10 ; differences of all 4 pixels
; calculate total sum
add r4, r4, r6 ; add positive differences to sum
sub r4, r4, r7 ; subtract negative differences from sum
; calculate sse
uxtb16 r7, r8 ; byte (two pixels) to halfwords
uxtb16 r10, r8, ror #8 ; another two pixels to halfwords
smlad r5, r7, r7, r5 ; dual signed multiply, add and accumulate (1)
; 2nd 4 pixels
ldr r6, [r0, #0x4] ; load 4 src pixels
ldr r7, [r2, #0x4] ; load 4 ref pixels
smlad r5, r10, r10, r5 ; dual signed multiply, add and accumulate (2)
usub8 r8, r6, r7 ; calculate difference
add r0, r0, r1 ; set src_ptr to next row
sel r10, r8, lr ; select bytes with positive difference
usub8 r9, r7, r6 ; calculate difference with reversed operands
add r2, r2, r3 ; set dst_ptr to next row
sel r8, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r6, r10, lr ; calculate sum of positive differences
usad8 r7, r8, lr ; calculate sum of negative differences
orr r8, r8, r10 ; differences of all 4 pixels
; calculate total sum
add r4, r4, r6 ; add positive differences to sum
sub r4, r4, r7 ; subtract negative differences from sum
; calculate sse
uxtb16 r7, r8 ; byte (two pixels) to halfwords
uxtb16 r10, r8, ror #8 ; another two pixels to halfwords
smlad r5, r7, r7, r5 ; dual signed multiply, add and accumulate (1)
subs r12, r12, #1 ; next row
smlad r5, r10, r10, r5 ; dual signed multiply, add and accumulate (2)
bne loop8x8
; return stuff
ldr r8, [sp, #32] ; get address of sse
mul r1, r4, r4 ; sum * sum
str r5, [r8] ; store sse
sub r0, r5, r1, ASR #6 ; return (sse - ((sum * sum) >> 6))
pop {r4-r10, pc}
ENDP
; r0 unsigned char *src_ptr
; r1 int source_stride
; r2 unsigned char *ref_ptr
; r3 int recon_stride
; stack unsigned int *sse
;
;note: Based on aom_variance16x16_media. In this function, sum is never used.
; So, we can remove this part of calculation.
|aom_mse16x16_media| PROC
push {r4-r9, lr}
pld [r0, r1, lsl #0]
pld [r2, r3, lsl #0]
mov r12, #16 ; set loop counter to 16 (=block height)
mov r4, #0 ; initialize sse = 0
loopmse
; 1st 4 pixels
ldr r5, [r0, #0x0] ; load 4 src pixels
ldr r6, [r2, #0x0] ; load 4 ref pixels
mov lr, #0 ; constant zero
usub8 r8, r5, r6 ; calculate difference
pld [r0, r1, lsl #1]
sel r7, r8, lr ; select bytes with positive difference
usub8 r9, r6, r5 ; calculate difference with reversed operands
pld [r2, r3, lsl #1]
sel r8, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r5, r7, lr ; calculate sum of positive differences
usad8 r6, r8, lr ; calculate sum of negative differences
orr r8, r8, r7 ; differences of all 4 pixels
ldr r5, [r0, #0x4] ; load 4 src pixels
; calculate sse
uxtb16 r6, r8 ; byte (two pixels) to halfwords
uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
; 2nd 4 pixels
ldr r6, [r2, #0x4] ; load 4 ref pixels
smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
usub8 r8, r5, r6 ; calculate difference
sel r7, r8, lr ; select bytes with positive difference
usub8 r9, r6, r5 ; calculate difference with reversed operands
sel r8, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r5, r7, lr ; calculate sum of positive differences
usad8 r6, r8, lr ; calculate sum of negative differences
orr r8, r8, r7 ; differences of all 4 pixels
ldr r5, [r0, #0x8] ; load 4 src pixels
; calculate sse
uxtb16 r6, r8 ; byte (two pixels) to halfwords
uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
; 3rd 4 pixels
ldr r6, [r2, #0x8] ; load 4 ref pixels
smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
usub8 r8, r5, r6 ; calculate difference
sel r7, r8, lr ; select bytes with positive difference
usub8 r9, r6, r5 ; calculate difference with reversed operands
sel r8, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r5, r7, lr ; calculate sum of positive differences
usad8 r6, r8, lr ; calculate sum of negative differences
orr r8, r8, r7 ; differences of all 4 pixels
ldr r5, [r0, #0xc] ; load 4 src pixels
; calculate sse
uxtb16 r6, r8 ; byte (two pixels) to halfwords
uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
; 4th 4 pixels
ldr r6, [r2, #0xc] ; load 4 ref pixels
smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
usub8 r8, r5, r6 ; calculate difference
add r0, r0, r1 ; set src_ptr to next row
sel r7, r8, lr ; select bytes with positive difference
usub8 r9, r6, r5 ; calculate difference with reversed operands
add r2, r2, r3 ; set dst_ptr to next row
sel r8, r9, lr ; select bytes with negative difference
; calculate partial sums
usad8 r5, r7, lr ; calculate sum of positive differences
usad8 r6, r8, lr ; calculate sum of negative differences
orr r8, r8, r7 ; differences of all 4 pixels
subs r12, r12, #1 ; next row
; calculate sse
uxtb16 r6, r8 ; byte (two pixels) to halfwords
uxtb16 r7, r8, ror #8 ; another two pixels to halfwords
smlad r4, r6, r6, r4 ; dual signed multiply, add and accumulate (1)
smlad r4, r7, r7, r4 ; dual signed multiply, add and accumulate (2)
bne loopmse
; return stuff
ldr r1, [sp, #28] ; get address of sse
mov r0, r4 ; return sse
str r4, [r1] ; store sse
pop {r4-r9, pc}
ENDP
END

42
third_party/aom/aom_dsp/avg.c vendored
View file

@ -13,26 +13,6 @@
#include "./aom_dsp_rtcd.h" #include "./aom_dsp_rtcd.h"
#include "aom_ports/mem.h" #include "aom_ports/mem.h"
unsigned int aom_avg_8x8_c(const uint8_t *src, int stride) {
int i, j;
int sum = 0;
for (i = 0; i < 8; ++i, src += stride)
for (j = 0; j < 8; sum += src[j], ++j) {
}
return ROUND_POWER_OF_TWO(sum, 6);
}
unsigned int aom_avg_4x4_c(const uint8_t *src, int stride) {
int i, j;
int sum = 0;
for (i = 0; i < 4; ++i, src += stride)
for (j = 0; j < 4; sum += src[j], ++j) {
}
return ROUND_POWER_OF_TWO(sum, 4);
}
// src_diff: first pass, 9 bit, dynamic range [-255, 255] // src_diff: first pass, 9 bit, dynamic range [-255, 255]
// second pass, 12 bit, dynamic range [-2040, 2040] // second pass, 12 bit, dynamic range [-2040, 2040]
static void hadamard_col8(const int16_t *src_diff, int src_stride, static void hadamard_col8(const int16_t *src_diff, int src_stride,
@ -192,28 +172,6 @@ void aom_minmax_8x8_c(const uint8_t *src, int src_stride, const uint8_t *ref,
} }
#if CONFIG_HIGHBITDEPTH #if CONFIG_HIGHBITDEPTH
unsigned int aom_highbd_avg_8x8_c(const uint8_t *src, int stride) {
int i, j;
int sum = 0;
const uint16_t *s = CONVERT_TO_SHORTPTR(src);
for (i = 0; i < 8; ++i, s += stride)
for (j = 0; j < 8; sum += s[j], ++j) {
}
return ROUND_POWER_OF_TWO(sum, 6);
}
unsigned int aom_highbd_avg_4x4_c(const uint8_t *src, int stride) {
int i, j;
int sum = 0;
const uint16_t *s = CONVERT_TO_SHORTPTR(src);
for (i = 0; i < 4; ++i, s += stride)
for (j = 0; j < 4; sum += s[j], ++j) {
}
return ROUND_POWER_OF_TWO(sum, 4);
}
void aom_highbd_minmax_8x8_c(const uint8_t *s8, int p, const uint8_t *d8, void aom_highbd_minmax_8x8_c(const uint8_t *s8, int p, const uint8_t *d8,
int dp, int *min, int *max) { int dp, int *min, int *max) {
int i, j; int i, j;

53
third_party/aom/aom_dsp/binary_codes_reader.c vendored
View file

@ -9,7 +9,7 @@
* PATENTS file, you can obtain it at www.aomedia.org/license/patent. * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/ */
#include "aom_dsp/bitreader.h" #include "aom_dsp/binary_codes_reader.h"
#include "av1/common/common.h" #include "av1/common/common.h"
@ -33,26 +33,28 @@ static uint16_t inv_recenter_finite_nonneg(uint16_t n, uint16_t r, uint16_t v) {
} }
} }
int16_t aom_read_primitive_symmetric(aom_reader *r, unsigned int mag_bits) { int16_t aom_read_primitive_symmetric_(aom_reader *r,
if (aom_read_bit(r, NULL)) { unsigned int mag_bits ACCT_STR_PARAM) {
int s = aom_read_bit(r, NULL); if (aom_read_bit(r, ACCT_STR_NAME)) {
int16_t x = aom_read_literal(r, mag_bits, NULL) + 1; int s = aom_read_bit(r, ACCT_STR_NAME);
int16_t x = aom_read_literal(r, mag_bits, ACCT_STR_NAME) + 1;
return (s > 0 ? -x : x); return (s > 0 ? -x : x);
} else { } else {
return 0; return 0;
} }
} }
uint16_t aom_read_primitive_quniform(aom_reader *r, uint16_t n) { uint16_t aom_read_primitive_quniform_(aom_reader *r,
uint16_t n ACCT_STR_PARAM) {
if (n <= 1) return 0; if (n <= 1) return 0;
const int l = get_msb(n - 1) + 1; const int l = get_msb(n - 1) + 1;
const int m = (1 << l) - n; const int m = (1 << l) - n;
const int v = aom_read_literal(r, l - 1, NULL); const int v = aom_read_literal(r, l - 1, ACCT_STR_NAME);
return v < m ? v : (v << 1) - m + aom_read_bit(r, NULL); return v < m ? v : (v << 1) - m + aom_read_bit(r, ACCT_STR_NAME);
} }
uint16_t aom_read_primitive_refbilevel(aom_reader *r, uint16_t n, uint16_t p, uint16_t aom_read_primitive_refbilevel_(aom_reader *r, uint16_t n, uint16_t p,
uint16_t ref) { uint16_t ref ACCT_STR_PARAM) {
if (n <= 1) return 0; if (n <= 1) return 0;
assert(p > 0 && p <= n); assert(p > 0 && p <= n);
assert(ref < n); assert(ref < n);
@ -64,10 +66,10 @@ uint16_t aom_read_primitive_refbilevel(aom_reader *r, uint16_t n, uint16_t p,
lolimit = n - p; lolimit = n - p;
} }
int v; int v;
if (aom_read_bit(r, NULL)) { if (aom_read_bit(r, ACCT_STR_NAME)) {
v = aom_read_primitive_quniform(r, p) + lolimit; v = aom_read_primitive_quniform(r, p, ACCT_STR_NAME) + lolimit;
} else { } else {
v = aom_read_primitive_quniform(r, n - p); v = aom_read_primitive_quniform(r, n - p, ACCT_STR_NAME);
if (v >= lolimit) v += p; if (v >= lolimit) v += p;
} }
return v; return v;
@ -75,7 +77,8 @@ uint16_t aom_read_primitive_refbilevel(aom_reader *r, uint16_t n, uint16_t p,
// Decode finite subexponential code that for a symbol v in [0, n-1] with // Decode finite subexponential code that for a symbol v in [0, n-1] with
// parameter k // parameter k
uint16_t aom_read_primitive_subexpfin(aom_reader *r, uint16_t n, uint16_t k) { uint16_t aom_read_primitive_subexpfin_(aom_reader *r, uint16_t n,
uint16_t k ACCT_STR_PARAM) {
int i = 0; int i = 0;
int mk = 0; int mk = 0;
uint16_t v; uint16_t v;
@ -83,14 +86,14 @@ uint16_t aom_read_primitive_subexpfin(aom_reader *r, uint16_t n, uint16_t k) {
int b = (i ? k + i - 1 : k); int b = (i ? k + i - 1 : k);
int a = (1 << b); int a = (1 << b);
if (n <= mk + 3 * a) { if (n <= mk + 3 * a) {
v = aom_read_primitive_quniform(r, n - mk) + mk; v = aom_read_primitive_quniform(r, n - mk, ACCT_STR_NAME) + mk;
break; break;
} else { } else {
if (aom_read_bit(r, NULL)) { if (aom_read_bit(r, ACCT_STR_NAME)) {
i = i + 1; i = i + 1;
mk += a; mk += a;
} else { } else {
v = aom_read_literal(r, b, NULL) + mk; v = aom_read_literal(r, b, ACCT_STR_NAME) + mk;
break; break;
} }
} }
@ -101,17 +104,19 @@ uint16_t aom_read_primitive_subexpfin(aom_reader *r, uint16_t n, uint16_t k) {
// Decode finite subexponential code that for a symbol v in [0, n-1] with // Decode finite subexponential code that for a symbol v in [0, n-1] with
// parameter k // parameter k
// based on a reference ref also in [0, n-1]. // based on a reference ref also in [0, n-1].
uint16_t aom_read_primitive_refsubexpfin(aom_reader *r, uint16_t n, uint16_t k, uint16_t aom_read_primitive_refsubexpfin_(aom_reader *r, uint16_t n, uint16_t k,
uint16_t ref) { uint16_t ref ACCT_STR_PARAM) {
return inv_recenter_finite_nonneg(n, ref, return inv_recenter_finite_nonneg(
aom_read_primitive_subexpfin(r, n, k)); n, ref, aom_read_primitive_subexpfin(r, n, k, ACCT_STR_NAME));
} }
// Decode finite subexponential code that for a symbol v in [-(n-1), n-1] with // Decode finite subexponential code that for a symbol v in [-(n-1), n-1] with
// parameter k based on a reference ref also in [-(n-1), n-1]. // parameter k based on a reference ref also in [-(n-1), n-1].
int16_t aom_read_signed_primitive_refsubexpfin(aom_reader *r, uint16_t n, int16_t aom_read_signed_primitive_refsubexpfin_(aom_reader *r, uint16_t n,
uint16_t k, int16_t ref) { uint16_t k,
int16_t ref ACCT_STR_PARAM) {
ref += n - 1; ref += n - 1;
const uint16_t scaled_n = (n << 1) - 1; const uint16_t scaled_n = (n << 1) - 1;
return aom_read_primitive_refsubexpfin(r, scaled_n, k, ref) - n + 1; return aom_read_primitive_refsubexpfin(r, scaled_n, k, ref, ACCT_STR_NAME) -
n + 1;
} }

34
third_party/aom/aom_dsp/binary_codes_reader.h vendored
View file

@ -21,16 +21,32 @@ extern "C" {
#include "aom/aom_integer.h" #include "aom/aom_integer.h"
#include "aom_dsp/bitreader.h" #include "aom_dsp/bitreader.h"
int16_t aom_read_primitive_symmetric(aom_reader *r, unsigned int mag_bits); #define aom_read_primitive_symmetric(r, n, ACCT_STR_NAME) \
aom_read_primitive_symmetric_(r, n ACCT_STR_ARG(ACCT_STR_NAME))
#define aom_read_primitive_quniform(r, n, ACCT_STR_NAME) \
aom_read_primitive_quniform_(r, n ACCT_STR_ARG(ACCT_STR_NAME))
#define aom_read_primitive_refbilevel(r, n, p, ref, ACCT_STR_NAME) \
aom_read_primitive_refbilevel_(r, n, p, ref ACCT_STR_ARG(ACCT_STR_NAME))
#define aom_read_primitive_subexpfin(r, n, k, ACCT_STR_NAME) \
aom_read_primitive_subexpfin_(r, n, k ACCT_STR_ARG(ACCT_STR_NAME))
#define aom_read_primitive_refsubexpfin(r, n, k, ref, ACCT_STR_NAME) \
aom_read_primitive_refsubexpfin_(r, n, k, ref ACCT_STR_ARG(ACCT_STR_NAME))
#define aom_read_signed_primitive_refsubexpfin(r, n, k, ref, ACCT_STR_NAME) \
aom_read_signed_primitive_refsubexpfin_(r, n, k, \
ref ACCT_STR_ARG(ACCT_STR_NAME))
uint16_t aom_read_primitive_quniform(aom_reader *r, uint16_t n); int16_t aom_read_primitive_symmetric_(aom_reader *r,
uint16_t aom_read_primitive_refbilevel(aom_reader *r, uint16_t n, uint16_t p, unsigned int mag_bits ACCT_STR_PARAM);
uint16_t ref); uint16_t aom_read_primitive_quniform_(aom_reader *r, uint16_t n ACCT_STR_PARAM);
uint16_t aom_read_primitive_subexpfin(aom_reader *r, uint16_t n, uint16_t k); uint16_t aom_read_primitive_refbilevel_(aom_reader *r, uint16_t n, uint16_t p,
uint16_t aom_read_primitive_refsubexpfin(aom_reader *r, uint16_t n, uint16_t k, uint16_t ref ACCT_STR_PARAM);
uint16_t ref); uint16_t aom_read_primitive_subexpfin_(aom_reader *r, uint16_t n,
int16_t aom_read_signed_primitive_refsubexpfin(aom_reader *r, uint16_t n, uint16_t k ACCT_STR_PARAM);
uint16_t k, int16_t ref); uint16_t aom_read_primitive_refsubexpfin_(aom_reader *r, uint16_t n, uint16_t k,
uint16_t ref ACCT_STR_PARAM);
int16_t aom_read_signed_primitive_refsubexpfin_(aom_reader *r, uint16_t n,
uint16_t k,
int16_t ref ACCT_STR_PARAM);
#ifdef __cplusplus #ifdef __cplusplus
} // extern "C" } // extern "C"
#endif #endif

62
third_party/aom/aom_dsp/bitreader.h vendored
View file

@ -16,18 +16,13 @@
#include <limits.h> #include <limits.h>
#include "./aom_config.h" #include "./aom_config.h"
#if CONFIG_EC_ADAPT && !CONFIG_EC_MULTISYMBOL
#error "CONFIG_EC_ADAPT is enabled without enabling CONFIG_EC_MULTISYMBOL."
#endif
#include "aom/aomdx.h" #include "aom/aomdx.h"
#include "aom/aom_integer.h" #include "aom/aom_integer.h"
#if CONFIG_ANS #if CONFIG_ANS
#include "aom_dsp/ansreader.h" #include "aom_dsp/ansreader.h"
#elif CONFIG_DAALA_EC
#include "aom_dsp/daalaboolreader.h"
#else #else
#include "aom_dsp/dkboolreader.h" #include "aom_dsp/daalaboolreader.h"
#endif #endif
#include "aom_dsp/prob.h" #include "aom_dsp/prob.h"
#include "av1/common/odintrin.h" #include "av1/common/odintrin.h"
@ -61,26 +56,20 @@ extern "C" {
#if CONFIG_ANS #if CONFIG_ANS
typedef struct AnsDecoder aom_reader; typedef struct AnsDecoder aom_reader;
#elif CONFIG_DAALA_EC
typedef struct daala_reader aom_reader;
#else #else
typedef struct aom_dk_reader aom_reader; typedef struct daala_reader aom_reader;
#endif #endif
static INLINE int aom_reader_init(aom_reader *r, const uint8_t *buffer, static INLINE int aom_reader_init(aom_reader *r, const uint8_t *buffer,
size_t size, aom_decrypt_cb decrypt_cb, size_t size, aom_decrypt_cb decrypt_cb,
void *decrypt_state) { void *decrypt_state) {
#if CONFIG_ANS
(void)decrypt_cb; (void)decrypt_cb;
(void)decrypt_state; (void)decrypt_state;
#if CONFIG_ANS
if (size > INT_MAX) return 1; if (size > INT_MAX) return 1;
return ans_read_init(r, buffer, (int)size); return ans_read_init(r, buffer, (int)size);
#elif CONFIG_DAALA_EC
(void)decrypt_cb;
(void)decrypt_state;
return aom_daala_reader_init(r, buffer, (int)size);
#else #else
return aom_dk_reader_init(r, buffer, size, decrypt_cb, decrypt_state); return aom_daala_reader_init(r, buffer, (int)size);
#endif #endif
} }
@ -89,20 +78,16 @@ static INLINE const uint8_t *aom_reader_find_end(aom_reader *r) {
(void)r; (void)r;
assert(0 && "Use the raw buffer size with ANS"); assert(0 && "Use the raw buffer size with ANS");
return NULL; return NULL;
#elif CONFIG_DAALA_EC
return aom_daala_reader_find_end(r);
#else #else
return aom_dk_reader_find_end(r); return aom_daala_reader_find_end(r);
#endif #endif
} }
static INLINE int aom_reader_has_error(aom_reader *r) { static INLINE int aom_reader_has_error(aom_reader *r) {
#if CONFIG_ANS #if CONFIG_ANS
return ans_reader_has_error(r); return ans_reader_has_error(r);
#elif CONFIG_DAALA_EC
return aom_daala_reader_has_error(r);
#else #else
return aom_dk_reader_has_error(r); return aom_daala_reader_has_error(r);
#endif #endif
} }
@ -112,10 +97,8 @@ static INLINE uint32_t aom_reader_tell(const aom_reader *r) {
(void)r; (void)r;
assert(0 && "aom_reader_tell() is unimplemented for ANS"); assert(0 && "aom_reader_tell() is unimplemented for ANS");
return 0; return 0;
#elif CONFIG_DAALA_EC
return aom_daala_reader_tell(r);
#else #else
return aom_dk_reader_tell(r); return aom_daala_reader_tell(r);
#endif #endif
} }
@ -125,10 +108,8 @@ static INLINE uint32_t aom_reader_tell_frac(const aom_reader *r) {
(void)r; (void)r;
assert(0 && "aom_reader_tell_frac() is unimplemented for ANS"); assert(0 && "aom_reader_tell_frac() is unimplemented for ANS");
return 0; return 0;
#elif CONFIG_DAALA_EC
return aom_daala_reader_tell_frac(r);
#else #else
return aom_dk_reader_tell_frac(r); return aom_daala_reader_tell_frac(r);
#endif #endif
} }
@ -155,10 +136,8 @@ static INLINE int aom_read_(aom_reader *r, int prob ACCT_STR_PARAM) {
int ret; int ret;
#if CONFIG_ANS #if CONFIG_ANS
ret = rabs_read(r, prob); ret = rabs_read(r, prob);
#elif CONFIG_DAALA_EC
ret = aom_daala_read(r, prob);
#else #else
ret = aom_dk_read(r, prob); ret = aom_daala_read(r, prob);
#endif #endif
#if CONFIG_ACCOUNTING #if CONFIG_ACCOUNTING
if (ACCT_STR_NAME) aom_process_accounting(r, ACCT_STR_NAME); if (ACCT_STR_NAME) aom_process_accounting(r, ACCT_STR_NAME);
@ -171,7 +150,7 @@ static INLINE int aom_read_bit_(aom_reader *r ACCT_STR_PARAM) {
int ret; int ret;
#if CONFIG_ANS #if CONFIG_ANS
ret = rabs_read_bit(r); // Non trivial optimization at half probability ret = rabs_read_bit(r); // Non trivial optimization at half probability
#elif CONFIG_DAALA_EC && CONFIG_RAWBITS #elif CONFIG_RAWBITS
// Note this uses raw bits and is not the same as aom_daala_read(r, 128); // Note this uses raw bits and is not the same as aom_daala_read(r, 128);
// Calls to this function are omitted from raw symbol accounting. // Calls to this function are omitted from raw symbol accounting.
ret = aom_daala_read_bit(r); ret = aom_daala_read_bit(r);
@ -194,28 +173,14 @@ static INLINE int aom_read_literal_(aom_reader *r, int bits ACCT_STR_PARAM) {
return literal; return literal;
} }
static INLINE int aom_read_tree_as_bits(aom_reader *r,
const aom_tree_index *tree,
const aom_prob *probs) {
aom_tree_index i = 0;
while ((i = tree[i + aom_read(r, probs[i >> 1], NULL)]) > 0) continue;
return -i;
}
#if CONFIG_EC_MULTISYMBOL
static INLINE int aom_read_cdf_(aom_reader *r, const aom_cdf_prob *cdf, static INLINE int aom_read_cdf_(aom_reader *r, const aom_cdf_prob *cdf,
int nsymbs ACCT_STR_PARAM) { int nsymbs ACCT_STR_PARAM) {
int ret; int ret;
#if CONFIG_ANS #if CONFIG_ANS
(void)nsymbs; (void)nsymbs;
ret = rans_read(r, cdf); ret = rans_read(r, cdf);
#elif CONFIG_DAALA_EC
ret = daala_read_symbol(r, cdf, nsymbs);
#else #else
#error \ ret = daala_read_symbol(r, cdf, nsymbs);
"CONFIG_EC_MULTISYMBOL is selected without a valid backing entropy " \
"coder. Enable daala_ec or ans for a valid configuration."
#endif #endif
#if CONFIG_ACCOUNTING #if CONFIG_ACCOUNTING
@ -253,16 +218,11 @@ static INLINE int aom_read_tree_as_cdf(aom_reader *r,
} while (i > 0); } while (i > 0);
return -i; return -i;
} }
#endif // CONFIG_EC_MULTISYMBOL
static INLINE int aom_read_tree_(aom_reader *r, const aom_tree_index *tree, static INLINE int aom_read_tree_(aom_reader *r, const aom_tree_index *tree,
const aom_prob *probs ACCT_STR_PARAM) { const aom_prob *probs ACCT_STR_PARAM) {
int ret; int ret;
#if CONFIG_EC_MULTISYMBOL
ret = aom_read_tree_as_cdf(r, tree, probs); ret = aom_read_tree_as_cdf(r, tree, probs);
#else
ret = aom_read_tree_as_bits(r, tree, probs);
#endif
#if CONFIG_ACCOUNTING #if CONFIG_ACCOUNTING
if (ACCT_STR_NAME) aom_process_accounting(r, ACCT_STR_NAME); if (ACCT_STR_NAME) aom_process_accounting(r, ACCT_STR_NAME);
#endif #endif

2
third_party/aom/aom_dsp/bitreader_buffer.c vendored
View file

@ -24,7 +24,7 @@ int aom_rb_read_bit(struct aom_read_bit_buffer *rb) {
rb->bit_offset = off + 1; rb->bit_offset = off + 1;
return bit; return bit;
} else { } else {
rb->error_handler(rb->error_handler_data); if (rb->error_handler) rb->error_handler(rb->error_handler_data);
return 0; return 0;
} }
} }

63
third_party/aom/aom_dsp/bitwriter.h vendored
View file

@ -14,16 +14,11 @@
#include <assert.h> #include <assert.h>
#include "./aom_config.h" #include "./aom_config.h"
#if CONFIG_EC_ADAPT && !CONFIG_EC_MULTISYMBOL
#error "CONFIG_EC_ADAPT is enabled without enabling CONFIG_EC_MULTISYMBOL"
#endif
#if CONFIG_ANS #if CONFIG_ANS
#include "aom_dsp/buf_ans.h" #include "aom_dsp/buf_ans.h"
#elif CONFIG_DAALA_EC
#include "aom_dsp/daalaboolwriter.h"
#else #else
#include "aom_dsp/dkboolwriter.h" #include "aom_dsp/daalaboolwriter.h"
#endif #endif
#include "aom_dsp/prob.h" #include "aom_dsp/prob.h"
@ -38,10 +33,8 @@ extern "C" {
#if CONFIG_ANS #if CONFIG_ANS
typedef struct BufAnsCoder aom_writer; typedef struct BufAnsCoder aom_writer;
#elif CONFIG_DAALA_EC
typedef struct daala_writer aom_writer;
#else #else
typedef struct aom_dk_writer aom_writer; typedef struct daala_writer aom_writer;
#endif #endif
typedef struct TOKEN_STATS { typedef struct TOKEN_STATS {
@ -72,10 +65,8 @@ static INLINE void aom_start_encode(aom_writer *bc, uint8_t *buffer) {
(void)bc; (void)bc;
(void)buffer; (void)buffer;
assert(0 && "buf_ans requires a more complicated startup procedure"); assert(0 && "buf_ans requires a more complicated startup procedure");
#elif CONFIG_DAALA_EC
aom_daala_start_encode(bc, buffer);
#else #else
aom_dk_start_encode(bc, buffer); aom_daala_start_encode(bc, buffer);
#endif #endif
} }
@ -83,20 +74,16 @@ static INLINE void aom_stop_encode(aom_writer *bc) {
#if CONFIG_ANS #if CONFIG_ANS
(void)bc; (void)bc;
assert(0 && "buf_ans requires a more complicated shutdown procedure"); assert(0 && "buf_ans requires a more complicated shutdown procedure");
#elif CONFIG_DAALA_EC
aom_daala_stop_encode(bc);
#else #else
aom_dk_stop_encode(bc); aom_daala_stop_encode(bc);
#endif #endif
} }
static INLINE void aom_write(aom_writer *br, int bit, int probability) { static INLINE void aom_write(aom_writer *br, int bit, int probability) {
#if CONFIG_ANS #if CONFIG_ANS
buf_rabs_write(br, bit, probability); buf_rabs_write(br, bit, probability);
#elif CONFIG_DAALA_EC
aom_daala_write(br, bit, probability);
#else #else
aom_dk_write(br, bit, probability); aom_daala_write(br, bit, probability);
#endif #endif
} }
@ -113,7 +100,7 @@ static INLINE void aom_write_record(aom_writer *br, int bit, int probability,
static INLINE void aom_write_bit(aom_writer *w, int bit) { static INLINE void aom_write_bit(aom_writer *w, int bit) {
#if CONFIG_ANS #if CONFIG_ANS
buf_rabs_write_bit(w, bit); buf_rabs_write_bit(w, bit);
#elif CONFIG_DAALA_EC && CONFIG_RAWBITS #elif CONFIG_RAWBITS
// Note this uses raw bits and is not the same as aom_daala_write(r, 128); // Note this uses raw bits and is not the same as aom_daala_write(r, 128);
aom_daala_write_bit(w, bit); aom_daala_write_bit(w, bit);
#else #else
@ -137,28 +124,6 @@ static INLINE void aom_write_literal(aom_writer *w, int data, int bits) {
for (bit = bits - 1; bit >= 0; bit--) aom_write_bit(w, 1 & (data >> bit)); for (bit = bits - 1; bit >= 0; bit--) aom_write_bit(w, 1 & (data >> bit));
} }
static INLINE void aom_write_tree_as_bits(aom_writer *w,
const aom_tree_index *tr,
const aom_prob *probs, int bits,
int len, aom_tree_index i) {
do {
const int bit = (bits >> --len) & 1;
aom_write(w, bit, probs[i >> 1]);
i = tr[i + bit];
} while (len);
}
static INLINE void aom_write_tree_as_bits_record(
aom_writer *w, const aom_tree_index *tr, const aom_prob *probs, int bits,
int len, aom_tree_index i, TOKEN_STATS *token_stats) {
do {
const int bit = (bits >> --len) & 1;
aom_write_record(w, bit, probs[i >> 1], token_stats);
i = tr[i + bit];
} while (len);
}
#if CONFIG_EC_MULTISYMBOL
static INLINE void aom_write_cdf(aom_writer *w, int symb, static INLINE void aom_write_cdf(aom_writer *w, int symb,
const aom_cdf_prob *cdf, int nsymbs) { const aom_cdf_prob *cdf, int nsymbs) {
#if CONFIG_ANS #if CONFIG_ANS
@ -167,12 +132,8 @@ static INLINE void aom_write_cdf(aom_writer *w, int symb,
const aom_cdf_prob cum_prob = symb > 0 ? cdf[symb - 1] : 0; const aom_cdf_prob cum_prob = symb > 0 ? cdf[symb - 1] : 0;
const aom_cdf_prob prob = cdf[symb] - cum_prob; const aom_cdf_prob prob = cdf[symb] - cum_prob;
buf_rans_write(w, cum_prob, prob); buf_rans_write(w, cum_prob, prob);
#elif CONFIG_DAALA_EC
daala_write_symbol(w, symb, cdf, nsymbs);
#else #else
#error \ daala_write_symbol(w, symb, cdf, nsymbs);
"CONFIG_EC_MULTISYMBOL is selected without a valid backing entropy " \
"coder. Enable daala_ec or ans for a valid configuration."
#endif #endif
} }
@ -223,16 +184,10 @@ static INLINE void aom_write_tree_as_cdf(aom_writer *w,
} while (len); } while (len);
} }
#endif // CONFIG_EC_MULTISYMBOL
static INLINE void aom_write_tree(aom_writer *w, const aom_tree_index *tree, static INLINE void aom_write_tree(aom_writer *w, const aom_tree_index *tree,
const aom_prob *probs, int bits, int len, const aom_prob *probs, int bits, int len,
aom_tree_index i) { aom_tree_index i) {
#if CONFIG_EC_MULTISYMBOL
aom_write_tree_as_cdf(w, tree, probs, bits, len, i); aom_write_tree_as_cdf(w, tree, probs, bits, len, i);
#else
aom_write_tree_as_bits(w, tree, probs, bits, len, i);
#endif
} }
static INLINE void aom_write_tree_record(aom_writer *w, static INLINE void aom_write_tree_record(aom_writer *w,
@ -240,12 +195,8 @@ static INLINE void aom_write_tree_record(aom_writer *w,
const aom_prob *probs, int bits, const aom_prob *probs, int bits,
int len, aom_tree_index i, int len, aom_tree_index i,
TOKEN_STATS *token_stats) { TOKEN_STATS *token_stats) {
#if CONFIG_EC_MULTISYMBOL
(void)token_stats; (void)token_stats;
aom_write_tree_as_cdf(w, tree, probs, bits, len, i); aom_write_tree_as_cdf(w, tree, probs, bits, len, i);
#else
aom_write_tree_as_bits_record(w, tree, probs, bits, len, i, token_stats);
#endif
} }
#ifdef __cplusplus #ifdef __cplusplus

110
third_party/aom/aom_dsp/dkboolreader.c vendored
View file

@ -1,110 +0,0 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include <stdlib.h>
#include "./aom_config.h"
#include "aom_dsp/dkboolreader.h"
#include "aom_dsp/prob.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_ports/mem.h"
#include "aom_mem/aom_mem.h"
#include "aom_util/endian_inl.h"
static INLINE int aom_dk_read_bit(struct aom_dk_reader *r) {
return aom_dk_read(r, 128); // aom_prob_half
}
int aom_dk_reader_init(struct aom_dk_reader *r, const uint8_t *buffer,
size_t size, aom_decrypt_cb decrypt_cb,
void *decrypt_state) {
if (size && !buffer) {
return 1;
} else {
r->buffer_end = buffer + size;
r->buffer_start = r->buffer = buffer;
r->value = 0;
r->count = -8;
r->range = 255;
r->decrypt_cb = decrypt_cb;
r->decrypt_state = decrypt_state;
aom_dk_reader_fill(r);
#if CONFIG_ACCOUNTING
r->accounting = NULL;
#endif
return aom_dk_read_bit(r) != 0; // marker bit
}
}
void aom_dk_reader_fill(struct aom_dk_reader *r) {
const uint8_t *const buffer_end = r->buffer_end;
const uint8_t *buffer = r->buffer;
const uint8_t *buffer_start = buffer;
BD_VALUE value = r->value;
int count = r->count;
const size_t bytes_left = buffer_end - buffer;
const size_t bits_left = bytes_left * CHAR_BIT;
int shift = BD_VALUE_SIZE - CHAR_BIT - (count + CHAR_BIT);
if (r->decrypt_cb) {
size_t n = AOMMIN(sizeof(r->clear_buffer), bytes_left);
r->decrypt_cb(r->decrypt_state, buffer, r->clear_buffer, (int)n);
buffer = r->clear_buffer;
buffer_start = r->clear_buffer;
}
if (bits_left > BD_VALUE_SIZE) {
const int bits = (shift & 0xfffffff8) + CHAR_BIT;
BD_VALUE nv;
BD_VALUE big_endian_values;
memcpy(&big_endian_values, buffer, sizeof(BD_VALUE));
#if SIZE_MAX == 0xffffffffffffffffULL
big_endian_values = HToBE64(big_endian_values);
#else
big_endian_values = HToBE32(big_endian_values);
#endif
nv = big_endian_values >> (BD_VALUE_SIZE - bits);
count += bits;
buffer += (bits >> 3);
value = r->value | (nv << (shift & 0x7));
} else {
const int bits_over = (int)(shift + CHAR_BIT - (int)bits_left);
int loop_end = 0;
if (bits_over >= 0) {
count += LOTS_OF_BITS;
loop_end = bits_over;
}
if (bits_over < 0 || bits_left) {
while (shift >= loop_end) {
count += CHAR_BIT;
value |= (BD_VALUE)*buffer++ << shift;
shift -= CHAR_BIT;
}
}
}
// NOTE: Variable 'buffer' may not relate to 'r->buffer' after decryption,
// so we increase 'r->buffer' by the amount that 'buffer' moved, rather than
// assign 'buffer' to 'r->buffer'.
r->buffer += buffer - buffer_start;
r->value = value;
r->count = count;
}
const uint8_t *aom_dk_reader_find_end(struct aom_dk_reader *r) {
// Find the end of the coded buffer
while (r->count > CHAR_BIT && r->count < BD_VALUE_SIZE) {
r->count -= CHAR_BIT;
r->buffer--;
}
return r->buffer;
}

181
third_party/aom/aom_dsp/dkboolreader.h vendored
View file

@ -1,181 +0,0 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#ifndef AOM_DSP_DKBOOLREADER_H_
#define AOM_DSP_DKBOOLREADER_H_
#include <assert.h>
#include <stddef.h>
#include <limits.h>
#include "./aom_config.h"
#if CONFIG_BITSTREAM_DEBUG
#include <assert.h>
#include <stdio.h>
#include "aom_util/debug_util.h"
#endif // CONFIG_BITSTREAM_DEBUG
#include "aom_ports/mem.h"
#include "aom/aomdx.h"
#include "aom/aom_integer.h"
#include "aom_dsp/prob.h"
#if CONFIG_ACCOUNTING
#include "av1/decoder/accounting.h"
#endif
#ifdef __cplusplus
extern "C" {
#endif
typedef size_t BD_VALUE;
#define BD_VALUE_SIZE ((int)sizeof(BD_VALUE) * CHAR_BIT)
// This is meant to be a large, positive constant that can still be efficiently
// loaded as an immediate (on platforms like ARM, for example).
// Even relatively modest values like 100 would work fine.
#define LOTS_OF_BITS 0x40000000
struct aom_dk_reader {
// Be careful when reordering this struct, it may impact the cache negatively.
BD_VALUE value;
unsigned int range;
int count;
const uint8_t *buffer_start;
const uint8_t *buffer_end;
const uint8_t *buffer;
aom_decrypt_cb decrypt_cb;
void *decrypt_state;
uint8_t clear_buffer[sizeof(BD_VALUE) + 1];
#if CONFIG_ACCOUNTING
Accounting *accounting;
#endif
};
int aom_dk_reader_init(struct aom_dk_reader *r, const uint8_t *buffer,
size_t size, aom_decrypt_cb decrypt_cb,
void *decrypt_state);
void aom_dk_reader_fill(struct aom_dk_reader *r);
const uint8_t *aom_dk_reader_find_end(struct aom_dk_reader *r);
static INLINE uint32_t aom_dk_reader_tell(const struct aom_dk_reader *r) {
const uint32_t bits_read =
(uint32_t)((r->buffer - r->buffer_start) * CHAR_BIT);
const int count =
(r->count < LOTS_OF_BITS) ? r->count : r->count - LOTS_OF_BITS;
assert(r->buffer >= r->buffer_start);
return bits_read - (count + CHAR_BIT);
}
/*The resolution of fractional-precision bit usage measurements, i.e.,
3 => 1/8th bits.*/
#define DK_BITRES (3)
static INLINE uint32_t aom_dk_reader_tell_frac(const struct aom_dk_reader *r) {
uint32_t num_bits;
uint32_t range;
int l;
int i;
num_bits = aom_dk_reader_tell(r) << DK_BITRES;
range = r->range;
l = 0;
for (i = DK_BITRES; i-- > 0;) {
int b;
range = range * range >> 7;
b = (int)(range >> 8);
l = l << 1 | b;
range >>= b;
}
return num_bits - l;
}
static INLINE int aom_dk_reader_has_error(struct aom_dk_reader *r) {
// Check if we have reached the end of the buffer.
//
// Variable 'count' stores the number of bits in the 'value' buffer, minus
// 8. The top byte is part of the algorithm, and the remainder is buffered
// to be shifted into it. So if count == 8, the top 16 bits of 'value' are
// occupied, 8 for the algorithm and 8 in the buffer.
//
// When reading a byte from the user's buffer, count is filled with 8 and
// one byte is filled into the value buffer. When we reach the end of the
// data, count is additionally filled with LOTS_OF_BITS. So when
// count == LOTS_OF_BITS - 1, the user's data has been exhausted.
//
// 1 if we have tried to decode bits after the end of stream was encountered.
// 0 No error.
return r->count > BD_VALUE_SIZE && r->count < LOTS_OF_BITS;
}
static INLINE int aom_dk_read(struct aom_dk_reader *r, int prob) {
unsigned int bit = 0;
BD_VALUE value;
BD_VALUE bigsplit;
int count;
unsigned int range;
unsigned int split = (r->range * prob + (256 - prob)) >> CHAR_BIT;
if (r->count < 0) aom_dk_reader_fill(r);
value = r->value;
count = r->count;
bigsplit = (BD_VALUE)split << (BD_VALUE_SIZE - CHAR_BIT);
range = split;
if (value >= bigsplit) {
range = r->range - split;
value = value - bigsplit;
bit = 1;
}
{
register int shift = aom_norm[range];
range <<= shift;
value <<= shift;
count -= shift;
}
r->value = value;
r->count = count;
r->range = range;
#if CONFIG_BITSTREAM_DEBUG
{
int ref_bit, ref_prob;
const int queue_r = bitstream_queue_get_read();
const int frame_idx = bitstream_queue_get_frame_read();
bitstream_queue_pop(&ref_bit, &ref_prob);
if (prob != ref_prob) {
fprintf(
stderr,
"\n *** prob error, frame_idx_r %d prob %d ref_prob %d queue_r %d\n",
frame_idx, prob, ref_prob, queue_r);
assert(0);
}
if ((int)bit != ref_bit) {
fprintf(stderr, "\n *** bit error, frame_idx_r %d bit %d ref_bit %d\n",
frame_idx, bit, ref_bit);
assert(0);
}
}
#endif // CONFIG_BITSTREAM_DEBUG
return bit;
}
#ifdef __cplusplus
} // extern "C"
#endif
#endif // AOM_DSP_DKBOOLREADER_H_

44
third_party/aom/aom_dsp/dkboolwriter.c vendored
View file

@ -1,44 +0,0 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include <assert.h>
#include "./dkboolwriter.h"
static INLINE void aom_dk_write_bit(aom_dk_writer *w, int bit) {
aom_dk_write(w, bit, 128); // aom_prob_half
}
void aom_dk_start_encode(aom_dk_writer *br, uint8_t *source) {
br->lowvalue = 0;
br->range = 255;
br->count = -24;
br->buffer = source;
br->pos = 0;
aom_dk_write_bit(br, 0);
}
void aom_dk_stop_encode(aom_dk_writer *br) {
int i;
#if CONFIG_BITSTREAM_DEBUG
bitstream_queue_set_skip_write(1);
#endif // CONFIG_BITSTREAM_DEBUG
for (i = 0; i < 32; i++) aom_dk_write_bit(br, 0);
#if CONFIG_BITSTREAM_DEBUG
bitstream_queue_set_skip_write(0);
#endif // CONFIG_BITSTREAM_DEBUG
// Ensure there's no ambigous collision with any index marker bytes
if ((br->buffer[br->pos - 1] & 0xe0) == 0xc0) br->buffer[br->pos++] = 0;
}

104
third_party/aom/aom_dsp/dkboolwriter.h vendored
View file

@ -1,104 +0,0 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#ifndef AOM_DSP_DKBOOLWRITER_H_
#define AOM_DSP_DKBOOLWRITER_H_
#include "./aom_config.h"
#if CONFIG_BITSTREAM_DEBUG
#include <stdio.h>
#include "aom_util/debug_util.h"
#endif // CONFIG_BITSTREAM_DEBUG
#include "aom_dsp/prob.h"
#include "aom_ports/mem.h"
#ifdef __cplusplus
extern "C" {
#endif
typedef struct aom_dk_writer {
unsigned int lowvalue;
unsigned int range;
int count;
unsigned int pos;
uint8_t *buffer;
} aom_dk_writer;
void aom_dk_start_encode(aom_dk_writer *bc, uint8_t *buffer);
void aom_dk_stop_encode(aom_dk_writer *bc);
static INLINE void aom_dk_write(aom_dk_writer *br, int bit, int probability) {
unsigned int split;
int count = br->count;
unsigned int range = br->range;
unsigned int lowvalue = br->lowvalue;
register int shift;
#if CONFIG_BITSTREAM_DEBUG
// int queue_r = 0;
// int frame_idx_r = 0;
// int queue_w = bitstream_queue_get_write();
// int frame_idx_w = bitstream_queue_get_frame_write();
// if (frame_idx_w == frame_idx_r && queue_w == queue_r) {
// fprintf(stderr, "\n *** bitstream queue at frame_idx_w %d queue_w %d\n",
// frame_idx_w, queue_w);
// }
bitstream_queue_push(bit, probability);
#endif // CONFIG_BITSTREAM_DEBUG
split = 1 + (((range - 1) * probability) >> 8);
range = split;
if (bit) {
lowvalue += split;
range = br->range - split;
}
shift = aom_norm[range];
range <<= shift;
count += shift;
if (count >= 0) {
int offset = shift - count;
if ((lowvalue << (offset - 1)) & 0x80000000) {
int x = br->pos - 1;
while (x >= 0 && br->buffer[x] == 0xff) {
br->buffer[x] = 0;
x--;
}
br->buffer[x] += 1;
}
br->buffer[br->pos++] = (lowvalue >> (24 - offset));
lowvalue <<= offset;
shift = count;
lowvalue &= 0xffffff;
count -= 8;
}
lowvalue <<= shift;
br->count = count;
br->lowvalue = lowvalue;
br->range = range;
}
#ifdef __cplusplus
} // extern "C"
#endif
#endif // AOM_DSP_DKBOOLWRITER_H_

162
third_party/aom/aom_dsp/intrapred.c vendored
View file

@ -208,33 +208,30 @@ static const int sm_weight_log2_scale = 8;
#if CONFIG_TX64X64 #if CONFIG_TX64X64
// max(block_size_wide[BLOCK_LARGEST], block_size_high[BLOCK_LARGEST]) // max(block_size_wide[BLOCK_LARGEST], block_size_high[BLOCK_LARGEST])
#define MAX_BLOCK_DIM 64 #define MAX_BLOCK_DIM 64
#define NUM_BLOCK_DIMS 6 // log2(MAX_BLOCK_DIM)
#else #else
#define MAX_BLOCK_DIM 32 #define MAX_BLOCK_DIM 32
#define NUM_BLOCK_DIMS 5
#endif // CONFIG_TX64X64 #endif // CONFIG_TX64X64
static const uint8_t sm_weight_arrays[NUM_BLOCK_DIMS][MAX_BLOCK_DIM] = { static const uint8_t sm_weight_arrays[2 * MAX_BLOCK_DIM] = {
// Unused, because we always offset by bs, which is at least 2.
0, 0,
// bs = 2 // bs = 2
{ 255, 128 }, 255, 128,
// bs = 4 // bs = 4
{ 255, 149, 85, 64 }, 255, 149, 85, 64,
// bs = 8 // bs = 8
{ 255, 197, 146, 105, 73, 50, 37, 32 }, 255, 197, 146, 105, 73, 50, 37, 32,
// bs = 16 // bs = 16
{ 255, 225, 196, 170, 145, 123, 102, 84, 68, 54, 43, 33, 26, 20, 17, 16 }, 255, 225, 196, 170, 145, 123, 102, 84, 68, 54, 43, 33, 26, 20, 17, 16,
// bs = 32 // bs = 32
{ 255, 240, 225, 210, 196, 182, 169, 157, 145, 133, 122, 111, 101, 92, 83, 74,
255, 240, 225, 210, 196, 182, 169, 157, 145, 133, 122, 66, 59, 52, 45, 39, 34, 29, 25, 21, 17, 14, 12, 10, 9, 8, 8,
111, 101, 92, 83, 74, 66, 59, 52, 45, 39, 34,
29, 25, 21, 17, 14, 12, 10, 9, 8, 8 },
#if CONFIG_TX64X64 #if CONFIG_TX64X64
// bs = 64 // bs = 64
{ 255, 248, 240, 233, 225, 218, 210, 203, 196, 189, 182, 176, 169, 255, 248, 240, 233, 225, 218, 210, 203, 196, 189, 182, 176, 169, 163, 156,
163, 156, 150, 144, 138, 133, 127, 121, 116, 111, 106, 101, 96, 150, 144, 138, 133, 127, 121, 116, 111, 106, 101, 96, 91, 86, 82, 77, 73, 69,
91, 86, 82, 77, 73, 69, 65, 61, 57, 54, 50, 47, 44, 65, 61, 57, 54, 50, 47, 44, 41, 38, 35, 32, 29, 27, 25, 22, 20, 18, 16, 15,
41, 38, 35, 32, 29, 27, 25, 22, 20, 18, 16, 15, 13, 13, 12, 10, 9, 8, 7, 6, 6, 5, 5, 4, 4, 4,
12, 10, 9, 8, 7, 6, 6, 5, 5, 4, 4, 4 },
#endif // CONFIG_TX64X64 #endif // CONFIG_TX64X64
}; };
@ -250,10 +247,7 @@ static INLINE void smooth_predictor(uint8_t *dst, ptrdiff_t stride, int bs,
const uint8_t *above, const uint8_t *left) { const uint8_t *above, const uint8_t *left) {
const uint8_t below_pred = left[bs - 1]; // estimated by bottom-left pixel const uint8_t below_pred = left[bs - 1]; // estimated by bottom-left pixel
const uint8_t right_pred = above[bs - 1]; // estimated by top-right pixel const uint8_t right_pred = above[bs - 1]; // estimated by top-right pixel
const int arr_index = get_msb(bs) - 1; const uint8_t *const sm_weights = sm_weight_arrays + bs;
assert(arr_index >= 0);
assert(arr_index < NUM_BLOCK_DIMS);
const uint8_t *const sm_weights = sm_weight_arrays[arr_index];
// scale = 2 * 2^sm_weight_log2_scale // scale = 2 * 2^sm_weight_log2_scale
const int log2_scale = 1 + sm_weight_log2_scale; const int log2_scale = 1 + sm_weight_log2_scale;
const uint16_t scale = (1 << sm_weight_log2_scale); const uint16_t scale = (1 << sm_weight_log2_scale);
@ -277,6 +271,64 @@ static INLINE void smooth_predictor(uint8_t *dst, ptrdiff_t stride, int bs,
} }
} }
#if CONFIG_SMOOTH_HV
static INLINE void smooth_v_predictor(uint8_t *dst, ptrdiff_t stride, int bs,
const uint8_t *above,
const uint8_t *left) {
const uint8_t below_pred = left[bs - 1]; // estimated by bottom-left pixel
const uint8_t *const sm_weights = sm_weight_arrays + bs;
// scale = 2^sm_weight_log2_scale
const int log2_scale = sm_weight_log2_scale;
const uint16_t scale = (1 << sm_weight_log2_scale);
sm_weights_sanity_checks(sm_weights, scale, log2_scale + sizeof(*dst));
int r;
for (r = 0; r < bs; r++) {
int c;
for (c = 0; c < bs; ++c) {
const uint8_t pixels[] = { above[c], below_pred };
const uint8_t weights[] = { sm_weights[r], scale - sm_weights[r] };
uint32_t this_pred = 0;
assert(scale >= sm_weights[r]);
int i;
for (i = 0; i < 2; ++i) {
this_pred += weights[i] * pixels[i];
}
dst[c] = clip_pixel(divide_round(this_pred, log2_scale));
}
dst += stride;
}
}
static INLINE void smooth_h_predictor(uint8_t *dst, ptrdiff_t stride, int bs,
const uint8_t *above,
const uint8_t *left) {
const uint8_t right_pred = above[bs - 1]; // estimated by top-right pixel
const uint8_t *const sm_weights = sm_weight_arrays + bs;
// scale = 2^sm_weight_log2_scale
const int log2_scale = sm_weight_log2_scale;
const uint16_t scale = (1 << sm_weight_log2_scale);
sm_weights_sanity_checks(sm_weights, scale, log2_scale + sizeof(*dst));
int r;
for (r = 0; r < bs; r++) {
int c;
for (c = 0; c < bs; ++c) {
const uint8_t pixels[] = { left[r], right_pred };
const uint8_t weights[] = { sm_weights[c], scale - sm_weights[c] };
uint32_t this_pred = 0;
assert(scale >= sm_weights[c]);
int i;
for (i = 0; i < 2; ++i) {
this_pred += weights[i] * pixels[i];
}
dst[c] = clip_pixel(divide_round(this_pred, log2_scale));
}
dst += stride;
}
}
#endif // CONFIG_SMOOTH_HV
#else #else
static INLINE void tm_predictor(uint8_t *dst, ptrdiff_t stride, int bs, static INLINE void tm_predictor(uint8_t *dst, ptrdiff_t stride, int bs,
@ -743,10 +795,7 @@ static INLINE void highbd_smooth_predictor(uint16_t *dst, ptrdiff_t stride,
const uint16_t *left, int bd) { const uint16_t *left, int bd) {
const uint16_t below_pred = left[bs - 1]; // estimated by bottom-left pixel const uint16_t below_pred = left[bs - 1]; // estimated by bottom-left pixel
const uint16_t right_pred = above[bs - 1]; // estimated by top-right pixel const uint16_t right_pred = above[bs - 1]; // estimated by top-right pixel
const int arr_index = get_msb(bs) - 1; const uint8_t *const sm_weights = sm_weight_arrays + bs;
assert(arr_index >= 0);
assert(arr_index < NUM_BLOCK_DIMS);
const uint8_t *const sm_weights = sm_weight_arrays[arr_index];
// scale = 2 * 2^sm_weight_log2_scale // scale = 2 * 2^sm_weight_log2_scale
const int log2_scale = 1 + sm_weight_log2_scale; const int log2_scale = 1 + sm_weight_log2_scale;
const uint16_t scale = (1 << sm_weight_log2_scale); const uint16_t scale = (1 << sm_weight_log2_scale);
@ -770,6 +819,64 @@ static INLINE void highbd_smooth_predictor(uint16_t *dst, ptrdiff_t stride,
} }
} }
#if CONFIG_SMOOTH_HV
static INLINE void highbd_smooth_v_predictor(uint16_t *dst, ptrdiff_t stride,
int bs, const uint16_t *above,
const uint16_t *left, int bd) {
const uint16_t below_pred = left[bs - 1]; // estimated by bottom-left pixel
const uint8_t *const sm_weights = sm_weight_arrays + bs;
// scale = 2^sm_weight_log2_scale
const int log2_scale = sm_weight_log2_scale;
const uint16_t scale = (1 << sm_weight_log2_scale);
sm_weights_sanity_checks(sm_weights, scale, log2_scale + sizeof(*dst));
int r;
for (r = 0; r < bs; r++) {
int c;
for (c = 0; c < bs; ++c) {
const uint16_t pixels[] = { above[c], below_pred };
const uint8_t weights[] = { sm_weights[r], scale - sm_weights[r] };
uint32_t this_pred = 0;
assert(scale >= sm_weights[r]);
int i;
for (i = 0; i < 2; ++i) {
this_pred += weights[i] * pixels[i];
}
dst[c] = clip_pixel_highbd(divide_round(this_pred, log2_scale), bd);
}
dst += stride;
}
}
static INLINE void highbd_smooth_h_predictor(uint16_t *dst, ptrdiff_t stride,
int bs, const uint16_t *above,
const uint16_t *left, int bd) {
const uint16_t right_pred = above[bs - 1]; // estimated by top-right pixel
const uint8_t *const sm_weights = sm_weight_arrays + bs;
// scale = 2^sm_weight_log2_scale
const int log2_scale = sm_weight_log2_scale;
const uint16_t scale = (1 << sm_weight_log2_scale);
sm_weights_sanity_checks(sm_weights, scale, log2_scale + sizeof(*dst));
int r;
for (r = 0; r < bs; r++) {
int c;
for (c = 0; c < bs; ++c) {
const uint16_t pixels[] = { left[r], right_pred };
const uint8_t weights[] = { sm_weights[c], scale - sm_weights[c] };
uint32_t this_pred = 0;
assert(scale >= sm_weights[c]);
int i;
for (i = 0; i < 2; ++i) {
this_pred += weights[i] * pixels[i];
}
dst[c] = clip_pixel_highbd(divide_round(this_pred, log2_scale), bd);
}
dst += stride;
}
}
#endif
#else #else
static INLINE void highbd_tm_predictor(uint16_t *dst, ptrdiff_t stride, int bs, static INLINE void highbd_tm_predictor(uint16_t *dst, ptrdiff_t stride, int bs,
const uint16_t *above, const uint16_t *above,
@ -879,6 +986,7 @@ static INLINE void highbd_dc_predictor(uint16_t *dst, ptrdiff_t stride, int bs,
intra_pred_sized(type, 16) \ intra_pred_sized(type, 16) \
intra_pred_sized(type, 32) \ intra_pred_sized(type, 32) \
intra_pred_sized(type, 64) \ intra_pred_sized(type, 64) \
intra_pred_highbd_sized(type, 2) \
intra_pred_highbd_sized(type, 4) \ intra_pred_highbd_sized(type, 4) \
intra_pred_highbd_sized(type, 8) \ intra_pred_highbd_sized(type, 8) \
intra_pred_highbd_sized(type, 16) \ intra_pred_highbd_sized(type, 16) \
@ -958,8 +1066,12 @@ intra_pred_above_4x4(d153)
intra_pred_allsizes(v) intra_pred_allsizes(v)
intra_pred_allsizes(h) intra_pred_allsizes(h)
#if CONFIG_ALT_INTRA #if CONFIG_ALT_INTRA
intra_pred_allsizes(paeth)
intra_pred_allsizes(smooth) intra_pred_allsizes(smooth)
#if CONFIG_SMOOTH_HV
intra_pred_allsizes(smooth_v)
intra_pred_allsizes(smooth_h)
#endif // CONFIG_SMOOTH_HV
intra_pred_allsizes(paeth)
#else #else
intra_pred_allsizes(tm) intra_pred_allsizes(tm)
#endif // CONFIG_ALT_INTRA #endif // CONFIG_ALT_INTRA

864
third_party/aom/aom_dsp/inv_txfm.c vendored
View file

@ -1442,4 +1442,868 @@ void aom_highbd_idct4x4_1_add_c(const tran_low_t *input, uint8_t *dest8,
} }
} }
void aom_highbd_idct8_c(const tran_low_t *input, tran_low_t *output, int bd) {
tran_low_t step1[8], step2[8];
tran_high_t temp1, temp2;
// stage 1
step1[0] = input[0];
step1[2] = input[4];
step1[1] = input[2];
step1[3] = input[6];
temp1 = input[1] * cospi_28_64 - input[7] * cospi_4_64;
temp2 = input[1] * cospi_4_64 + input[7] * cospi_28_64;
step1[4] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[7] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[5] * cospi_12_64 - input[3] * cospi_20_64;
temp2 = input[5] * cospi_20_64 + input[3] * cospi_12_64;
step1[5] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[6] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
// stage 2 & stage 3 - even half
aom_highbd_idct4_c(step1, step1, bd);
// stage 2 - odd half
step2[4] = HIGHBD_WRAPLOW(step1[4] + step1[5], bd);
step2[5] = HIGHBD_WRAPLOW(step1[4] - step1[5], bd);
step2[6] = HIGHBD_WRAPLOW(-step1[6] + step1[7], bd);
step2[7] = HIGHBD_WRAPLOW(step1[6] + step1[7], bd);
// stage 3 - odd half
step1[4] = step2[4];
temp1 = (step2[6] - step2[5]) * cospi_16_64;
temp2 = (step2[5] + step2[6]) * cospi_16_64;
step1[5] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[6] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[7] = step2[7];
// stage 4
output[0] = HIGHBD_WRAPLOW(step1[0] + step1[7], bd);
output[1] = HIGHBD_WRAPLOW(step1[1] + step1[6], bd);
output[2] = HIGHBD_WRAPLOW(step1[2] + step1[5], bd);
output[3] = HIGHBD_WRAPLOW(step1[3] + step1[4], bd);
output[4] = HIGHBD_WRAPLOW(step1[3] - step1[4], bd);
output[5] = HIGHBD_WRAPLOW(step1[2] - step1[5], bd);
output[6] = HIGHBD_WRAPLOW(step1[1] - step1[6], bd);
output[7] = HIGHBD_WRAPLOW(step1[0] - step1[7], bd);
}
void aom_highbd_iadst4_c(const tran_low_t *input, tran_low_t *output, int bd) {
tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;
tran_low_t x0 = input[0];
tran_low_t x1 = input[1];
tran_low_t x2 = input[2];
tran_low_t x3 = input[3];
(void)bd;
if (!(x0 | x1 | x2 | x3)) {
memset(output, 0, 4 * sizeof(*output));
return;
}
s0 = sinpi_1_9 * x0;
s1 = sinpi_2_9 * x0;
s2 = sinpi_3_9 * x1;
s3 = sinpi_4_9 * x2;
s4 = sinpi_1_9 * x2;
s5 = sinpi_2_9 * x3;
s6 = sinpi_4_9 * x3;
s7 = (tran_high_t)HIGHBD_WRAPLOW(x0 - x2 + x3, bd);
s0 = s0 + s3 + s5;
s1 = s1 - s4 - s6;
s3 = s2;
s2 = sinpi_3_9 * s7;
// 1-D transform scaling factor is sqrt(2).
// The overall dynamic range is 14b (input) + 14b (multiplication scaling)
// + 1b (addition) = 29b.
// Hence the output bit depth is 15b.
output[0] = HIGHBD_WRAPLOW(dct_const_round_shift(s0 + s3), bd);
output[1] = HIGHBD_WRAPLOW(dct_const_round_shift(s1 + s3), bd);
output[2] = HIGHBD_WRAPLOW(dct_const_round_shift(s2), bd);
output[3] = HIGHBD_WRAPLOW(dct_const_round_shift(s0 + s1 - s3), bd);
}
void aom_highbd_iadst8_c(const tran_low_t *input, tran_low_t *output, int bd) {
tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;
tran_low_t x0 = input[7];
tran_low_t x1 = input[0];
tran_low_t x2 = input[5];
tran_low_t x3 = input[2];
tran_low_t x4 = input[3];
tran_low_t x5 = input[4];
tran_low_t x6 = input[1];
tran_low_t x7 = input[6];
(void)bd;
if (!(x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7)) {
memset(output, 0, 8 * sizeof(*output));
return;
}
// stage 1
s0 = cospi_2_64 * x0 + cospi_30_64 * x1;
s1 = cospi_30_64 * x0 - cospi_2_64 * x1;
s2 = cospi_10_64 * x2 + cospi_22_64 * x3;
s3 = cospi_22_64 * x2 - cospi_10_64 * x3;
s4 = cospi_18_64 * x4 + cospi_14_64 * x5;
s5 = cospi_14_64 * x4 - cospi_18_64 * x5;
s6 = cospi_26_64 * x6 + cospi_6_64 * x7;
s7 = cospi_6_64 * x6 - cospi_26_64 * x7;
x0 = HIGHBD_WRAPLOW(dct_const_round_shift(s0 + s4), bd);
x1 = HIGHBD_WRAPLOW(dct_const_round_shift(s1 + s5), bd);
x2 = HIGHBD_WRAPLOW(dct_const_round_shift(s2 + s6), bd);
x3 = HIGHBD_WRAPLOW(dct_const_round_shift(s3 + s7), bd);
x4 = HIGHBD_WRAPLOW(dct_const_round_shift(s0 - s4), bd);
x5 = HIGHBD_WRAPLOW(dct_const_round_shift(s1 - s5), bd);
x6 = HIGHBD_WRAPLOW(dct_const_round_shift(s2 - s6), bd);
x7 = HIGHBD_WRAPLOW(dct_const_round_shift(s3 - s7), bd);
// stage 2
s0 = x0;
s1 = x1;
s2 = x2;
s3 = x3;
s4 = cospi_8_64 * x4 + cospi_24_64 * x5;
s5 = cospi_24_64 * x4 - cospi_8_64 * x5;
s6 = -cospi_24_64 * x6 + cospi_8_64 * x7;
s7 = cospi_8_64 * x6 + cospi_24_64 * x7;
x0 = HIGHBD_WRAPLOW(s0 + s2, bd);
x1 = HIGHBD_WRAPLOW(s1 + s3, bd);
x2 = HIGHBD_WRAPLOW(s0 - s2, bd);
x3 = HIGHBD_WRAPLOW(s1 - s3, bd);
x4 = HIGHBD_WRAPLOW(dct_const_round_shift(s4 + s6), bd);
x5 = HIGHBD_WRAPLOW(dct_const_round_shift(s5 + s7), bd);
x6 = HIGHBD_WRAPLOW(dct_const_round_shift(s4 - s6), bd);
x7 = HIGHBD_WRAPLOW(dct_const_round_shift(s5 - s7), bd);
// stage 3
s2 = cospi_16_64 * (x2 + x3);
s3 = cospi_16_64 * (x2 - x3);
s6 = cospi_16_64 * (x6 + x7);
s7 = cospi_16_64 * (x6 - x7);
x2 = HIGHBD_WRAPLOW(dct_const_round_shift(s2), bd);
x3 = HIGHBD_WRAPLOW(dct_const_round_shift(s3), bd);
x6 = HIGHBD_WRAPLOW(dct_const_round_shift(s6), bd);
x7 = HIGHBD_WRAPLOW(dct_const_round_shift(s7), bd);
output[0] = HIGHBD_WRAPLOW(x0, bd);
output[1] = HIGHBD_WRAPLOW(-x4, bd);
output[2] = HIGHBD_WRAPLOW(x6, bd);
output[3] = HIGHBD_WRAPLOW(-x2, bd);
output[4] = HIGHBD_WRAPLOW(x3, bd);
output[5] = HIGHBD_WRAPLOW(-x7, bd);
output[6] = HIGHBD_WRAPLOW(x5, bd);
output[7] = HIGHBD_WRAPLOW(-x1, bd);
}
void aom_highbd_idct16_c(const tran_low_t *input, tran_low_t *output, int bd) {
tran_low_t step1[16], step2[16];
tran_high_t temp1, temp2;
(void)bd;
// stage 1
step1[0] = input[0 / 2];
step1[1] = input[16 / 2];
step1[2] = input[8 / 2];
step1[3] = input[24 / 2];
step1[4] = input[4 / 2];
step1[5] = input[20 / 2];
step1[6] = input[12 / 2];
step1[7] = input[28 / 2];
step1[8] = input[2 / 2];
step1[9] = input[18 / 2];
step1[10] = input[10 / 2];
step1[11] = input[26 / 2];
step1[12] = input[6 / 2];
step1[13] = input[22 / 2];
step1[14] = input[14 / 2];
step1[15] = input[30 / 2];
// stage 2
step2[0] = step1[0];
step2[1] = step1[1];
step2[2] = step1[2];
step2[3] = step1[3];
step2[4] = step1[4];
step2[5] = step1[5];
step2[6] = step1[6];
step2[7] = step1[7];
temp1 = step1[8] * cospi_30_64 - step1[15] * cospi_2_64;
temp2 = step1[8] * cospi_2_64 + step1[15] * cospi_30_64;
step2[8] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[15] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[9] * cospi_14_64 - step1[14] * cospi_18_64;
temp2 = step1[9] * cospi_18_64 + step1[14] * cospi_14_64;
step2[9] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[14] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[10] * cospi_22_64 - step1[13] * cospi_10_64;
temp2 = step1[10] * cospi_10_64 + step1[13] * cospi_22_64;
step2[10] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[13] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[11] * cospi_6_64 - step1[12] * cospi_26_64;
temp2 = step1[11] * cospi_26_64 + step1[12] * cospi_6_64;
step2[11] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[12] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
// stage 3
step1[0] = step2[0];
step1[1] = step2[1];
step1[2] = step2[2];
step1[3] = step2[3];
temp1 = step2[4] * cospi_28_64 - step2[7] * cospi_4_64;
temp2 = step2[4] * cospi_4_64 + step2[7] * cospi_28_64;
step1[4] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[7] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step2[5] * cospi_12_64 - step2[6] * cospi_20_64;
temp2 = step2[5] * cospi_20_64 + step2[6] * cospi_12_64;
step1[5] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[6] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[8] = HIGHBD_WRAPLOW(step2[8] + step2[9], bd);
step1[9] = HIGHBD_WRAPLOW(step2[8] - step2[9], bd);
step1[10] = HIGHBD_WRAPLOW(-step2[10] + step2[11], bd);
step1[11] = HIGHBD_WRAPLOW(step2[10] + step2[11], bd);
step1[12] = HIGHBD_WRAPLOW(step2[12] + step2[13], bd);
step1[13] = HIGHBD_WRAPLOW(step2[12] - step2[13], bd);
step1[14] = HIGHBD_WRAPLOW(-step2[14] + step2[15], bd);
step1[15] = HIGHBD_WRAPLOW(step2[14] + step2[15], bd);
// stage 4
temp1 = (step1[0] + step1[1]) * cospi_16_64;
temp2 = (step1[0] - step1[1]) * cospi_16_64;
step2[0] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[1] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[2] * cospi_24_64 - step1[3] * cospi_8_64;
temp2 = step1[2] * cospi_8_64 + step1[3] * cospi_24_64;
step2[2] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[3] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step2[4] = HIGHBD_WRAPLOW(step1[4] + step1[5], bd);
step2[5] = HIGHBD_WRAPLOW(step1[4] - step1[5], bd);
step2[6] = HIGHBD_WRAPLOW(-step1[6] + step1[7], bd);
step2[7] = HIGHBD_WRAPLOW(step1[6] + step1[7], bd);
step2[8] = step1[8];
step2[15] = step1[15];
temp1 = -step1[9] * cospi_8_64 + step1[14] * cospi_24_64;
temp2 = step1[9] * cospi_24_64 + step1[14] * cospi_8_64;
step2[9] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[14] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = -step1[10] * cospi_24_64 - step1[13] * cospi_8_64;
temp2 = -step1[10] * cospi_8_64 + step1[13] * cospi_24_64;
step2[10] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[13] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step2[11] = step1[11];
step2[12] = step1[12];
// stage 5
step1[0] = HIGHBD_WRAPLOW(step2[0] + step2[3], bd);
step1[1] = HIGHBD_WRAPLOW(step2[1] + step2[2], bd);
step1[2] = HIGHBD_WRAPLOW(step2[1] - step2[2], bd);
step1[3] = HIGHBD_WRAPLOW(step2[0] - step2[3], bd);
step1[4] = step2[4];
temp1 = (step2[6] - step2[5]) * cospi_16_64;
temp2 = (step2[5] + step2[6]) * cospi_16_64;
step1[5] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[6] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[7] = step2[7];
step1[8] = HIGHBD_WRAPLOW(step2[8] + step2[11], bd);
step1[9] = HIGHBD_WRAPLOW(step2[9] + step2[10], bd);
step1[10] = HIGHBD_WRAPLOW(step2[9] - step2[10], bd);
step1[11] = HIGHBD_WRAPLOW(step2[8] - step2[11], bd);
step1[12] = HIGHBD_WRAPLOW(-step2[12] + step2[15], bd);
step1[13] = HIGHBD_WRAPLOW(-step2[13] + step2[14], bd);
step1[14] = HIGHBD_WRAPLOW(step2[13] + step2[14], bd);
step1[15] = HIGHBD_WRAPLOW(step2[12] + step2[15], bd);
// stage 6
step2[0] = HIGHBD_WRAPLOW(step1[0] + step1[7], bd);
step2[1] = HIGHBD_WRAPLOW(step1[1] + step1[6], bd);
step2[2] = HIGHBD_WRAPLOW(step1[2] + step1[5], bd);
step2[3] = HIGHBD_WRAPLOW(step1[3] + step1[4], bd);
step2[4] = HIGHBD_WRAPLOW(step1[3] - step1[4], bd);
step2[5] = HIGHBD_WRAPLOW(step1[2] - step1[5], bd);
step2[6] = HIGHBD_WRAPLOW(step1[1] - step1[6], bd);
step2[7] = HIGHBD_WRAPLOW(step1[0] - step1[7], bd);
step2[8] = step1[8];
step2[9] = step1[9];
temp1 = (-step1[10] + step1[13]) * cospi_16_64;
temp2 = (step1[10] + step1[13]) * cospi_16_64;
step2[10] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[13] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = (-step1[11] + step1[12]) * cospi_16_64;
temp2 = (step1[11] + step1[12]) * cospi_16_64;
step2[11] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[12] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step2[14] = step1[14];
step2[15] = step1[15];
// stage 7
output[0] = HIGHBD_WRAPLOW(step2[0] + step2[15], bd);
output[1] = HIGHBD_WRAPLOW(step2[1] + step2[14], bd);
output[2] = HIGHBD_WRAPLOW(step2[2] + step2[13], bd);
output[3] = HIGHBD_WRAPLOW(step2[3] + step2[12], bd);
output[4] = HIGHBD_WRAPLOW(step2[4] + step2[11], bd);
output[5] = HIGHBD_WRAPLOW(step2[5] + step2[10], bd);
output[6] = HIGHBD_WRAPLOW(step2[6] + step2[9], bd);
output[7] = HIGHBD_WRAPLOW(step2[7] + step2[8], bd);
output[8] = HIGHBD_WRAPLOW(step2[7] - step2[8], bd);
output[9] = HIGHBD_WRAPLOW(step2[6] - step2[9], bd);
output[10] = HIGHBD_WRAPLOW(step2[5] - step2[10], bd);
output[11] = HIGHBD_WRAPLOW(step2[4] - step2[11], bd);
output[12] = HIGHBD_WRAPLOW(step2[3] - step2[12], bd);
output[13] = HIGHBD_WRAPLOW(step2[2] - step2[13], bd);
output[14] = HIGHBD_WRAPLOW(step2[1] - step2[14], bd);
output[15] = HIGHBD_WRAPLOW(step2[0] - step2[15], bd);
}
void aom_highbd_iadst16_c(const tran_low_t *input, tran_low_t *output, int bd) {
tran_high_t s0, s1, s2, s3, s4, s5, s6, s7, s8;
tran_high_t s9, s10, s11, s12, s13, s14, s15;
tran_low_t x0 = input[15];
tran_low_t x1 = input[0];
tran_low_t x2 = input[13];
tran_low_t x3 = input[2];
tran_low_t x4 = input[11];
tran_low_t x5 = input[4];
tran_low_t x6 = input[9];
tran_low_t x7 = input[6];
tran_low_t x8 = input[7];
tran_low_t x9 = input[8];
tran_low_t x10 = input[5];
tran_low_t x11 = input[10];
tran_low_t x12 = input[3];
tran_low_t x13 = input[12];
tran_low_t x14 = input[1];
tran_low_t x15 = input[14];
(void)bd;
if (!(x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7 | x8 | x9 | x10 | x11 | x12 |
x13 | x14 | x15)) {
memset(output, 0, 16 * sizeof(*output));
return;
}
// stage 1
s0 = x0 * cospi_1_64 + x1 * cospi_31_64;
s1 = x0 * cospi_31_64 - x1 * cospi_1_64;
s2 = x2 * cospi_5_64 + x3 * cospi_27_64;
s3 = x2 * cospi_27_64 - x3 * cospi_5_64;
s4 = x4 * cospi_9_64 + x5 * cospi_23_64;
s5 = x4 * cospi_23_64 - x5 * cospi_9_64;
s6 = x6 * cospi_13_64 + x7 * cospi_19_64;
s7 = x6 * cospi_19_64 - x7 * cospi_13_64;
s8 = x8 * cospi_17_64 + x9 * cospi_15_64;
s9 = x8 * cospi_15_64 - x9 * cospi_17_64;
s10 = x10 * cospi_21_64 + x11 * cospi_11_64;
s11 = x10 * cospi_11_64 - x11 * cospi_21_64;
s12 = x12 * cospi_25_64 + x13 * cospi_7_64;
s13 = x12 * cospi_7_64 - x13 * cospi_25_64;
s14 = x14 * cospi_29_64 + x15 * cospi_3_64;
s15 = x14 * cospi_3_64 - x15 * cospi_29_64;
x0 = HIGHBD_WRAPLOW(dct_const_round_shift(s0 + s8), bd);
x1 = HIGHBD_WRAPLOW(dct_const_round_shift(s1 + s9), bd);
x2 = HIGHBD_WRAPLOW(dct_const_round_shift(s2 + s10), bd);
x3 = HIGHBD_WRAPLOW(dct_const_round_shift(s3 + s11), bd);
x4 = HIGHBD_WRAPLOW(dct_const_round_shift(s4 + s12), bd);
x5 = HIGHBD_WRAPLOW(dct_const_round_shift(s5 + s13), bd);
x6 = HIGHBD_WRAPLOW(dct_const_round_shift(s6 + s14), bd);
x7 = HIGHBD_WRAPLOW(dct_const_round_shift(s7 + s15), bd);
x8 = HIGHBD_WRAPLOW(dct_const_round_shift(s0 - s8), bd);
x9 = HIGHBD_WRAPLOW(dct_const_round_shift(s1 - s9), bd);
x10 = HIGHBD_WRAPLOW(dct_const_round_shift(s2 - s10), bd);
x11 = HIGHBD_WRAPLOW(dct_const_round_shift(s3 - s11), bd);
x12 = HIGHBD_WRAPLOW(dct_const_round_shift(s4 - s12), bd);
x13 = HIGHBD_WRAPLOW(dct_const_round_shift(s5 - s13), bd);
x14 = HIGHBD_WRAPLOW(dct_const_round_shift(s6 - s14), bd);
x15 = HIGHBD_WRAPLOW(dct_const_round_shift(s7 - s15), bd);
// stage 2
s0 = x0;
s1 = x1;
s2 = x2;
s3 = x3;
s4 = x4;
s5 = x5;
s6 = x6;
s7 = x7;
s8 = x8 * cospi_4_64 + x9 * cospi_28_64;
s9 = x8 * cospi_28_64 - x9 * cospi_4_64;
s10 = x10 * cospi_20_64 + x11 * cospi_12_64;
s11 = x10 * cospi_12_64 - x11 * cospi_20_64;
s12 = -x12 * cospi_28_64 + x13 * cospi_4_64;
s13 = x12 * cospi_4_64 + x13 * cospi_28_64;
s14 = -x14 * cospi_12_64 + x15 * cospi_20_64;
s15 = x14 * cospi_20_64 + x15 * cospi_12_64;
x0 = HIGHBD_WRAPLOW(s0 + s4, bd);
x1 = HIGHBD_WRAPLOW(s1 + s5, bd);
x2 = HIGHBD_WRAPLOW(s2 + s6, bd);
x3 = HIGHBD_WRAPLOW(s3 + s7, bd);
x4 = HIGHBD_WRAPLOW(s0 - s4, bd);
x5 = HIGHBD_WRAPLOW(s1 - s5, bd);
x6 = HIGHBD_WRAPLOW(s2 - s6, bd);
x7 = HIGHBD_WRAPLOW(s3 - s7, bd);
x8 = HIGHBD_WRAPLOW(dct_const_round_shift(s8 + s12), bd);
x9 = HIGHBD_WRAPLOW(dct_const_round_shift(s9 + s13), bd);
x10 = HIGHBD_WRAPLOW(dct_const_round_shift(s10 + s14), bd);
x11 = HIGHBD_WRAPLOW(dct_const_round_shift(s11 + s15), bd);
x12 = HIGHBD_WRAPLOW(dct_const_round_shift(s8 - s12), bd);
x13 = HIGHBD_WRAPLOW(dct_const_round_shift(s9 - s13), bd);
x14 = HIGHBD_WRAPLOW(dct_const_round_shift(s10 - s14), bd);
x15 = HIGHBD_WRAPLOW(dct_const_round_shift(s11 - s15), bd);
// stage 3
s0 = x0;
s1 = x1;
s2 = x2;
s3 = x3;
s4 = x4 * cospi_8_64 + x5 * cospi_24_64;
s5 = x4 * cospi_24_64 - x5 * cospi_8_64;
s6 = -x6 * cospi_24_64 + x7 * cospi_8_64;
s7 = x6 * cospi_8_64 + x7 * cospi_24_64;
s8 = x8;
s9 = x9;
s10 = x10;
s11 = x11;
s12 = x12 * cospi_8_64 + x13 * cospi_24_64;
s13 = x12 * cospi_24_64 - x13 * cospi_8_64;
s14 = -x14 * cospi_24_64 + x15 * cospi_8_64;
s15 = x14 * cospi_8_64 + x15 * cospi_24_64;
x0 = HIGHBD_WRAPLOW(s0 + s2, bd);
x1 = HIGHBD_WRAPLOW(s1 + s3, bd);
x2 = HIGHBD_WRAPLOW(s0 - s2, bd);
x3 = HIGHBD_WRAPLOW(s1 - s3, bd);
x4 = HIGHBD_WRAPLOW(dct_const_round_shift(s4 + s6), bd);
x5 = HIGHBD_WRAPLOW(dct_const_round_shift(s5 + s7), bd);
x6 = HIGHBD_WRAPLOW(dct_const_round_shift(s4 - s6), bd);
x7 = HIGHBD_WRAPLOW(dct_const_round_shift(s5 - s7), bd);
x8 = HIGHBD_WRAPLOW(s8 + s10, bd);
x9 = HIGHBD_WRAPLOW(s9 + s11, bd);
x10 = HIGHBD_WRAPLOW(s8 - s10, bd);
x11 = HIGHBD_WRAPLOW(s9 - s11, bd);
x12 = HIGHBD_WRAPLOW(dct_const_round_shift(s12 + s14), bd);
x13 = HIGHBD_WRAPLOW(dct_const_round_shift(s13 + s15), bd);
x14 = HIGHBD_WRAPLOW(dct_const_round_shift(s12 - s14), bd);
x15 = HIGHBD_WRAPLOW(dct_const_round_shift(s13 - s15), bd);
// stage 4
s2 = (-cospi_16_64) * (x2 + x3);
s3 = cospi_16_64 * (x2 - x3);
s6 = cospi_16_64 * (x6 + x7);
s7 = cospi_16_64 * (-x6 + x7);
s10 = cospi_16_64 * (x10 + x11);
s11 = cospi_16_64 * (-x10 + x11);
s14 = (-cospi_16_64) * (x14 + x15);
s15 = cospi_16_64 * (x14 - x15);
x2 = HIGHBD_WRAPLOW(dct_const_round_shift(s2), bd);
x3 = HIGHBD_WRAPLOW(dct_const_round_shift(s3), bd);
x6 = HIGHBD_WRAPLOW(dct_const_round_shift(s6), bd);
x7 = HIGHBD_WRAPLOW(dct_const_round_shift(s7), bd);
x10 = HIGHBD_WRAPLOW(dct_const_round_shift(s10), bd);
x11 = HIGHBD_WRAPLOW(dct_const_round_shift(s11), bd);
x14 = HIGHBD_WRAPLOW(dct_const_round_shift(s14), bd);
x15 = HIGHBD_WRAPLOW(dct_const_round_shift(s15), bd);
output[0] = HIGHBD_WRAPLOW(x0, bd);
output[1] = HIGHBD_WRAPLOW(-x8, bd);
output[2] = HIGHBD_WRAPLOW(x12, bd);
output[3] = HIGHBD_WRAPLOW(-x4, bd);
output[4] = HIGHBD_WRAPLOW(x6, bd);
output[5] = HIGHBD_WRAPLOW(x14, bd);
output[6] = HIGHBD_WRAPLOW(x10, bd);
output[7] = HIGHBD_WRAPLOW(x2, bd);
output[8] = HIGHBD_WRAPLOW(x3, bd);
output[9] = HIGHBD_WRAPLOW(x11, bd);
output[10] = HIGHBD_WRAPLOW(x15, bd);
output[11] = HIGHBD_WRAPLOW(x7, bd);
output[12] = HIGHBD_WRAPLOW(x5, bd);
output[13] = HIGHBD_WRAPLOW(-x13, bd);
output[14] = HIGHBD_WRAPLOW(x9, bd);
output[15] = HIGHBD_WRAPLOW(-x1, bd);
}
void aom_highbd_idct32_c(const tran_low_t *input, tran_low_t *output, int bd) {
tran_low_t step1[32], step2[32];
tran_high_t temp1, temp2;
(void)bd;
// stage 1
step1[0] = input[0];
step1[1] = input[16];
step1[2] = input[8];
step1[3] = input[24];
step1[4] = input[4];
step1[5] = input[20];
step1[6] = input[12];
step1[7] = input[28];
step1[8] = input[2];
step1[9] = input[18];
step1[10] = input[10];
step1[11] = input[26];
step1[12] = input[6];
step1[13] = input[22];
step1[14] = input[14];
step1[15] = input[30];
temp1 = input[1] * cospi_31_64 - input[31] * cospi_1_64;
temp2 = input[1] * cospi_1_64 + input[31] * cospi_31_64;
step1[16] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[31] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[17] * cospi_15_64 - input[15] * cospi_17_64;
temp2 = input[17] * cospi_17_64 + input[15] * cospi_15_64;
step1[17] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[30] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[9] * cospi_23_64 - input[23] * cospi_9_64;
temp2 = input[9] * cospi_9_64 + input[23] * cospi_23_64;
step1[18] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[29] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[25] * cospi_7_64 - input[7] * cospi_25_64;
temp2 = input[25] * cospi_25_64 + input[7] * cospi_7_64;
step1[19] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[28] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[5] * cospi_27_64 - input[27] * cospi_5_64;
temp2 = input[5] * cospi_5_64 + input[27] * cospi_27_64;
step1[20] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[27] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[21] * cospi_11_64 - input[11] * cospi_21_64;
temp2 = input[21] * cospi_21_64 + input[11] * cospi_11_64;
step1[21] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[26] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[13] * cospi_19_64 - input[19] * cospi_13_64;
temp2 = input[13] * cospi_13_64 + input[19] * cospi_19_64;
step1[22] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[25] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = input[29] * cospi_3_64 - input[3] * cospi_29_64;
temp2 = input[29] * cospi_29_64 + input[3] * cospi_3_64;
step1[23] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[24] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
// stage 2
step2[0] = step1[0];
step2[1] = step1[1];
step2[2] = step1[2];
step2[3] = step1[3];
step2[4] = step1[4];
step2[5] = step1[5];
step2[6] = step1[6];
step2[7] = step1[7];
temp1 = step1[8] * cospi_30_64 - step1[15] * cospi_2_64;
temp2 = step1[8] * cospi_2_64 + step1[15] * cospi_30_64;
step2[8] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[15] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[9] * cospi_14_64 - step1[14] * cospi_18_64;
temp2 = step1[9] * cospi_18_64 + step1[14] * cospi_14_64;
step2[9] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[14] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[10] * cospi_22_64 - step1[13] * cospi_10_64;
temp2 = step1[10] * cospi_10_64 + step1[13] * cospi_22_64;
step2[10] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[13] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[11] * cospi_6_64 - step1[12] * cospi_26_64;
temp2 = step1[11] * cospi_26_64 + step1[12] * cospi_6_64;
step2[11] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[12] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step2[16] = HIGHBD_WRAPLOW(step1[16] + step1[17], bd);
step2[17] = HIGHBD_WRAPLOW(step1[16] - step1[17], bd);
step2[18] = HIGHBD_WRAPLOW(-step1[18] + step1[19], bd);
step2[19] = HIGHBD_WRAPLOW(step1[18] + step1[19], bd);
step2[20] = HIGHBD_WRAPLOW(step1[20] + step1[21], bd);
step2[21] = HIGHBD_WRAPLOW(step1[20] - step1[21], bd);
step2[22] = HIGHBD_WRAPLOW(-step1[22] + step1[23], bd);
step2[23] = HIGHBD_WRAPLOW(step1[22] + step1[23], bd);
step2[24] = HIGHBD_WRAPLOW(step1[24] + step1[25], bd);
step2[25] = HIGHBD_WRAPLOW(step1[24] - step1[25], bd);
step2[26] = HIGHBD_WRAPLOW(-step1[26] + step1[27], bd);
step2[27] = HIGHBD_WRAPLOW(step1[26] + step1[27], bd);
step2[28] = HIGHBD_WRAPLOW(step1[28] + step1[29], bd);
step2[29] = HIGHBD_WRAPLOW(step1[28] - step1[29], bd);
step2[30] = HIGHBD_WRAPLOW(-step1[30] + step1[31], bd);
step2[31] = HIGHBD_WRAPLOW(step1[30] + step1[31], bd);
// stage 3
step1[0] = step2[0];
step1[1] = step2[1];
step1[2] = step2[2];
step1[3] = step2[3];
temp1 = step2[4] * cospi_28_64 - step2[7] * cospi_4_64;
temp2 = step2[4] * cospi_4_64 + step2[7] * cospi_28_64;
step1[4] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[7] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step2[5] * cospi_12_64 - step2[6] * cospi_20_64;
temp2 = step2[5] * cospi_20_64 + step2[6] * cospi_12_64;
step1[5] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[6] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[8] = HIGHBD_WRAPLOW(step2[8] + step2[9], bd);
step1[9] = HIGHBD_WRAPLOW(step2[8] - step2[9], bd);
step1[10] = HIGHBD_WRAPLOW(-step2[10] + step2[11], bd);
step1[11] = HIGHBD_WRAPLOW(step2[10] + step2[11], bd);
step1[12] = HIGHBD_WRAPLOW(step2[12] + step2[13], bd);
step1[13] = HIGHBD_WRAPLOW(step2[12] - step2[13], bd);
step1[14] = HIGHBD_WRAPLOW(-step2[14] + step2[15], bd);
step1[15] = HIGHBD_WRAPLOW(step2[14] + step2[15], bd);
step1[16] = step2[16];
step1[31] = step2[31];
temp1 = -step2[17] * cospi_4_64 + step2[30] * cospi_28_64;
temp2 = step2[17] * cospi_28_64 + step2[30] * cospi_4_64;
step1[17] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[30] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = -step2[18] * cospi_28_64 - step2[29] * cospi_4_64;
temp2 = -step2[18] * cospi_4_64 + step2[29] * cospi_28_64;
step1[18] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[29] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[19] = step2[19];
step1[20] = step2[20];
temp1 = -step2[21] * cospi_20_64 + step2[26] * cospi_12_64;
temp2 = step2[21] * cospi_12_64 + step2[26] * cospi_20_64;
step1[21] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[26] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = -step2[22] * cospi_12_64 - step2[25] * cospi_20_64;
temp2 = -step2[22] * cospi_20_64 + step2[25] * cospi_12_64;
step1[22] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[25] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[23] = step2[23];
step1[24] = step2[24];
step1[27] = step2[27];
step1[28] = step2[28];
// stage 4
temp1 = (step1[0] + step1[1]) * cospi_16_64;
temp2 = (step1[0] - step1[1]) * cospi_16_64;
step2[0] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[1] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = step1[2] * cospi_24_64 - step1[3] * cospi_8_64;
temp2 = step1[2] * cospi_8_64 + step1[3] * cospi_24_64;
step2[2] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[3] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step2[4] = HIGHBD_WRAPLOW(step1[4] + step1[5], bd);
step2[5] = HIGHBD_WRAPLOW(step1[4] - step1[5], bd);
step2[6] = HIGHBD_WRAPLOW(-step1[6] + step1[7], bd);
step2[7] = HIGHBD_WRAPLOW(step1[6] + step1[7], bd);
step2[8] = step1[8];
step2[15] = step1[15];
temp1 = -step1[9] * cospi_8_64 + step1[14] * cospi_24_64;
temp2 = step1[9] * cospi_24_64 + step1[14] * cospi_8_64;
step2[9] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[14] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = -step1[10] * cospi_24_64 - step1[13] * cospi_8_64;
temp2 = -step1[10] * cospi_8_64 + step1[13] * cospi_24_64;
step2[10] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[13] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step2[11] = step1[11];
step2[12] = step1[12];
step2[16] = HIGHBD_WRAPLOW(step1[16] + step1[19], bd);
step2[17] = HIGHBD_WRAPLOW(step1[17] + step1[18], bd);
step2[18] = HIGHBD_WRAPLOW(step1[17] - step1[18], bd);
step2[19] = HIGHBD_WRAPLOW(step1[16] - step1[19], bd);
step2[20] = HIGHBD_WRAPLOW(-step1[20] + step1[23], bd);
step2[21] = HIGHBD_WRAPLOW(-step1[21] + step1[22], bd);
step2[22] = HIGHBD_WRAPLOW(step1[21] + step1[22], bd);
step2[23] = HIGHBD_WRAPLOW(step1[20] + step1[23], bd);
step2[24] = HIGHBD_WRAPLOW(step1[24] + step1[27], bd);
step2[25] = HIGHBD_WRAPLOW(step1[25] + step1[26], bd);
step2[26] = HIGHBD_WRAPLOW(step1[25] - step1[26], bd);
step2[27] = HIGHBD_WRAPLOW(step1[24] - step1[27], bd);
step2[28] = HIGHBD_WRAPLOW(-step1[28] + step1[31], bd);
step2[29] = HIGHBD_WRAPLOW(-step1[29] + step1[30], bd);
step2[30] = HIGHBD_WRAPLOW(step1[29] + step1[30], bd);
step2[31] = HIGHBD_WRAPLOW(step1[28] + step1[31], bd);
// stage 5
step1[0] = HIGHBD_WRAPLOW(step2[0] + step2[3], bd);
step1[1] = HIGHBD_WRAPLOW(step2[1] + step2[2], bd);
step1[2] = HIGHBD_WRAPLOW(step2[1] - step2[2], bd);
step1[3] = HIGHBD_WRAPLOW(step2[0] - step2[3], bd);
step1[4] = step2[4];
temp1 = (step2[6] - step2[5]) * cospi_16_64;
temp2 = (step2[5] + step2[6]) * cospi_16_64;
step1[5] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[6] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[7] = step2[7];
step1[8] = HIGHBD_WRAPLOW(step2[8] + step2[11], bd);
step1[9] = HIGHBD_WRAPLOW(step2[9] + step2[10], bd);
step1[10] = HIGHBD_WRAPLOW(step2[9] - step2[10], bd);
step1[11] = HIGHBD_WRAPLOW(step2[8] - step2[11], bd);
step1[12] = HIGHBD_WRAPLOW(-step2[12] + step2[15], bd);
step1[13] = HIGHBD_WRAPLOW(-step2[13] + step2[14], bd);
step1[14] = HIGHBD_WRAPLOW(step2[13] + step2[14], bd);
step1[15] = HIGHBD_WRAPLOW(step2[12] + step2[15], bd);
step1[16] = step2[16];
step1[17] = step2[17];
temp1 = -step2[18] * cospi_8_64 + step2[29] * cospi_24_64;
temp2 = step2[18] * cospi_24_64 + step2[29] * cospi_8_64;
step1[18] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[29] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = -step2[19] * cospi_8_64 + step2[28] * cospi_24_64;
temp2 = step2[19] * cospi_24_64 + step2[28] * cospi_8_64;
step1[19] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[28] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = -step2[20] * cospi_24_64 - step2[27] * cospi_8_64;
temp2 = -step2[20] * cospi_8_64 + step2[27] * cospi_24_64;
step1[20] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[27] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = -step2[21] * cospi_24_64 - step2[26] * cospi_8_64;
temp2 = -step2[21] * cospi_8_64 + step2[26] * cospi_24_64;
step1[21] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[26] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[22] = step2[22];
step1[23] = step2[23];
step1[24] = step2[24];
step1[25] = step2[25];
step1[30] = step2[30];
step1[31] = step2[31];
// stage 6
step2[0] = HIGHBD_WRAPLOW(step1[0] + step1[7], bd);
step2[1] = HIGHBD_WRAPLOW(step1[1] + step1[6], bd);
step2[2] = HIGHBD_WRAPLOW(step1[2] + step1[5], bd);
step2[3] = HIGHBD_WRAPLOW(step1[3] + step1[4], bd);
step2[4] = HIGHBD_WRAPLOW(step1[3] - step1[4], bd);
step2[5] = HIGHBD_WRAPLOW(step1[2] - step1[5], bd);
step2[6] = HIGHBD_WRAPLOW(step1[1] - step1[6], bd);
step2[7] = HIGHBD_WRAPLOW(step1[0] - step1[7], bd);
step2[8] = step1[8];
step2[9] = step1[9];
temp1 = (-step1[10] + step1[13]) * cospi_16_64;
temp2 = (step1[10] + step1[13]) * cospi_16_64;
step2[10] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[13] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = (-step1[11] + step1[12]) * cospi_16_64;
temp2 = (step1[11] + step1[12]) * cospi_16_64;
step2[11] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step2[12] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step2[14] = step1[14];
step2[15] = step1[15];
step2[16] = HIGHBD_WRAPLOW(step1[16] + step1[23], bd);
step2[17] = HIGHBD_WRAPLOW(step1[17] + step1[22], bd);
step2[18] = HIGHBD_WRAPLOW(step1[18] + step1[21], bd);
step2[19] = HIGHBD_WRAPLOW(step1[19] + step1[20], bd);
step2[20] = HIGHBD_WRAPLOW(step1[19] - step1[20], bd);
step2[21] = HIGHBD_WRAPLOW(step1[18] - step1[21], bd);
step2[22] = HIGHBD_WRAPLOW(step1[17] - step1[22], bd);
step2[23] = HIGHBD_WRAPLOW(step1[16] - step1[23], bd);
step2[24] = HIGHBD_WRAPLOW(-step1[24] + step1[31], bd);
step2[25] = HIGHBD_WRAPLOW(-step1[25] + step1[30], bd);
step2[26] = HIGHBD_WRAPLOW(-step1[26] + step1[29], bd);
step2[27] = HIGHBD_WRAPLOW(-step1[27] + step1[28], bd);
step2[28] = HIGHBD_WRAPLOW(step1[27] + step1[28], bd);
step2[29] = HIGHBD_WRAPLOW(step1[26] + step1[29], bd);
step2[30] = HIGHBD_WRAPLOW(step1[25] + step1[30], bd);
step2[31] = HIGHBD_WRAPLOW(step1[24] + step1[31], bd);
// stage 7
step1[0] = HIGHBD_WRAPLOW(step2[0] + step2[15], bd);
step1[1] = HIGHBD_WRAPLOW(step2[1] + step2[14], bd);
step1[2] = HIGHBD_WRAPLOW(step2[2] + step2[13], bd);
step1[3] = HIGHBD_WRAPLOW(step2[3] + step2[12], bd);
step1[4] = HIGHBD_WRAPLOW(step2[4] + step2[11], bd);
step1[5] = HIGHBD_WRAPLOW(step2[5] + step2[10], bd);
step1[6] = HIGHBD_WRAPLOW(step2[6] + step2[9], bd);
step1[7] = HIGHBD_WRAPLOW(step2[7] + step2[8], bd);
step1[8] = HIGHBD_WRAPLOW(step2[7] - step2[8], bd);
step1[9] = HIGHBD_WRAPLOW(step2[6] - step2[9], bd);
step1[10] = HIGHBD_WRAPLOW(step2[5] - step2[10], bd);
step1[11] = HIGHBD_WRAPLOW(step2[4] - step2[11], bd);
step1[12] = HIGHBD_WRAPLOW(step2[3] - step2[12], bd);
step1[13] = HIGHBD_WRAPLOW(step2[2] - step2[13], bd);
step1[14] = HIGHBD_WRAPLOW(step2[1] - step2[14], bd);
step1[15] = HIGHBD_WRAPLOW(step2[0] - step2[15], bd);
step1[16] = step2[16];
step1[17] = step2[17];
step1[18] = step2[18];
step1[19] = step2[19];
temp1 = (-step2[20] + step2[27]) * cospi_16_64;
temp2 = (step2[20] + step2[27]) * cospi_16_64;
step1[20] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[27] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = (-step2[21] + step2[26]) * cospi_16_64;
temp2 = (step2[21] + step2[26]) * cospi_16_64;
step1[21] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[26] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = (-step2[22] + step2[25]) * cospi_16_64;
temp2 = (step2[22] + step2[25]) * cospi_16_64;
step1[22] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[25] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
temp1 = (-step2[23] + step2[24]) * cospi_16_64;
temp2 = (step2[23] + step2[24]) * cospi_16_64;
step1[23] = HIGHBD_WRAPLOW(dct_const_round_shift(temp1), bd);
step1[24] = HIGHBD_WRAPLOW(dct_const_round_shift(temp2), bd);
step1[28] = step2[28];
step1[29] = step2[29];
step1[30] = step2[30];
step1[31] = step2[31];
// final stage
output[0] = HIGHBD_WRAPLOW(step1[0] + step1[31], bd);
output[1] = HIGHBD_WRAPLOW(step1[1] + step1[30], bd);
output[2] = HIGHBD_WRAPLOW(step1[2] + step1[29], bd);
output[3] = HIGHBD_WRAPLOW(step1[3] + step1[28], bd);
output[4] = HIGHBD_WRAPLOW(step1[4] + step1[27], bd);
output[5] = HIGHBD_WRAPLOW(step1[5] + step1[26], bd);
output[6] = HIGHBD_WRAPLOW(step1[6] + step1[25], bd);
output[7] = HIGHBD_WRAPLOW(step1[7] + step1[24], bd);
output[8] = HIGHBD_WRAPLOW(step1[8] + step1[23], bd);
output[9] = HIGHBD_WRAPLOW(step1[9] + step1[22], bd);
output[10] = HIGHBD_WRAPLOW(step1[10] + step1[21], bd);
output[11] = HIGHBD_WRAPLOW(step1[11] + step1[20], bd);
output[12] = HIGHBD_WRAPLOW(step1[12] + step1[19], bd);
output[13] = HIGHBD_WRAPLOW(step1[13] + step1[18], bd);
output[14] = HIGHBD_WRAPLOW(step1[14] + step1[17], bd);
output[15] = HIGHBD_WRAPLOW(step1[15] + step1[16], bd);
output[16] = HIGHBD_WRAPLOW(step1[15] - step1[16], bd);
output[17] = HIGHBD_WRAPLOW(step1[14] - step1[17], bd);
output[18] = HIGHBD_WRAPLOW(step1[13] - step1[18], bd);
output[19] = HIGHBD_WRAPLOW(step1[12] - step1[19], bd);
output[20] = HIGHBD_WRAPLOW(step1[11] - step1[20], bd);
output[21] = HIGHBD_WRAPLOW(step1[10] - step1[21], bd);
output[22] = HIGHBD_WRAPLOW(step1[9] - step1[22], bd);
output[23] = HIGHBD_WRAPLOW(step1[8] - step1[23], bd);
output[24] = HIGHBD_WRAPLOW(step1[7] - step1[24], bd);
output[25] = HIGHBD_WRAPLOW(step1[6] - step1[25], bd);
output[26] = HIGHBD_WRAPLOW(step1[5] - step1[26], bd);
output[27] = HIGHBD_WRAPLOW(step1[4] - step1[27], bd);
output[28] = HIGHBD_WRAPLOW(step1[3] - step1[28], bd);
output[29] = HIGHBD_WRAPLOW(step1[2] - step1[29], bd);
output[30] = HIGHBD_WRAPLOW(step1[1] - step1[30], bd);
output[31] = HIGHBD_WRAPLOW(step1[0] - step1[31], bd);
}
#endif // CONFIG_HIGHBITDEPTH #endif // CONFIG_HIGHBITDEPTH

110
third_party/aom/aom_dsp/loopfilter.c vendored
View file

@ -149,10 +149,15 @@ void aom_lpf_horizontal_4_c(uint8_t *s, int p /* pitch */,
const uint8_t *blimit, const uint8_t *limit, const uint8_t *blimit, const uint8_t *limit,
const uint8_t *thresh) { const uint8_t *thresh) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
#if !CONFIG_PARALLEL_DEBLOCKING #if !CONFIG_PARALLEL_DEBLOCKING
const uint8_t p3 = s[-4 * p], p2 = s[-3 * p], p1 = s[-2 * p], p0 = s[-p]; const uint8_t p3 = s[-4 * p], p2 = s[-3 * p], p1 = s[-2 * p], p0 = s[-p];
const uint8_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p]; const uint8_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p];
@ -179,10 +184,15 @@ void aom_lpf_horizontal_4_dual_c(uint8_t *s, int p, const uint8_t *blimit0,
void aom_lpf_vertical_4_c(uint8_t *s, int pitch, const uint8_t *blimit, void aom_lpf_vertical_4_c(uint8_t *s, int pitch, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh) { const uint8_t *limit, const uint8_t *thresh) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
#if !CONFIG_PARALLEL_DEBLOCKING #if !CONFIG_PARALLEL_DEBLOCKING
const uint8_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1]; const uint8_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1];
const uint8_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3]; const uint8_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3];
@ -206,7 +216,7 @@ void aom_lpf_vertical_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0,
aom_lpf_vertical_4_c(s + 8 * pitch, pitch, blimit1, limit1, thresh1); aom_lpf_vertical_4_c(s + 8 * pitch, pitch, blimit1, limit1, thresh1);
} }
static INLINE void filter8(int8_t mask, uint8_t thresh, uint8_t flat, static INLINE void filter8(int8_t mask, uint8_t thresh, int8_t flat,
uint8_t *op3, uint8_t *op2, uint8_t *op1, uint8_t *op3, uint8_t *op2, uint8_t *op1,
uint8_t *op0, uint8_t *oq0, uint8_t *oq1, uint8_t *op0, uint8_t *oq0, uint8_t *oq1,
uint8_t *oq2, uint8_t *oq3) { uint8_t *oq2, uint8_t *oq3) {
@ -229,10 +239,15 @@ static INLINE void filter8(int8_t mask, uint8_t thresh, uint8_t flat,
void aom_lpf_horizontal_8_c(uint8_t *s, int p, const uint8_t *blimit, void aom_lpf_horizontal_8_c(uint8_t *s, int p, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh) { const uint8_t *limit, const uint8_t *thresh) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
const uint8_t p3 = s[-4 * p], p2 = s[-3 * p], p1 = s[-2 * p], p0 = s[-p]; const uint8_t p3 = s[-4 * p], p2 = s[-3 * p], p1 = s[-2 * p], p0 = s[-p];
const uint8_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p]; const uint8_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p];
@ -256,8 +271,13 @@ void aom_lpf_horizontal_8_dual_c(uint8_t *s, int p, const uint8_t *blimit0,
void aom_lpf_vertical_8_c(uint8_t *s, int pitch, const uint8_t *blimit, void aom_lpf_vertical_8_c(uint8_t *s, int pitch, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh) { const uint8_t *limit, const uint8_t *thresh) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
const uint8_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1]; const uint8_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1];
const uint8_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3]; const uint8_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3];
const int8_t mask = const int8_t mask =
@ -278,8 +298,8 @@ void aom_lpf_vertical_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0,
} }
#if PARALLEL_DEBLOCKING_11_TAP #if PARALLEL_DEBLOCKING_11_TAP
static INLINE void filter12(int8_t mask, uint8_t thresh, uint8_t flat, static INLINE void filter12(int8_t mask, uint8_t thresh, int8_t flat,
uint8_t flat2, uint8_t *op5, uint8_t *op4, int8_t flat2, uint8_t *op5, uint8_t *op4,
uint8_t *op3, uint8_t *op2, uint8_t *op1, uint8_t *op3, uint8_t *op2, uint8_t *op1,
uint8_t *op0, uint8_t *oq0, uint8_t *oq1, uint8_t *op0, uint8_t *oq0, uint8_t *oq1,
uint8_t *oq2, uint8_t *oq3, uint8_t *oq4, uint8_t *oq2, uint8_t *oq3, uint8_t *oq4,
@ -308,8 +328,8 @@ static INLINE void filter12(int8_t mask, uint8_t thresh, uint8_t flat,
#endif #endif
#if PARALLEL_DEBLOCKING_9_TAP #if PARALLEL_DEBLOCKING_9_TAP
static INLINE void filter10(int8_t mask, uint8_t thresh, uint8_t flat, static INLINE void filter10(int8_t mask, uint8_t thresh, int8_t flat,
uint8_t flat2, uint8_t *op4, uint8_t *op3, int8_t flat2, uint8_t *op4, uint8_t *op3,
uint8_t *op2, uint8_t *op1, uint8_t *op0, uint8_t *op2, uint8_t *op1, uint8_t *op0,
uint8_t *oq0, uint8_t *oq1, uint8_t *oq2, uint8_t *oq0, uint8_t *oq1, uint8_t *oq2,
uint8_t *oq3, uint8_t *oq4) { uint8_t *oq3, uint8_t *oq4) {
@ -332,8 +352,8 @@ static INLINE void filter10(int8_t mask, uint8_t thresh, uint8_t flat,
} }
#endif #endif
static INLINE void filter16(int8_t mask, uint8_t thresh, uint8_t flat, static INLINE void filter16(int8_t mask, uint8_t thresh, int8_t flat,
uint8_t flat2, uint8_t *op7, uint8_t *op6, int8_t flat2, uint8_t *op7, uint8_t *op6,
uint8_t *op5, uint8_t *op4, uint8_t *op3, uint8_t *op5, uint8_t *op4, uint8_t *op3,
uint8_t *op2, uint8_t *op1, uint8_t *op0, uint8_t *op2, uint8_t *op1, uint8_t *op0,
uint8_t *oq0, uint8_t *oq1, uint8_t *oq2, uint8_t *oq0, uint8_t *oq1, uint8_t *oq2,
@ -390,10 +410,15 @@ static void mb_lpf_horizontal_edge_w(uint8_t *s, int p, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *limit,
const uint8_t *thresh, int count) { const uint8_t *thresh, int count) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int step = 4;
#else
int step = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8 * count; ++i) { for (i = 0; i < step * count; ++i) {
const uint8_t p7 = s[-8 * p], p6 = s[-7 * p], p5 = s[-6 * p], const uint8_t p7 = s[-8 * p], p6 = s[-7 * p], p5 = s[-6 * p],
p4 = s[-5 * p], p3 = s[-4 * p], p2 = s[-3 * p], p4 = s[-5 * p], p3 = s[-4 * p], p2 = s[-3 * p],
p1 = s[-2 * p], p0 = s[-p]; p1 = s[-2 * p], p0 = s[-p];
@ -436,7 +461,11 @@ void aom_lpf_horizontal_edge_8_c(uint8_t *s, int p, const uint8_t *blimit,
void aom_lpf_horizontal_edge_16_c(uint8_t *s, int p, const uint8_t *blimit, void aom_lpf_horizontal_edge_16_c(uint8_t *s, int p, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh) { const uint8_t *limit, const uint8_t *thresh) {
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 1);
#else
mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 2); mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 2);
#endif
} }
static void mb_lpf_vertical_edge_w(uint8_t *s, int p, const uint8_t *blimit, static void mb_lpf_vertical_edge_w(uint8_t *s, int p, const uint8_t *blimit,
@ -478,7 +507,11 @@ static void mb_lpf_vertical_edge_w(uint8_t *s, int p, const uint8_t *blimit,
void aom_lpf_vertical_16_c(uint8_t *s, int p, const uint8_t *blimit, void aom_lpf_vertical_16_c(uint8_t *s, int p, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh) { const uint8_t *limit, const uint8_t *thresh) {
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 4);
#else
mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 8); mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 8);
#endif
} }
void aom_lpf_vertical_16_dual_c(uint8_t *s, int p, const uint8_t *blimit, void aom_lpf_vertical_16_dual_c(uint8_t *s, int p, const uint8_t *blimit,
@ -596,10 +629,15 @@ void aom_highbd_lpf_horizontal_4_c(uint16_t *s, int p /* pitch */,
const uint8_t *blimit, const uint8_t *limit, const uint8_t *blimit, const uint8_t *limit,
const uint8_t *thresh, int bd) { const uint8_t *thresh, int bd) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
#if !CONFIG_PARALLEL_DEBLOCKING #if !CONFIG_PARALLEL_DEBLOCKING
const uint16_t p3 = s[-4 * p]; const uint16_t p3 = s[-4 * p];
const uint16_t p2 = s[-3 * p]; const uint16_t p2 = s[-3 * p];
@ -636,10 +674,15 @@ void aom_highbd_lpf_vertical_4_c(uint16_t *s, int pitch, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh, const uint8_t *limit, const uint8_t *thresh,
int bd) { int bd) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
#if !CONFIG_PARALLEL_DEBLOCKING #if !CONFIG_PARALLEL_DEBLOCKING
const uint16_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1]; const uint16_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1];
const uint16_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3]; const uint16_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3];
@ -665,7 +708,7 @@ void aom_highbd_lpf_vertical_4_dual_c(
bd); bd);
} }
static INLINE void highbd_filter8(int8_t mask, uint8_t thresh, uint8_t flat, static INLINE void highbd_filter8(int8_t mask, uint8_t thresh, int8_t flat,
uint16_t *op3, uint16_t *op2, uint16_t *op1, uint16_t *op3, uint16_t *op2, uint16_t *op1,
uint16_t *op0, uint16_t *oq0, uint16_t *oq1, uint16_t *op0, uint16_t *oq0, uint16_t *oq1,
uint16_t *oq2, uint16_t *oq3, int bd) { uint16_t *oq2, uint16_t *oq3, int bd) {
@ -689,10 +732,15 @@ void aom_highbd_lpf_horizontal_8_c(uint16_t *s, int p, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh, const uint8_t *limit, const uint8_t *thresh,
int bd) { int bd) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
const uint16_t p3 = s[-4 * p], p2 = s[-3 * p], p1 = s[-2 * p], p0 = s[-p]; const uint16_t p3 = s[-4 * p], p2 = s[-3 * p], p1 = s[-2 * p], p0 = s[-p];
const uint16_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p]; const uint16_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p];
@ -718,8 +766,13 @@ void aom_highbd_lpf_vertical_8_c(uint16_t *s, int pitch, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh, const uint8_t *limit, const uint8_t *thresh,
int bd) { int bd) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int count = 4;
#else
int count = 8;
#endif
for (i = 0; i < 8; ++i) { for (i = 0; i < count; ++i) {
const uint16_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1]; const uint16_t p3 = s[-4], p2 = s[-3], p1 = s[-2], p0 = s[-1];
const uint16_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3]; const uint16_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3];
const int8_t mask = const int8_t mask =
@ -741,8 +794,8 @@ void aom_highbd_lpf_vertical_8_dual_c(
bd); bd);
} }
static INLINE void highbd_filter16(int8_t mask, uint8_t thresh, uint8_t flat, static INLINE void highbd_filter16(int8_t mask, uint8_t thresh, int8_t flat,
uint8_t flat2, uint16_t *op7, uint16_t *op6, int8_t flat2, uint16_t *op7, uint16_t *op6,
uint16_t *op5, uint16_t *op4, uint16_t *op3, uint16_t *op5, uint16_t *op4, uint16_t *op3,
uint16_t *op2, uint16_t *op1, uint16_t *op0, uint16_t *op2, uint16_t *op1, uint16_t *op0,
uint16_t *oq0, uint16_t *oq1, uint16_t *oq2, uint16_t *oq0, uint16_t *oq1, uint16_t *oq2,
@ -813,10 +866,15 @@ static void highbd_mb_lpf_horizontal_edge_w(uint16_t *s, int p,
const uint8_t *thresh, int count, const uint8_t *thresh, int count,
int bd) { int bd) {
int i; int i;
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
int step = 4;
#else
int step = 8;
#endif
// loop filter designed to work using chars so that we can make maximum use // loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions. // of 8 bit simd instructions.
for (i = 0; i < 8 * count; ++i) { for (i = 0; i < step * count; ++i) {
const uint16_t p3 = s[-4 * p]; const uint16_t p3 = s[-4 * p];
const uint16_t p2 = s[-3 * p]; const uint16_t p2 = s[-3 * p];
const uint16_t p1 = s[-2 * p]; const uint16_t p1 = s[-2 * p];
@ -852,7 +910,11 @@ void aom_highbd_lpf_horizontal_edge_16_c(uint16_t *s, int p,
const uint8_t *blimit, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *limit,
const uint8_t *thresh, int bd) { const uint8_t *thresh, int bd) {
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
highbd_mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 1, bd);
#else
highbd_mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 2, bd); highbd_mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 2, bd);
#endif
} }
static void highbd_mb_lpf_vertical_edge_w(uint16_t *s, int p, static void highbd_mb_lpf_vertical_edge_w(uint16_t *s, int p,
@ -888,13 +950,21 @@ static void highbd_mb_lpf_vertical_edge_w(uint16_t *s, int p,
void aom_highbd_lpf_vertical_16_c(uint16_t *s, int p, const uint8_t *blimit, void aom_highbd_lpf_vertical_16_c(uint16_t *s, int p, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh, const uint8_t *limit, const uint8_t *thresh,
int bd) { int bd) {
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
highbd_mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 4, bd);
#else
highbd_mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 8, bd); highbd_mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 8, bd);
#endif
} }
void aom_highbd_lpf_vertical_16_dual_c(uint16_t *s, int p, void aom_highbd_lpf_vertical_16_dual_c(uint16_t *s, int p,
const uint8_t *blimit, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *limit,
const uint8_t *thresh, int bd) { const uint8_t *thresh, int bd) {
#if CONFIG_PARALLEL_DEBLOCKING && CONFIG_CB4X4
highbd_mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 8, bd);
#else
highbd_mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 16, bd); highbd_mb_lpf_vertical_edge_w(s, p, blimit, limit, thresh, 16, bd);
#endif
} }
#endif // CONFIG_HIGHBITDEPTH #endif // CONFIG_HIGHBITDEPTH

57
third_party/aom/aom_dsp/mips/avg_msa.c vendored
View file

@ -1,57 +0,0 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include "./aom_dsp_rtcd.h"
#include "aom_dsp/mips/macros_msa.h"
uint32_t aom_avg_8x8_msa(const uint8_t *src, int32_t src_stride) {
uint32_t sum_out;
v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
v8u16 sum0, sum1, sum2, sum3, sum4, sum5, sum6, sum7;
v4u32 sum = { 0 };
LD_UB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
HADD_UB4_UH(src0, src1, src2, src3, sum0, sum1, sum2, sum3);
HADD_UB4_UH(src4, src5, src6, src7, sum4, sum5, sum6, sum7);
ADD4(sum0, sum1, sum2, sum3, sum4, sum5, sum6, sum7, sum0, sum2, sum4, sum6);
ADD2(sum0, sum2, sum4, sum6, sum0, sum4);
sum0 += sum4;
sum = __msa_hadd_u_w(sum0, sum0);
sum0 = (v8u16)__msa_pckev_h((v8i16)sum, (v8i16)sum);
sum = __msa_hadd_u_w(sum0, sum0);
sum = (v4u32)__msa_srari_w((v4i32)sum, 6);
sum_out = __msa_copy_u_w((v4i32)sum, 0);
return sum_out;
}
uint32_t aom_avg_4x4_msa(const uint8_t *src, int32_t src_stride) {
uint32_t sum_out;
uint32_t src0, src1, src2, src3;
v16u8 vec = { 0 };
v8u16 sum0;
v4u32 sum1;
v2u64 sum2;
LW4(src, src_stride, src0, src1, src2, src3);
INSERT_W4_UB(src0, src1, src2, src3, vec);
sum0 = __msa_hadd_u_h(vec, vec);
sum1 = __msa_hadd_u_w(sum0, sum0);
sum0 = (v8u16)__msa_pckev_h((v8i16)sum1, (v8i16)sum1);
sum1 = __msa_hadd_u_w(sum0, sum0);
sum2 = __msa_hadd_u_d(sum1, sum1);
sum1 = (v4u32)__msa_srari_w((v4i32)sum2, 4);
sum_out = __msa_copy_u_w((v4i32)sum1, 0);
return sum_out;
}

17
third_party/aom/aom_dsp/prob.c vendored
View file

@ -11,25 +11,10 @@
#include "./aom_config.h" #include "./aom_config.h"
#if CONFIG_EC_MULTISYMBOL
#include <string.h> #include <string.h>
#endif
#include "aom_dsp/prob.h" #include "aom_dsp/prob.h"
const uint8_t aom_norm[256] = {
0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
3, 3, 3, 3, 3, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
};
static unsigned int tree_merge_probs_impl(unsigned int i, static unsigned int tree_merge_probs_impl(unsigned int i,
const aom_tree_index *tree, const aom_tree_index *tree,
const aom_prob *pre_probs, const aom_prob *pre_probs,
@ -53,7 +38,6 @@ void aom_tree_merge_probs(const aom_tree_index *tree, const aom_prob *pre_probs,
tree_merge_probs_impl(0, tree, pre_probs, counts, probs); tree_merge_probs_impl(0, tree, pre_probs, counts, probs);
} }
#if CONFIG_EC_MULTISYMBOL
typedef struct tree_node tree_node; typedef struct tree_node tree_node;
struct tree_node { struct tree_node {
@ -233,4 +217,3 @@ void av1_indices_from_tree(int *ind, int *inv, const aom_tree_index *tree) {
int stack_index = 0; int stack_index = 0;
tree_to_index(&stack_index, ind, inv, tree, 0, 0); tree_to_index(&stack_index, ind, inv, tree, 0, 0);
} }
#endif

12
third_party/aom/aom_dsp/prob.h vendored
View file

@ -20,7 +20,7 @@
#include "aom_ports/bitops.h" #include "aom_ports/bitops.h"
#include "aom_ports/mem.h" #include "aom_ports/mem.h"
#if CONFIG_DAALA_EC #if !CONFIG_ANS
#include "aom_dsp/entcode.h" #include "aom_dsp/entcode.h"
#endif #endif
@ -33,14 +33,12 @@ typedef uint8_t aom_prob;
// TODO(negge): Rename this aom_prob once we remove vpxbool. // TODO(negge): Rename this aom_prob once we remove vpxbool.
typedef uint16_t aom_cdf_prob; typedef uint16_t aom_cdf_prob;
#if CONFIG_EC_MULTISYMBOL
#define CDF_SIZE(x) ((x) + 1) #define CDF_SIZE(x) ((x) + 1)
#endif
#define CDF_PROB_BITS 15 #define CDF_PROB_BITS 15
#define CDF_PROB_TOP (1 << CDF_PROB_BITS) #define CDF_PROB_TOP (1 << CDF_PROB_BITS)
#if CONFIG_DAALA_EC #if !CONFIG_ANS
#define AOM_ICDF OD_ICDF #define AOM_ICDF OD_ICDF
#else #else
#define AOM_ICDF(x) (x) #define AOM_ICDF(x) (x)
@ -117,7 +115,6 @@ static INLINE aom_prob mode_mv_merge_probs(aom_prob pre_prob,
void aom_tree_merge_probs(const aom_tree_index *tree, const aom_prob *pre_probs, void aom_tree_merge_probs(const aom_tree_index *tree, const aom_prob *pre_probs,
const unsigned int *counts, aom_prob *probs); const unsigned int *counts, aom_prob *probs);
#if CONFIG_EC_MULTISYMBOL
int tree_to_cdf(const aom_tree_index *tree, const aom_prob *probs, int tree_to_cdf(const aom_tree_index *tree, const aom_prob *probs,
aom_tree_index root, aom_cdf_prob *cdf, aom_tree_index *ind, aom_tree_index root, aom_cdf_prob *cdf, aom_tree_index *ind,
int *pth, int *len); int *pth, int *len);
@ -150,9 +147,6 @@ static INLINE void av1_tree_to_cdf(const aom_tree_index *tree,
} while (0) } while (0)
void av1_indices_from_tree(int *ind, int *inv, const aom_tree_index *tree); void av1_indices_from_tree(int *ind, int *inv, const aom_tree_index *tree);
#endif
DECLARE_ALIGNED(16, extern const uint8_t, aom_norm[256]);
#if CONFIG_EC_ADAPT #if CONFIG_EC_ADAPT
static INLINE void update_cdf(aom_cdf_prob *cdf, int val, int nsymbs) { static INLINE void update_cdf(aom_cdf_prob *cdf, int val, int nsymbs) {
@ -165,7 +159,7 @@ static INLINE void update_cdf(aom_cdf_prob *cdf, int val, int nsymbs) {
tmp = AOM_ICDF(tmp0); tmp = AOM_ICDF(tmp0);
diff = ((CDF_PROB_TOP - (nsymbs << rate2)) >> rate) << rate; diff = ((CDF_PROB_TOP - (nsymbs << rate2)) >> rate) << rate;
// Single loop (faster) // Single loop (faster)
#if CONFIG_DAALA_EC && CONFIG_EC_SMALLMUL #if !CONFIG_ANS && CONFIG_EC_SMALLMUL
for (i = 0; i < nsymbs - 1; ++i, tmp -= tmp0) { for (i = 0; i < nsymbs - 1; ++i, tmp -= tmp0) {
tmp -= (i == val ? diff : 0); tmp -= (i == val ? diff : 0);
cdf[i] += ((tmp - cdf[i]) >> rate); cdf[i] += ((tmp - cdf[i]) >> rate);

54
third_party/aom/aom_dsp/sad.c vendored
View file

@ -16,6 +16,7 @@
#include "aom/aom_integer.h" #include "aom/aom_integer.h"
#include "aom_ports/mem.h" #include "aom_ports/mem.h"
#include "aom_dsp/blend.h"
/* Sum the difference between every corresponding element of the buffers. */ /* Sum the difference between every corresponding element of the buffers. */
static INLINE unsigned int sad(const uint8_t *a, int a_stride, const uint8_t *b, static INLINE unsigned int sad(const uint8_t *a, int a_stride, const uint8_t *b,
@ -311,15 +312,20 @@ highbd_sadMxNx4D(4, 4)
#if CONFIG_AV1 && CONFIG_EXT_INTER #if CONFIG_AV1 && CONFIG_EXT_INTER
static INLINE static INLINE
unsigned int masked_sad(const uint8_t *a, int a_stride, const uint8_t *b, unsigned int masked_sad(const uint8_t *src, int src_stride,
const uint8_t *a, int a_stride, const uint8_t *b,
int b_stride, const uint8_t *m, int m_stride, int b_stride, const uint8_t *m, int m_stride,
int width, int height) { int width, int height) {
int y, x; int y, x;
unsigned int sad = 0; unsigned int sad = 0;
for (y = 0; y < height; y++) { for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) sad += m[x] * abs(a[x] - b[x]); for (x = 0; x < width; x++) {
const uint8_t pred = AOM_BLEND_A64(m[x], a[x], b[x]);
sad += abs(pred - src[x]);
}
src += src_stride;
a += a_stride; a += a_stride;
b += b_stride; b += b_stride;
m += m_stride; m += m_stride;
@ -329,12 +335,17 @@ highbd_sadMxNx4D(4, 4)
return sad; return sad;
} }
#define MASKSADMxN(m, n) \ #define MASKSADMxN(m, n) \
unsigned int aom_masked_sad##m##x##n##_c( \ unsigned int aom_masked_sad##m##x##n##_c( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \ const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
const uint8_t *msk, int msk_stride) { \ const uint8_t *second_pred, const uint8_t *msk, int msk_stride, \
return masked_sad(src, src_stride, ref, ref_stride, msk, msk_stride, m, \ int invert_mask) { \
n); \ if (!invert_mask) \
return masked_sad(src, src_stride, ref, ref_stride, second_pred, m, msk, \
msk_stride, m, n); \
else \
return masked_sad(src, src_stride, second_pred, m, ref, ref_stride, msk, \
msk_stride, m, n); \
} }
/* clang-format off */ /* clang-format off */
@ -360,18 +371,24 @@ MASKSADMxN(4, 4)
#if CONFIG_HIGHBITDEPTH #if CONFIG_HIGHBITDEPTH
static INLINE static INLINE
unsigned int highbd_masked_sad(const uint8_t *a8, int a_stride, unsigned int highbd_masked_sad(const uint8_t *src8, int src_stride,
const uint8_t *a8, int a_stride,
const uint8_t *b8, int b_stride, const uint8_t *b8, int b_stride,
const uint8_t *m, int m_stride, int width, const uint8_t *m, int m_stride, int width,
int height) { int height) {
int y, x; int y, x;
unsigned int sad = 0; unsigned int sad = 0;
const uint16_t *src = CONVERT_TO_SHORTPTR(src8);
const uint16_t *a = CONVERT_TO_SHORTPTR(a8); const uint16_t *a = CONVERT_TO_SHORTPTR(a8);
const uint16_t *b = CONVERT_TO_SHORTPTR(b8); const uint16_t *b = CONVERT_TO_SHORTPTR(b8);
for (y = 0; y < height; y++) { for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) sad += m[x] * abs(a[x] - b[x]); for (x = 0; x < width; x++) {
const uint16_t pred = AOM_BLEND_A64(m[x], a[x], b[x]);
sad += abs(pred - src[x]);
}
src += src_stride;
a += a_stride; a += a_stride;
b += b_stride; b += b_stride;
m += m_stride; m += m_stride;
@ -381,12 +398,17 @@ MASKSADMxN(4, 4)
return sad; return sad;
} }
#define HIGHBD_MASKSADMXN(m, n) \ #define HIGHBD_MASKSADMXN(m, n) \
unsigned int aom_highbd_masked_sad##m##x##n##_c( \ unsigned int aom_highbd_masked_sad##m##x##n##_c( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \ const uint8_t *src8, int src_stride, const uint8_t *ref8, \
const uint8_t *msk, int msk_stride) { \ int ref_stride, const uint8_t *second_pred8, const uint8_t *msk, \
return highbd_masked_sad(src, src_stride, ref, ref_stride, msk, \ int msk_stride, int invert_mask) { \
msk_stride, m, n); \ if (!invert_mask) \
return highbd_masked_sad(src8, src_stride, ref8, ref_stride, \
second_pred8, m, msk, msk_stride, m, n); \
else \
return highbd_masked_sad(src8, src_stride, second_pred8, m, ref8, \
ref_stride, msk, msk_stride, m, n); \
} }
#if CONFIG_EXT_PARTITION #if CONFIG_EXT_PARTITION

4
third_party/aom/aom_dsp/simd/v64_intrinsics.h vendored
View file

@ -60,7 +60,9 @@ SIMD_INLINE void v64_store_aligned(void *p, v64 a) {
c_v64_store_aligned(p, a); c_v64_store_aligned(p, a);
} }
SIMD_INLINE v64 v64_align(v64 a, v64 b, c) { return c_v64_align(a, b, c); } SIMD_INLINE v64 v64_align(v64 a, v64 b, unsigned int c) {
return c_v64_align(a, b, c);
}
SIMD_INLINE v64 v64_zero() { return c_v64_zero(); } SIMD_INLINE v64 v64_zero() { return c_v64_zero(); }
SIMD_INLINE v64 v64_dup_8(uint8_t x) { return c_v64_dup_8(x); } SIMD_INLINE v64 v64_dup_8(uint8_t x) { return c_v64_dup_8(x); }

367
third_party/aom/aom_dsp/variance.c vendored
View file

@ -18,6 +18,7 @@
#include "aom_dsp/variance.h" #include "aom_dsp/variance.h"
#include "aom_dsp/aom_filter.h" #include "aom_dsp/aom_filter.h"
#include "aom_dsp/blend.h"
uint32_t aom_get4x4sse_cs_c(const uint8_t *a, int a_stride, const uint8_t *b, uint32_t aom_get4x4sse_cs_c(const uint8_t *a, int a_stride, const uint8_t *b,
int b_stride) { int b_stride) {
@ -672,297 +673,215 @@ void aom_highbd_comp_avg_upsampled_pred_c(uint16_t *comp_pred,
#endif // CONFIG_HIGHBITDEPTH #endif // CONFIG_HIGHBITDEPTH
#if CONFIG_AV1 && CONFIG_EXT_INTER #if CONFIG_AV1 && CONFIG_EXT_INTER
void masked_variance(const uint8_t *a, int a_stride, const uint8_t *b, void aom_comp_mask_pred_c(uint8_t *comp_pred, const uint8_t *pred, int width,
int b_stride, const uint8_t *m, int m_stride, int w, int h, int height, const uint8_t *ref, int ref_stride,
unsigned int *sse, int *sum) { const uint8_t *mask, int mask_stride,
int invert_mask) {
int i, j; int i, j;
int64_t sum64 = 0; for (i = 0; i < height; ++i) {
uint64_t sse64 = 0; for (j = 0; j < width; ++j) {
if (!invert_mask)
for (i = 0; i < h; i++) { comp_pred[j] = AOM_BLEND_A64(mask[j], ref[j], pred[j]);
for (j = 0; j < w; j++) { else
const int diff = (a[j] - b[j]) * (m[j]); comp_pred[j] = AOM_BLEND_A64(mask[j], pred[j], ref[j]);
sum64 += diff;
sse64 += diff * diff;
} }
comp_pred += width;
a += a_stride; pred += width;
b += b_stride; ref += ref_stride;
m += m_stride; mask += mask_stride;
} }
sum64 = (sum64 >= 0) ? sum64 : -sum64;
*sum = (int)ROUND_POWER_OF_TWO(sum64, 6);
*sse = (uint32_t)ROUND_POWER_OF_TWO(sse64, 12);
} }
#define MASK_VAR(W, H) \ void aom_comp_mask_upsampled_pred_c(uint8_t *comp_pred, const uint8_t *pred,
unsigned int aom_masked_variance##W##x##H##_c( \ int width, int height, const uint8_t *ref,
const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, \ int ref_stride, const uint8_t *mask,
const uint8_t *m, int m_stride, unsigned int *sse) { \ int mask_stride, int invert_mask) {
int sum; \ int i, j;
masked_variance(a, a_stride, b, b_stride, m, m_stride, W, H, sse, &sum); \ int stride = ref_stride << 3;
return *sse - (unsigned int)(((int64_t)sum * sum) / (W * H)); \
for (i = 0; i < height; i++) {
for (j = 0; j < width; j++) {
if (!invert_mask)
comp_pred[j] = AOM_BLEND_A64(mask[j], ref[(j << 3)], pred[j]);
else
comp_pred[j] = AOM_BLEND_A64(mask[j], pred[j], ref[(j << 3)]);
}
comp_pred += width;
pred += width;
ref += stride;
mask += mask_stride;
} }
}
#define MASK_SUBPIX_VAR(W, H) \ #define MASK_SUBPIX_VAR(W, H) \
unsigned int aom_masked_sub_pixel_variance##W##x##H##_c( \ unsigned int aom_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, int xoffset, int yoffset, \ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
const uint8_t *dst, int dst_stride, const uint8_t *msk, int msk_stride, \ const uint8_t *ref, int ref_stride, const uint8_t *second_pred, \
const uint8_t *msk, int msk_stride, int invert_mask, \
unsigned int *sse) { \ unsigned int *sse) { \
uint16_t fdata3[(H + 1) * W]; \ uint16_t fdata3[(H + 1) * W]; \
uint8_t temp2[H * W]; \ uint8_t temp2[H * W]; \
DECLARE_ALIGNED(16, uint8_t, temp3[H * W]); \
\ \
var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, H + 1, W, \ var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, H + 1, W, \
bilinear_filters_2t[xoffset]); \ bilinear_filters_2t[xoffset]); \
var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \ var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
bilinear_filters_2t[yoffset]); \ bilinear_filters_2t[yoffset]); \
\ \
return aom_masked_variance##W##x##H##_c(temp2, W, dst, dst_stride, msk, \ aom_comp_mask_pred_c(temp3, second_pred, W, H, temp2, W, msk, msk_stride, \
msk_stride, sse); \ invert_mask); \
return aom_variance##W##x##H##_c(temp3, W, ref, ref_stride, sse); \
} }
MASK_VAR(4, 4)
MASK_SUBPIX_VAR(4, 4) MASK_SUBPIX_VAR(4, 4)
MASK_VAR(4, 8)
MASK_SUBPIX_VAR(4, 8) MASK_SUBPIX_VAR(4, 8)
MASK_VAR(8, 4)
MASK_SUBPIX_VAR(8, 4) MASK_SUBPIX_VAR(8, 4)
MASK_VAR(8, 8)
MASK_SUBPIX_VAR(8, 8) MASK_SUBPIX_VAR(8, 8)
MASK_VAR(8, 16)
MASK_SUBPIX_VAR(8, 16) MASK_SUBPIX_VAR(8, 16)
MASK_VAR(16, 8)
MASK_SUBPIX_VAR(16, 8) MASK_SUBPIX_VAR(16, 8)
MASK_VAR(16, 16)
MASK_SUBPIX_VAR(16, 16) MASK_SUBPIX_VAR(16, 16)
MASK_VAR(16, 32)
MASK_SUBPIX_VAR(16, 32) MASK_SUBPIX_VAR(16, 32)
MASK_VAR(32, 16)
MASK_SUBPIX_VAR(32, 16) MASK_SUBPIX_VAR(32, 16)
MASK_VAR(32, 32)
MASK_SUBPIX_VAR(32, 32) MASK_SUBPIX_VAR(32, 32)
MASK_VAR(32, 64)
MASK_SUBPIX_VAR(32, 64) MASK_SUBPIX_VAR(32, 64)
MASK_VAR(64, 32)
MASK_SUBPIX_VAR(64, 32) MASK_SUBPIX_VAR(64, 32)
MASK_VAR(64, 64)
MASK_SUBPIX_VAR(64, 64) MASK_SUBPIX_VAR(64, 64)
#if CONFIG_EXT_PARTITION #if CONFIG_EXT_PARTITION
MASK_VAR(64, 128)
MASK_SUBPIX_VAR(64, 128) MASK_SUBPIX_VAR(64, 128)
MASK_VAR(128, 64)
MASK_SUBPIX_VAR(128, 64) MASK_SUBPIX_VAR(128, 64)
MASK_VAR(128, 128)
MASK_SUBPIX_VAR(128, 128) MASK_SUBPIX_VAR(128, 128)
#endif // CONFIG_EXT_PARTITION #endif // CONFIG_EXT_PARTITION
#if CONFIG_HIGHBITDEPTH #if CONFIG_HIGHBITDEPTH
void highbd_masked_variance64(const uint8_t *a8, int a_stride, void aom_highbd_comp_mask_pred_c(uint16_t *comp_pred, const uint8_t *pred8,
const uint8_t *b8, int b_stride, const uint8_t *m, int width, int height, const uint8_t *ref8,
int m_stride, int w, int h, uint64_t *sse, int ref_stride, const uint8_t *mask,
int64_t *sum) { int mask_stride, int invert_mask) {
int i, j; int i, j;
uint16_t *a = CONVERT_TO_SHORTPTR(a8); uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
uint16_t *b = CONVERT_TO_SHORTPTR(b8); uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
for (i = 0; i < height; ++i) {
*sum = 0; for (j = 0; j < width; ++j) {
*sse = 0; if (!invert_mask)
comp_pred[j] = AOM_BLEND_A64(mask[j], ref[j], pred[j]);
for (i = 0; i < h; i++) { else
for (j = 0; j < w; j++) { comp_pred[j] = AOM_BLEND_A64(mask[j], pred[j], ref[j]);
const int diff = (a[j] - b[j]) * (m[j]);
*sum += (int64_t)diff;
*sse += (int64_t)diff * diff;
} }
comp_pred += width;
a += a_stride; pred += width;
b += b_stride; ref += ref_stride;
m += m_stride; mask += mask_stride;
} }
*sum = (*sum >= 0) ? *sum : -*sum;
*sum = ROUND_POWER_OF_TWO(*sum, 6);
*sse = ROUND_POWER_OF_TWO(*sse, 12);
} }
void highbd_masked_variance(const uint8_t *a8, int a_stride, const uint8_t *b8, void aom_highbd_comp_mask_upsampled_pred_c(uint16_t *comp_pred,
int b_stride, const uint8_t *m, int m_stride, int w, const uint8_t *pred8, int width,
int h, unsigned int *sse, int *sum) { int height, const uint8_t *ref8,
int64_t sum64; int ref_stride, const uint8_t *mask,
uint64_t sse64; int mask_stride, int invert_mask) {
highbd_masked_variance64(a8, a_stride, b8, b_stride, m, m_stride, w, h, int i, j;
&sse64, &sum64); int stride = ref_stride << 3;
*sum = (int)sum64;
*sse = (unsigned int)sse64; uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
for (i = 0; i < height; ++i) {
for (j = 0; j < width; ++j) {
if (!invert_mask)
comp_pred[j] = AOM_BLEND_A64(mask[j], ref[j << 3], pred[j]);
else
comp_pred[j] = AOM_BLEND_A64(mask[j], pred[j], ref[j << 3]);
}
comp_pred += width;
pred += width;
ref += stride;
mask += mask_stride;
}
} }
void highbd_10_masked_variance(const uint8_t *a8, int a_stride, #define HIGHBD_MASK_SUBPIX_VAR(W, H) \
const uint8_t *b8, int b_stride, unsigned int aom_highbd_8_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *m, int m_stride, int w, int h, const uint8_t *src, int src_stride, int xoffset, int yoffset, \
unsigned int *sse, int *sum) { const uint8_t *ref, int ref_stride, const uint8_t *second_pred, \
int64_t sum64; const uint8_t *msk, int msk_stride, int invert_mask, \
uint64_t sse64; unsigned int *sse) { \
highbd_masked_variance64(a8, a_stride, b8, b_stride, m, m_stride, w, h, uint16_t fdata3[(H + 1) * W]; \
&sse64, &sum64); uint16_t temp2[H * W]; \
*sum = (int)ROUND_POWER_OF_TWO(sum64, 2); DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
*sse = (unsigned int)ROUND_POWER_OF_TWO(sse64, 4); \
} aom_highbd_var_filter_block2d_bil_first_pass( \
src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
void highbd_12_masked_variance(const uint8_t *a8, int a_stride, aom_highbd_var_filter_block2d_bil_second_pass( \
const uint8_t *b8, int b_stride, fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
const uint8_t *m, int m_stride, int w, int h, \
unsigned int *sse, int *sum) { aom_highbd_comp_mask_pred_c(temp3, second_pred, W, H, \
int64_t sum64; CONVERT_TO_BYTEPTR(temp2), W, msk, msk_stride, \
uint64_t sse64; invert_mask); \
highbd_masked_variance64(a8, a_stride, b8, b_stride, m, m_stride, w, h, \
&sse64, &sum64); return aom_highbd_8_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, \
*sum = (int)ROUND_POWER_OF_TWO(sum64, 4); ref, ref_stride, sse); \
*sse = (unsigned int)ROUND_POWER_OF_TWO(sse64, 8); } \
} \
unsigned int aom_highbd_10_masked_sub_pixel_variance##W##x##H##_c( \
#define HIGHBD_MASK_VAR(W, H) \ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
unsigned int aom_highbd_masked_variance##W##x##H##_c( \ const uint8_t *ref, int ref_stride, const uint8_t *second_pred, \
const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, \ const uint8_t *msk, int msk_stride, int invert_mask, \
const uint8_t *m, int m_stride, unsigned int *sse) { \ unsigned int *sse) { \
int sum; \ uint16_t fdata3[(H + 1) * W]; \
highbd_masked_variance(a, a_stride, b, b_stride, m, m_stride, W, H, sse, \ uint16_t temp2[H * W]; \
&sum); \ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
return *sse - (unsigned int)(((int64_t)sum * sum) / (W * H)); \ \
} \ aom_highbd_var_filter_block2d_bil_first_pass( \
\ src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
unsigned int aom_highbd_10_masked_variance##W##x##H##_c( \ aom_highbd_var_filter_block2d_bil_second_pass( \
const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, \ fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
const uint8_t *m, int m_stride, unsigned int *sse) { \ \
int sum; \ aom_highbd_comp_mask_pred_c(temp3, second_pred, W, H, \
int64_t var; \ CONVERT_TO_BYTEPTR(temp2), W, msk, msk_stride, \
highbd_10_masked_variance(a, a_stride, b, b_stride, m, m_stride, W, H, \ invert_mask); \
sse, &sum); \ \
var = (int64_t)(*sse) - (((int64_t)sum * sum) / (W * H)); \ return aom_highbd_10_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, \
return (var >= 0) ? (uint32_t)var : 0; \ ref, ref_stride, sse); \
} \ } \
\ \
unsigned int aom_highbd_12_masked_variance##W##x##H##_c( \ unsigned int aom_highbd_12_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, \ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
const uint8_t *m, int m_stride, unsigned int *sse) { \ const uint8_t *ref, int ref_stride, const uint8_t *second_pred, \
int sum; \ const uint8_t *msk, int msk_stride, int invert_mask, \
int64_t var; \ unsigned int *sse) { \
highbd_12_masked_variance(a, a_stride, b, b_stride, m, m_stride, W, H, \ uint16_t fdata3[(H + 1) * W]; \
sse, &sum); \ uint16_t temp2[H * W]; \
var = (int64_t)(*sse) - (((int64_t)sum * sum) / (W * H)); \ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
return (var >= 0) ? (uint32_t)var : 0; \ \
aom_highbd_var_filter_block2d_bil_first_pass( \
src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
aom_highbd_var_filter_block2d_bil_second_pass( \
fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
\
aom_highbd_comp_mask_pred_c(temp3, second_pred, W, H, \
CONVERT_TO_BYTEPTR(temp2), W, msk, msk_stride, \
invert_mask); \
\
return aom_highbd_12_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, \
ref, ref_stride, sse); \
} }
#define HIGHBD_MASK_SUBPIX_VAR(W, H) \
unsigned int aom_highbd_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, int xoffset, int yoffset, \
const uint8_t *dst, int dst_stride, const uint8_t *msk, int msk_stride, \
unsigned int *sse) { \
uint16_t fdata3[(H + 1) * W]; \
uint16_t temp2[H * W]; \
\
aom_highbd_var_filter_block2d_bil_first_pass( \
src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
aom_highbd_var_filter_block2d_bil_second_pass( \
fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
\
return aom_highbd_masked_variance##W##x##H##_c( \
CONVERT_TO_BYTEPTR(temp2), W, dst, dst_stride, msk, msk_stride, sse); \
} \
\
unsigned int aom_highbd_10_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, int xoffset, int yoffset, \
const uint8_t *dst, int dst_stride, const uint8_t *msk, int msk_stride, \
unsigned int *sse) { \
uint16_t fdata3[(H + 1) * W]; \
uint16_t temp2[H * W]; \
\
aom_highbd_var_filter_block2d_bil_first_pass( \
src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
aom_highbd_var_filter_block2d_bil_second_pass( \
fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
\
return aom_highbd_10_masked_variance##W##x##H##_c( \
CONVERT_TO_BYTEPTR(temp2), W, dst, dst_stride, msk, msk_stride, sse); \
} \
\
unsigned int aom_highbd_12_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, int xoffset, int yoffset, \
const uint8_t *dst, int dst_stride, const uint8_t *msk, int msk_stride, \
unsigned int *sse) { \
uint16_t fdata3[(H + 1) * W]; \
uint16_t temp2[H * W]; \
\
aom_highbd_var_filter_block2d_bil_first_pass( \
src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
aom_highbd_var_filter_block2d_bil_second_pass( \
fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
\
return aom_highbd_12_masked_variance##W##x##H##_c( \
CONVERT_TO_BYTEPTR(temp2), W, dst, dst_stride, msk, msk_stride, sse); \
}
HIGHBD_MASK_VAR(4, 4)
HIGHBD_MASK_SUBPIX_VAR(4, 4) HIGHBD_MASK_SUBPIX_VAR(4, 4)
HIGHBD_MASK_VAR(4, 8)
HIGHBD_MASK_SUBPIX_VAR(4, 8) HIGHBD_MASK_SUBPIX_VAR(4, 8)
HIGHBD_MASK_VAR(8, 4)
HIGHBD_MASK_SUBPIX_VAR(8, 4) HIGHBD_MASK_SUBPIX_VAR(8, 4)
HIGHBD_MASK_VAR(8, 8)
HIGHBD_MASK_SUBPIX_VAR(8, 8) HIGHBD_MASK_SUBPIX_VAR(8, 8)
HIGHBD_MASK_VAR(8, 16)
HIGHBD_MASK_SUBPIX_VAR(8, 16) HIGHBD_MASK_SUBPIX_VAR(8, 16)
HIGHBD_MASK_VAR(16, 8)
HIGHBD_MASK_SUBPIX_VAR(16, 8) HIGHBD_MASK_SUBPIX_VAR(16, 8)
HIGHBD_MASK_VAR(16, 16)
HIGHBD_MASK_SUBPIX_VAR(16, 16) HIGHBD_MASK_SUBPIX_VAR(16, 16)
HIGHBD_MASK_VAR(16, 32)
HIGHBD_MASK_SUBPIX_VAR(16, 32) HIGHBD_MASK_SUBPIX_VAR(16, 32)
HIGHBD_MASK_VAR(32, 16)
HIGHBD_MASK_SUBPIX_VAR(32, 16) HIGHBD_MASK_SUBPIX_VAR(32, 16)
HIGHBD_MASK_VAR(32, 32)
HIGHBD_MASK_SUBPIX_VAR(32, 32) HIGHBD_MASK_SUBPIX_VAR(32, 32)
HIGHBD_MASK_VAR(32, 64)
HIGHBD_MASK_SUBPIX_VAR(32, 64) HIGHBD_MASK_SUBPIX_VAR(32, 64)
HIGHBD_MASK_VAR(64, 32)
HIGHBD_MASK_SUBPIX_VAR(64, 32) HIGHBD_MASK_SUBPIX_VAR(64, 32)
HIGHBD_MASK_VAR(64, 64)
HIGHBD_MASK_SUBPIX_VAR(64, 64) HIGHBD_MASK_SUBPIX_VAR(64, 64)
#if CONFIG_EXT_PARTITION #if CONFIG_EXT_PARTITION
HIGHBD_MASK_VAR(64, 128)
HIGHBD_MASK_SUBPIX_VAR(64, 128) HIGHBD_MASK_SUBPIX_VAR(64, 128)
HIGHBD_MASK_VAR(128, 64)
HIGHBD_MASK_SUBPIX_VAR(128, 64) HIGHBD_MASK_SUBPIX_VAR(128, 64)
HIGHBD_MASK_VAR(128, 128)
HIGHBD_MASK_SUBPIX_VAR(128, 128) HIGHBD_MASK_SUBPIX_VAR(128, 128)
#endif // CONFIG_EXT_PARTITION #endif // CONFIG_EXT_PARTITION
#endif // CONFIG_HIGHBITDEPTH #endif // CONFIG_HIGHBITDEPTH

13
third_party/aom/aom_dsp/variance.h vendored
View file

@ -57,15 +57,13 @@ typedef unsigned int (*aom_subp_avg_variance_fn_t)(
#if CONFIG_AV1 && CONFIG_EXT_INTER #if CONFIG_AV1 && CONFIG_EXT_INTER
typedef unsigned int (*aom_masked_sad_fn_t)(const uint8_t *src, int src_stride, typedef unsigned int (*aom_masked_sad_fn_t)(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride, const uint8_t *ref, int ref_stride,
const uint8_t *msk_ptr, const uint8_t *second_pred,
int msk_stride); const uint8_t *msk, int msk_stride,
typedef unsigned int (*aom_masked_variance_fn_t)( int invert_mask);
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride,
const uint8_t *msk, int msk_stride, unsigned int *sse);
typedef unsigned int (*aom_masked_subpixvariance_fn_t)( typedef unsigned int (*aom_masked_subpixvariance_fn_t)(
const uint8_t *src, int src_stride, int xoffset, int yoffset, const uint8_t *src, int src_stride, int xoffset, int yoffset,
const uint8_t *ref, int ref_stride, const uint8_t *msk, int msk_stride, const uint8_t *ref, int ref_stride, const uint8_t *second_pred,
unsigned int *sse); const uint8_t *msk, int msk_stride, int invert_mask, unsigned int *sse);
#endif // CONFIG_AV1 && CONFIG_EXT_INTER #endif // CONFIG_AV1 && CONFIG_EXT_INTER
#if CONFIG_AV1 && CONFIG_MOTION_VAR #if CONFIG_AV1 && CONFIG_MOTION_VAR
@ -94,7 +92,6 @@ typedef struct aom_variance_vtable {
aom_sad_multi_d_fn_t sdx4df; aom_sad_multi_d_fn_t sdx4df;
#if CONFIG_EXT_INTER #if CONFIG_EXT_INTER
aom_masked_sad_fn_t msdf; aom_masked_sad_fn_t msdf;
aom_masked_variance_fn_t mvf;
aom_masked_subpixvariance_fn_t msvf; aom_masked_subpixvariance_fn_t msvf;
#endif // CONFIG_EXT_INTER #endif // CONFIG_EXT_INTER
#if CONFIG_MOTION_VAR #if CONFIG_MOTION_VAR

195
third_party/aom/aom_dsp/x86/aom_convolve_hip_sse2.c vendored Normal file
View file

@ -0,0 +1,195 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include <emmintrin.h>
#include <assert.h>
#include "./aom_dsp_rtcd.h"
#include "aom_dsp/aom_convolve.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/aom_filter.h"
void aom_convolve8_add_src_hip_sse2(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4,
int w, int h) {
const int bd = 8;
assert(x_step_q4 == 16 && y_step_q4 == 16);
assert(!(w & 7));
(void)x_step_q4;
(void)y_step_q4;
uint16_t temp[(MAX_SB_SIZE + SUBPEL_TAPS - 1) * MAX_SB_SIZE];
int intermediate_height = h + SUBPEL_TAPS - 1;
int i, j;
const int center_tap = ((SUBPEL_TAPS - 1) / 2);
const uint8_t *const src_ptr = src - center_tap * src_stride - center_tap;
const __m128i zero = _mm_setzero_si128();
// Add an offset to account for the "add_src" part of the convolve function.
const __m128i offset = _mm_insert_epi16(zero, 1 << FILTER_BITS, 3);
/* Horizontal filter */
{
const __m128i coeffs_x =
_mm_add_epi16(_mm_loadu_si128((__m128i *)filter_x), offset);
// coeffs 0 1 0 1 2 3 2 3
const __m128i tmp_0 = _mm_unpacklo_epi32(coeffs_x, coeffs_x);
// coeffs 4 5 4 5 6 7 6 7
const __m128i tmp_1 = _mm_unpackhi_epi32(coeffs_x, coeffs_x);
// coeffs 0 1 0 1 0 1 0 1
const __m128i coeff_01 = _mm_unpacklo_epi64(tmp_0, tmp_0);
// coeffs 2 3 2 3 2 3 2 3
const __m128i coeff_23 = _mm_unpackhi_epi64(tmp_0, tmp_0);
// coeffs 4 5 4 5 4 5 4 5
const __m128i coeff_45 = _mm_unpacklo_epi64(tmp_1, tmp_1);
// coeffs 6 7 6 7 6 7 6 7
const __m128i coeff_67 = _mm_unpackhi_epi64(tmp_1, tmp_1);
const __m128i round_const =
_mm_set1_epi32((1 << (FILTER_BITS - EXTRAPREC_BITS - 1)) +
(1 << (bd + FILTER_BITS - 1)));
for (i = 0; i < intermediate_height; ++i) {
for (j = 0; j < w; j += 8) {
const __m128i data =
_mm_loadu_si128((__m128i *)&src_ptr[i * src_stride + j]);
// Filter even-index pixels
const __m128i src_0 = _mm_unpacklo_epi8(data, zero);
const __m128i res_0 = _mm_madd_epi16(src_0, coeff_01);
const __m128i src_2 = _mm_unpacklo_epi8(_mm_srli_si128(data, 2), zero);
const __m128i res_2 = _mm_madd_epi16(src_2, coeff_23);
const __m128i src_4 = _mm_unpacklo_epi8(_mm_srli_si128(data, 4), zero);
const __m128i res_4 = _mm_madd_epi16(src_4, coeff_45);
const __m128i src_6 = _mm_unpacklo_epi8(_mm_srli_si128(data, 6), zero);
const __m128i res_6 = _mm_madd_epi16(src_6, coeff_67);
__m128i res_even = _mm_add_epi32(_mm_add_epi32(res_0, res_4),
_mm_add_epi32(res_2, res_6));
res_even = _mm_srai_epi32(_mm_add_epi32(res_even, round_const),
FILTER_BITS - EXTRAPREC_BITS);
// Filter odd-index pixels
const __m128i src_1 = _mm_unpacklo_epi8(_mm_srli_si128(data, 1), zero);
const __m128i res_1 = _mm_madd_epi16(src_1, coeff_01);
const __m128i src_3 = _mm_unpacklo_epi8(_mm_srli_si128(data, 3), zero);
const __m128i res_3 = _mm_madd_epi16(src_3, coeff_23);
const __m128i src_5 = _mm_unpacklo_epi8(_mm_srli_si128(data, 5), zero);
const __m128i res_5 = _mm_madd_epi16(src_5, coeff_45);
const __m128i src_7 = _mm_unpacklo_epi8(_mm_srli_si128(data, 7), zero);
const __m128i res_7 = _mm_madd_epi16(src_7, coeff_67);
__m128i res_odd = _mm_add_epi32(_mm_add_epi32(res_1, res_5),
_mm_add_epi32(res_3, res_7));
res_odd = _mm_srai_epi32(_mm_add_epi32(res_odd, round_const),
FILTER_BITS - EXTRAPREC_BITS);
// Pack in the column order 0, 2, 4, 6, 1, 3, 5, 7
__m128i res = _mm_packs_epi32(res_even, res_odd);
res = _mm_min_epi16(_mm_max_epi16(res, zero),
_mm_set1_epi16(EXTRAPREC_CLAMP_LIMIT(bd) - 1));
_mm_storeu_si128((__m128i *)&temp[i * MAX_SB_SIZE + j], res);
}
}
}
/* Vertical filter */
{
const __m128i coeffs_y =
_mm_add_epi16(_mm_loadu_si128((__m128i *)filter_y), offset);
// coeffs 0 1 0 1 2 3 2 3
const __m128i tmp_0 = _mm_unpacklo_epi32(coeffs_y, coeffs_y);
// coeffs 4 5 4 5 6 7 6 7
const __m128i tmp_1 = _mm_unpackhi_epi32(coeffs_y, coeffs_y);
// coeffs 0 1 0 1 0 1 0 1
const __m128i coeff_01 = _mm_unpacklo_epi64(tmp_0, tmp_0);
// coeffs 2 3 2 3 2 3 2 3
const __m128i coeff_23 = _mm_unpackhi_epi64(tmp_0, tmp_0);
// coeffs 4 5 4 5 4 5 4 5
const __m128i coeff_45 = _mm_unpacklo_epi64(tmp_1, tmp_1);
// coeffs 6 7 6 7 6 7 6 7
const __m128i coeff_67 = _mm_unpackhi_epi64(tmp_1, tmp_1);
const __m128i round_const =
_mm_set1_epi32((1 << (FILTER_BITS + EXTRAPREC_BITS - 1)) -
(1 << (bd + FILTER_BITS + EXTRAPREC_BITS - 1)));
for (i = 0; i < h; ++i) {
for (j = 0; j < w; j += 8) {
// Filter even-index pixels
const uint16_t *data = &temp[i * MAX_SB_SIZE + j];
const __m128i src_0 =
_mm_unpacklo_epi16(*(__m128i *)(data + 0 * MAX_SB_SIZE),
*(__m128i *)(data + 1 * MAX_SB_SIZE));
const __m128i src_2 =
_mm_unpacklo_epi16(*(__m128i *)(data + 2 * MAX_SB_SIZE),
*(__m128i *)(data + 3 * MAX_SB_SIZE));
const __m128i src_4 =
_mm_unpacklo_epi16(*(__m128i *)(data + 4 * MAX_SB_SIZE),
*(__m128i *)(data + 5 * MAX_SB_SIZE));
const __m128i src_6 =
_mm_unpacklo_epi16(*(__m128i *)(data + 6 * MAX_SB_SIZE),
*(__m128i *)(data + 7 * MAX_SB_SIZE));
const __m128i res_0 = _mm_madd_epi16(src_0, coeff_01);
const __m128i res_2 = _mm_madd_epi16(src_2, coeff_23);
const __m128i res_4 = _mm_madd_epi16(src_4, coeff_45);
const __m128i res_6 = _mm_madd_epi16(src_6, coeff_67);
const __m128i res_even = _mm_add_epi32(_mm_add_epi32(res_0, res_2),
_mm_add_epi32(res_4, res_6));
// Filter odd-index pixels
const __m128i src_1 =
_mm_unpackhi_epi16(*(__m128i *)(data + 0 * MAX_SB_SIZE),
*(__m128i *)(data + 1 * MAX_SB_SIZE));
const __m128i src_3 =
_mm_unpackhi_epi16(*(__m128i *)(data + 2 * MAX_SB_SIZE),
*(__m128i *)(data + 3 * MAX_SB_SIZE));
const __m128i src_5 =
_mm_unpackhi_epi16(*(__m128i *)(data + 4 * MAX_SB_SIZE),
*(__m128i *)(data + 5 * MAX_SB_SIZE));
const __m128i src_7 =
_mm_unpackhi_epi16(*(__m128i *)(data + 6 * MAX_SB_SIZE),
*(__m128i *)(data + 7 * MAX_SB_SIZE));
const __m128i res_1 = _mm_madd_epi16(src_1, coeff_01);
const __m128i res_3 = _mm_madd_epi16(src_3, coeff_23);
const __m128i res_5 = _mm_madd_epi16(src_5, coeff_45);
const __m128i res_7 = _mm_madd_epi16(src_7, coeff_67);
const __m128i res_odd = _mm_add_epi32(_mm_add_epi32(res_1, res_3),
_mm_add_epi32(res_5, res_7));
// Rearrange pixels back into the order 0 ... 7
const __m128i res_lo = _mm_unpacklo_epi32(res_even, res_odd);
const __m128i res_hi = _mm_unpackhi_epi32(res_even, res_odd);
const __m128i res_lo_round = _mm_srai_epi32(
_mm_add_epi32(res_lo, round_const), FILTER_BITS + EXTRAPREC_BITS);
const __m128i res_hi_round = _mm_srai_epi32(
_mm_add_epi32(res_hi, round_const), FILTER_BITS + EXTRAPREC_BITS);
const __m128i res_16bit = _mm_packs_epi32(res_lo_round, res_hi_round);
__m128i res_8bit = _mm_packus_epi16(res_16bit, res_16bit);
__m128i *const p = (__m128i *)&dst[i * dst_stride + j];
_mm_storel_epi64(p, res_8bit);
}
}
}
}

203
third_party/aom/aom_dsp/x86/aom_highbd_convolve_hip_ssse3.c vendored Normal file
View file

@ -0,0 +1,203 @@
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include <tmmintrin.h>
#include <assert.h>
#include "./aom_dsp_rtcd.h"
#include "aom_dsp/aom_convolve.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/aom_filter.h"
#if EXTRAPREC_BITS > 2
#error "Highbd high-prec convolve filter only supports EXTRAPREC_BITS <= 2"
#error "(need to use 32-bit intermediates for EXTRAPREC_BITS > 2)"
#endif
void aom_highbd_convolve8_add_src_hip_ssse3(
const uint8_t *src8, ptrdiff_t src_stride, uint8_t *dst8,
ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4,
const int16_t *filter_y, int y_step_q4, int w, int h, int bd) {
assert(x_step_q4 == 16 && y_step_q4 == 16);
assert(!(w & 7));
(void)x_step_q4;
(void)y_step_q4;
const uint16_t *const src = CONVERT_TO_SHORTPTR(src8);
uint16_t *const dst = CONVERT_TO_SHORTPTR(dst8);
uint16_t temp[(MAX_SB_SIZE + SUBPEL_TAPS - 1) * MAX_SB_SIZE];
int intermediate_height = h + SUBPEL_TAPS - 1;
int i, j;
const int center_tap = ((SUBPEL_TAPS - 1) / 2);
const uint16_t *const src_ptr = src - center_tap * src_stride - center_tap;
const __m128i zero = _mm_setzero_si128();
// Add an offset to account for the "add_src" part of the convolve function.
const __m128i offset = _mm_insert_epi16(zero, 1 << FILTER_BITS, 3);
/* Horizontal filter */
{
const __m128i coeffs_x =
_mm_add_epi16(_mm_loadu_si128((__m128i *)filter_x), offset);
// coeffs 0 1 0 1 2 3 2 3
const __m128i tmp_0 = _mm_unpacklo_epi32(coeffs_x, coeffs_x);
// coeffs 4 5 4 5 6 7 6 7
const __m128i tmp_1 = _mm_unpackhi_epi32(coeffs_x, coeffs_x);
// coeffs 0 1 0 1 0 1 0 1
const __m128i coeff_01 = _mm_unpacklo_epi64(tmp_0, tmp_0);
// coeffs 2 3 2 3 2 3 2 3
const __m128i coeff_23 = _mm_unpackhi_epi64(tmp_0, tmp_0);
// coeffs 4 5 4 5 4 5 4 5
const __m128i coeff_45 = _mm_unpacklo_epi64(tmp_1, tmp_1);
// coeffs 6 7 6 7 6 7 6 7
const __m128i coeff_67 = _mm_unpackhi_epi64(tmp_1, tmp_1);
const __m128i round_const =
_mm_set1_epi32((1 << (FILTER_BITS - EXTRAPREC_BITS - 1)) +
(1 << (bd + FILTER_BITS - 1)));
for (i = 0; i < intermediate_height; ++i) {
for (j = 0; j < w; j += 8) {
const __m128i data =
_mm_loadu_si128((__m128i *)&src_ptr[i * src_stride + j]);
const __m128i data2 =
_mm_loadu_si128((__m128i *)&src_ptr[i * src_stride + j + 8]);
// Filter even-index pixels
const __m128i res_0 = _mm_madd_epi16(data, coeff_01);
const __m128i res_2 =
_mm_madd_epi16(_mm_alignr_epi8(data2, data, 4), coeff_23);
const __m128i res_4 =
_mm_madd_epi16(_mm_alignr_epi8(data2, data, 8), coeff_45);
const __m128i res_6 =
_mm_madd_epi16(_mm_alignr_epi8(data2, data, 12), coeff_67);
__m128i res_even = _mm_add_epi32(_mm_add_epi32(res_0, res_4),
_mm_add_epi32(res_2, res_6));
res_even = _mm_srai_epi32(_mm_add_epi32(res_even, round_const),
FILTER_BITS - EXTRAPREC_BITS);
// Filter odd-index pixels
const __m128i res_1 =
_mm_madd_epi16(_mm_alignr_epi8(data2, data, 2), coeff_01);
const __m128i res_3 =
_mm_madd_epi16(_mm_alignr_epi8(data2, data, 6), coeff_23);
const __m128i res_5 =
_mm_madd_epi16(_mm_alignr_epi8(data2, data, 10), coeff_45);
const __m128i res_7 =
_mm_madd_epi16(_mm_alignr_epi8(data2, data, 14), coeff_67);
__m128i res_odd = _mm_add_epi32(_mm_add_epi32(res_1, res_5),
_mm_add_epi32(res_3, res_7));
res_odd = _mm_srai_epi32(_mm_add_epi32(res_odd, round_const),
FILTER_BITS - EXTRAPREC_BITS);
// Pack in the column order 0, 2, 4, 6, 1, 3, 5, 7
const __m128i maxval = _mm_set1_epi16((EXTRAPREC_CLAMP_LIMIT(bd)) - 1);
__m128i res = _mm_packs_epi32(res_even, res_odd);
res = _mm_min_epi16(_mm_max_epi16(res, zero), maxval);
_mm_storeu_si128((__m128i *)&temp[i * MAX_SB_SIZE + j], res);
}
}
}
/* Vertical filter */
{
const __m128i coeffs_y =
_mm_add_epi16(_mm_loadu_si128((__m128i *)filter_y), offset);
// coeffs 0 1 0 1 2 3 2 3
const __m128i tmp_0 = _mm_unpacklo_epi32(coeffs_y, coeffs_y);
// coeffs 4 5 4 5 6 7 6 7
const __m128i tmp_1 = _mm_unpackhi_epi32(coeffs_y, coeffs_y);
// coeffs 0 1 0 1 0 1 0 1
const __m128i coeff_01 = _mm_unpacklo_epi64(tmp_0, tmp_0);
// coeffs 2 3 2 3 2 3 2 3
const __m128i coeff_23 = _mm_unpackhi_epi64(tmp_0, tmp_0);
// coeffs 4 5 4 5 4 5 4 5
const __m128i coeff_45 = _mm_unpacklo_epi64(tmp_1, tmp_1);
// coeffs 6 7 6 7 6 7 6 7
const __m128i coeff_67 = _mm_unpackhi_epi64(tmp_1, tmp_1);
const __m128i round_const =
_mm_set1_epi32((1 << (FILTER_BITS + EXTRAPREC_BITS - 1)) -
(1 << (bd + FILTER_BITS + EXTRAPREC_BITS - 1)));
for (i = 0; i < h; ++i) {
for (j = 0; j < w; j += 8) {
// Filter even-index pixels
const uint16_t *data = &temp[i * MAX_SB_SIZE + j];
const __m128i src_0 =
_mm_unpacklo_epi16(*(__m128i *)(data + 0 * MAX_SB_SIZE),
*(__m128i *)(data + 1 * MAX_SB_SIZE));
const __m128i src_2 =
_mm_unpacklo_epi16(*(__m128i *)(data + 2 * MAX_SB_SIZE),
*(__m128i *)(data + 3 * MAX_SB_SIZE));
const __m128i src_4 =
_mm_unpacklo_epi16(*(__m128i *)(data + 4 * MAX_SB_SIZE),
*(__m128i *)(data + 5 * MAX_SB_SIZE));
const __m128i src_6 =
_mm_unpacklo_epi16(*(__m128i *)(data + 6 * MAX_SB_SIZE),
*(__m128i *)(data + 7 * MAX_SB_SIZE));
const __m128i res_0 = _mm_madd_epi16(src_0, coeff_01);
const __m128i res_2 = _mm_madd_epi16(src_2, coeff_23);
const __m128i res_4 = _mm_madd_epi16(src_4, coeff_45);
const __m128i res_6 = _mm_madd_epi16(src_6, coeff_67);
const __m128i res_even = _mm_add_epi32(_mm_add_epi32(res_0, res_2),
_mm_add_epi32(res_4, res_6));
// Filter odd-index pixels
const __m128i src_1 =
_mm_unpackhi_epi16(*(__m128i *)(data + 0 * MAX_SB_SIZE),
*(__m128i *)(data + 1 * MAX_SB_SIZE));
const __m128i src_3 =
_mm_unpackhi_epi16(*(__m128i *)(data + 2 * MAX_SB_SIZE),
*(__m128i *)(data + 3 * MAX_SB_SIZE));
const __m128i src_5 =
_mm_unpackhi_epi16(*(__m128i *)(data + 4 * MAX_SB_SIZE),
*(__m128i *)(data + 5 * MAX_SB_SIZE));
const __m128i src_7 =
_mm_unpackhi_epi16(*(__m128i *)(data + 6 * MAX_SB_SIZE),
*(__m128i *)(data + 7 * MAX_SB_SIZE));
const __m128i res_1 = _mm_madd_epi16(src_1, coeff_01);
const __m128i res_3 = _mm_madd_epi16(src_3, coeff_23);
const __m128i res_5 = _mm_madd_epi16(src_5, coeff_45);
const __m128i res_7 = _mm_madd_epi16(src_7, coeff_67);
const __m128i res_odd = _mm_add_epi32(_mm_add_epi32(res_1, res_3),
_mm_add_epi32(res_5, res_7));
// Rearrange pixels back into the order 0 ... 7
const __m128i res_lo = _mm_unpacklo_epi32(res_even, res_odd);
const __m128i res_hi = _mm_unpackhi_epi32(res_even, res_odd);
const __m128i res_lo_round = _mm_srai_epi32(
_mm_add_epi32(res_lo, round_const), FILTER_BITS + EXTRAPREC_BITS);
const __m128i res_hi_round = _mm_srai_epi32(
_mm_add_epi32(res_hi, round_const), FILTER_BITS + EXTRAPREC_BITS);
const __m128i maxval = _mm_set1_epi16((1 << bd) - 1);
__m128i res_16bit = _mm_packs_epi32(res_lo_round, res_hi_round);
res_16bit = _mm_min_epi16(_mm_max_epi16(res_16bit, zero), maxval);
__m128i *const p = (__m128i *)&dst[i * dst_stride + j];
_mm_storeu_si128(p, res_16bit);
}
}
}
}

46
third_party/aom/aom_dsp/x86/avg_intrin_sse2.c vendored
View file

@ -94,52 +94,6 @@ void aom_minmax_8x8_sse2(const uint8_t *s, int p, const uint8_t *d, int dp,
*min = _mm_extract_epi16(minabsdiff, 0); *min = _mm_extract_epi16(minabsdiff, 0);
} }
unsigned int aom_avg_8x8_sse2(const uint8_t *s, int p) {
__m128i s0, s1, u0;
unsigned int avg = 0;
u0 = _mm_setzero_si128();
s0 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s)), u0);
s1 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s + p)), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s + 2 * p)), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s + 3 * p)), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s + 4 * p)), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s + 5 * p)), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s + 6 * p)), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(s + 7 * p)), u0);
s0 = _mm_adds_epu16(s0, s1);
s0 = _mm_adds_epu16(s0, _mm_srli_si128(s0, 8));
s0 = _mm_adds_epu16(s0, _mm_srli_epi64(s0, 32));
s0 = _mm_adds_epu16(s0, _mm_srli_epi64(s0, 16));
avg = _mm_extract_epi16(s0, 0);
return (avg + 32) >> 6;
}
unsigned int aom_avg_4x4_sse2(const uint8_t *s, int p) {
__m128i s0, s1, u0;
unsigned int avg = 0;
u0 = _mm_setzero_si128();
s0 = _mm_unpacklo_epi8(xx_loadl_32(s), u0);
s1 = _mm_unpacklo_epi8(xx_loadl_32(s + p), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(xx_loadl_32(s + 2 * p), u0);
s0 = _mm_adds_epu16(s0, s1);
s1 = _mm_unpacklo_epi8(xx_loadl_32(s + 3 * p), u0);
s0 = _mm_adds_epu16(s0, s1);
s0 = _mm_adds_epu16(s0, _mm_srli_si128(s0, 4));
s0 = _mm_adds_epu16(s0, _mm_srli_epi64(s0, 16));
avg = _mm_extract_epi16(s0, 0);
return (avg + 8) >> 4;
}
static void hadamard_col8_sse2(__m128i *in, int iter) { static void hadamard_col8_sse2(__m128i *in, int iter) {
__m128i a0 = in[0]; __m128i a0 = in[0];
__m128i a1 = in[1]; __m128i a1 = in[1];

660
third_party/aom/aom_dsp/x86/highbd_convolve_avx2.c vendored
View file

@ -14,30 +14,6 @@
#include "./aom_dsp_rtcd.h" #include "./aom_dsp_rtcd.h"
#include "aom_dsp/x86/convolve.h" #include "aom_dsp/x86/convolve.h"
#define CONV8_ROUNDING_BITS (7)
static const uint8_t signal_pattern_0[32] = { 0, 1, 2, 3, 2, 3, 4, 5, 4, 5, 6,
7, 6, 7, 8, 9, 0, 1, 2, 3, 2, 3,
4, 5, 4, 5, 6, 7, 6, 7, 8, 9 };
static const uint8_t signal_pattern_1[32] = { 4, 5, 6, 7, 6, 7, 8, 9,
8, 9, 10, 11, 10, 11, 12, 13,
4, 5, 6, 7, 6, 7, 8, 9,
8, 9, 10, 11, 10, 11, 12, 13 };
static const uint8_t signal_pattern_2[32] = { 6, 7, 8, 9, 8, 9, 10, 11,
10, 11, 12, 13, 12, 13, 14, 15,
6, 7, 8, 9, 8, 9, 10, 11,
10, 11, 12, 13, 12, 13, 14, 15 };
static const uint32_t signal_index[8] = { 2, 3, 4, 5, 2, 3, 4, 5 };
typedef enum { PACK_8x1, PACK_8x2, PACK_16x1 } PixelPackFormat;
typedef void (*WritePixels)(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst,
ptrdiff_t pitch);
// ----------------------------------------------------------------------------- // -----------------------------------------------------------------------------
// Copy and average // Copy and average
@ -216,6 +192,27 @@ void aom_highbd_convolve_avg_avx2(const uint8_t *src8, ptrdiff_t src_stride,
} }
} }
// -----------------------------------------------------------------------------
// Horizontal and vertical filtering
#define CONV8_ROUNDING_BITS (7)
static const uint8_t signal_pattern_0[32] = { 0, 1, 2, 3, 2, 3, 4, 5, 4, 5, 6,
7, 6, 7, 8, 9, 0, 1, 2, 3, 2, 3,
4, 5, 4, 5, 6, 7, 6, 7, 8, 9 };
static const uint8_t signal_pattern_1[32] = { 4, 5, 6, 7, 6, 7, 8, 9,
8, 9, 10, 11, 10, 11, 12, 13,
4, 5, 6, 7, 6, 7, 8, 9,
8, 9, 10, 11, 10, 11, 12, 13 };
static const uint8_t signal_pattern_2[32] = { 6, 7, 8, 9, 8, 9, 10, 11,
10, 11, 12, 13, 12, 13, 14, 15,
6, 7, 8, 9, 8, 9, 10, 11,
10, 11, 12, 13, 12, 13, 14, 15 };
static const uint32_t signal_index[8] = { 2, 3, 4, 5, 2, 3, 4, 5 };
// ----------------------------------------------------------------------------- // -----------------------------------------------------------------------------
// Horizontal Filtering // Horizontal Filtering
@ -248,52 +245,30 @@ static INLINE void pack_16_pixels(const __m256i *s0, const __m256i *s1,
x[7] = _mm256_permute2x128_si256(pp[1], pp[5], 0x31); x[7] = _mm256_permute2x128_si256(pp[1], pp[5], 0x31);
} }
static INLINE void pack_pixels_with_format(const uint16_t *src, static INLINE void pack_8x1_pixels(const uint16_t *src, __m256i *x) {
PixelPackFormat fmt, __m256i pp[8];
ptrdiff_t stride, __m256i *x) { __m256i s0;
switch (fmt) { s0 = _mm256_loadu_si256((const __m256i *)src);
case PACK_8x1: { pack_pixels(&s0, pp);
__m256i pp[8]; x[0] = _mm256_permute2x128_si256(pp[0], pp[2], 0x30);
__m256i s0; x[1] = _mm256_permute2x128_si256(pp[1], pp[3], 0x30);
s0 = _mm256_loadu_si256((const __m256i *)src); x[2] = _mm256_permute2x128_si256(pp[2], pp[0], 0x30);
pack_pixels(&s0, pp); x[3] = _mm256_permute2x128_si256(pp[3], pp[1], 0x30);
x[0] = _mm256_permute2x128_si256(pp[0], pp[2], 0x30);
x[1] = _mm256_permute2x128_si256(pp[1], pp[3], 0x30);
x[2] = _mm256_permute2x128_si256(pp[2], pp[0], 0x30);
x[3] = _mm256_permute2x128_si256(pp[3], pp[1], 0x30);
break;
}
case PACK_8x2: {
__m256i s0, s1;
s0 = _mm256_loadu_si256((const __m256i *)src);
s1 = _mm256_loadu_si256((const __m256i *)(src + stride));
pack_16_pixels(&s0, &s1, x);
break;
}
case PACK_16x1: {
__m256i s0, s1;
s0 = _mm256_loadu_si256((const __m256i *)src);
s1 = _mm256_loadu_si256((const __m256i *)(src + 8));
pack_16_pixels(&s0, &s1, x);
break;
}
default: { assert(0); }
}
} }
static INLINE void pack_8x1_pixels(const uint16_t *src, const ptrdiff_t pitch, static INLINE void pack_8x2_pixels(const uint16_t *src, ptrdiff_t stride,
__m256i *x /*x[4]*/) { __m256i *x) {
pack_pixels_with_format(src, PACK_8x1, pitch, x); __m256i s0, s1;
s0 = _mm256_loadu_si256((const __m256i *)src);
s1 = _mm256_loadu_si256((const __m256i *)(src + stride));
pack_16_pixels(&s0, &s1, x);
} }
static INLINE void pack_8x2_pixels(const uint16_t *src, const ptrdiff_t pitch, static INLINE void pack_16x1_pixels(const uint16_t *src, __m256i *x) {
__m256i *x /*x[8]*/) { __m256i s0, s1;
pack_pixels_with_format(src, PACK_8x2, pitch, x); s0 = _mm256_loadu_si256((const __m256i *)src);
} s1 = _mm256_loadu_si256((const __m256i *)(src + 8));
pack_16_pixels(&s0, &s1, x);
static INLINE void pack_16x1_pixels(const uint16_t *src, const ptrdiff_t pitch,
__m256i *x /*x[8]*/) {
pack_pixels_with_format(src, PACK_16x1, pitch, x);
} }
// Note: // Note:
@ -323,51 +298,49 @@ static INLINE void filter_8x1_pixels(const __m256i *sig /*sig[4]*/,
a0 = _mm256_madd_epi16(fil[1], sig[1]); a0 = _mm256_madd_epi16(fil[1], sig[1]);
a1 = _mm256_madd_epi16(fil[2], sig[2]); a1 = _mm256_madd_epi16(fil[2], sig[2]);
const __m256i min = _mm256_min_epi32(a0, a1); {
a = _mm256_add_epi32(a, min); const __m256i min = _mm256_min_epi32(a0, a1);
a = _mm256_add_epi32(a, min);
const __m256i max = _mm256_max_epi32(a0, a1); }
a = _mm256_add_epi32(a, max); {
const __m256i max = _mm256_max_epi32(a0, a1);
const __m256i rounding = _mm256_set1_epi32(1 << (CONV8_ROUNDING_BITS - 1)); a = _mm256_add_epi32(a, max);
a = _mm256_add_epi32(a, rounding); }
*y = _mm256_srai_epi32(a, CONV8_ROUNDING_BITS); {
const __m256i rounding = _mm256_set1_epi32(1 << (CONV8_ROUNDING_BITS - 1));
a = _mm256_add_epi32(a, rounding);
*y = _mm256_srai_epi32(a, CONV8_ROUNDING_BITS);
}
} }
static void write_8x1_pixels(const __m256i *y, const __m256i *z, static INLINE void store_8x1_pixels(const __m256i *y, const __m256i *mask,
const __m256i *mask, uint16_t *dst, uint16_t *dst) {
ptrdiff_t pitch) {
const __m128i a0 = _mm256_castsi256_si128(*y); const __m128i a0 = _mm256_castsi256_si128(*y);
const __m128i a1 = _mm256_extractf128_si256(*y, 1); const __m128i a1 = _mm256_extractf128_si256(*y, 1);
__m128i res = _mm_packus_epi32(a0, a1); __m128i res = _mm_packus_epi32(a0, a1);
(void)z;
(void)pitch;
res = _mm_min_epi16(res, _mm256_castsi256_si128(*mask)); res = _mm_min_epi16(res, _mm256_castsi256_si128(*mask));
_mm_storeu_si128((__m128i *)dst, res); _mm_storeu_si128((__m128i *)dst, res);
} }
static void write_8x2_pixels(const __m256i *y0, const __m256i *y1, static INLINE void store_8x2_pixels(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst, const __m256i *mask, uint16_t *dst,
ptrdiff_t pitch) { ptrdiff_t pitch) {
__m256i a = _mm256_packus_epi32(*y0, *y1); __m256i a = _mm256_packus_epi32(*y0, *y1);
a = _mm256_min_epi16(a, *mask); a = _mm256_min_epi16(a, *mask);
_mm_storeu_si128((__m128i *)dst, _mm256_castsi256_si128(a)); _mm_storeu_si128((__m128i *)dst, _mm256_castsi256_si128(a));
_mm_storeu_si128((__m128i *)(dst + pitch), _mm256_extractf128_si256(a, 1)); _mm_storeu_si128((__m128i *)(dst + pitch), _mm256_extractf128_si256(a, 1));
} }
static void write_16x1_pixels(const __m256i *y0, const __m256i *y1, static INLINE void store_16x1_pixels(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst, const __m256i *mask, uint16_t *dst) {
ptrdiff_t dst_pitch) {
(void)dst_pitch;
__m256i a = _mm256_packus_epi32(*y0, *y1); __m256i a = _mm256_packus_epi32(*y0, *y1);
a = _mm256_min_epi16(a, *mask); a = _mm256_min_epi16(a, *mask);
_mm256_storeu_si256((__m256i *)dst, a); _mm256_storeu_si256((__m256i *)dst, a);
} }
static void filter_block_width8_horiz( static void aom_highbd_filter_block1d8_h8_avx2(
const uint16_t *src_ptr, ptrdiff_t src_pitch, const WritePixels write_8x1, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
const WritePixels write_8x2, uint16_t *dst_ptr, ptrdiff_t dst_pitch, ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
uint32_t height, const int16_t *filter, int bd) {
__m256i signal[8], res0, res1; __m256i signal[8], res0, res1;
const __m256i max = _mm256_set1_epi16((1 << bd) - 1); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
@ -379,32 +352,22 @@ static void filter_block_width8_horiz(
pack_8x2_pixels(src_ptr, src_pitch, signal); pack_8x2_pixels(src_ptr, src_pitch, signal);
filter_8x1_pixels(signal, ff, &res0); filter_8x1_pixels(signal, ff, &res0);
filter_8x1_pixels(&signal[4], ff, &res1); filter_8x1_pixels(&signal[4], ff, &res1);
write_8x2(&res0, &res1, &max, dst_ptr, dst_pitch); store_8x2_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
height -= 2; height -= 2;
src_ptr += src_pitch << 1; src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1; dst_ptr += dst_pitch << 1;
} while (height > 1); } while (height > 1);
if (height > 0) { if (height > 0) {
pack_8x1_pixels(src_ptr, src_pitch, signal); pack_8x1_pixels(src_ptr, signal);
filter_8x1_pixels(signal, ff, &res0); filter_8x1_pixels(signal, ff, &res0);
write_8x1(&res0, &res1, &max, dst_ptr, dst_pitch); store_8x1_pixels(&res0, &max, dst_ptr);
} }
} }
static void aom_highbd_filter_block1d8_h8_avx2( static void aom_highbd_filter_block1d16_h8_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_horiz(src, src_pitch, write_8x1_pixels, write_8x2_pixels,
dst, dst_pitch, height, filter, bd);
}
static void filter_block_width16_horiz(const uint16_t *src_ptr,
ptrdiff_t src_pitch,
const WritePixels write_16x1,
uint16_t *dst_ptr, ptrdiff_t dst_pitch,
uint32_t height, const int16_t *filter,
int bd) {
__m256i signal[8], res0, res1; __m256i signal[8], res0, res1;
const __m256i max = _mm256_set1_epi16((1 << bd) - 1); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
@ -413,23 +376,17 @@ static void filter_block_width16_horiz(const uint16_t *src_ptr,
src_ptr -= 3; src_ptr -= 3;
do { do {
pack_16x1_pixels(src_ptr, src_pitch, signal); pack_16x1_pixels(src_ptr, signal);
filter_8x1_pixels(signal, ff, &res0); filter_8x1_pixels(signal, ff, &res0);
filter_8x1_pixels(&signal[4], ff, &res1); filter_8x1_pixels(&signal[4], ff, &res1);
write_16x1(&res0, &res1, &max, dst_ptr, dst_pitch); store_16x1_pixels(&res0, &res1, &max, dst_ptr);
height -= 1; height -= 1;
src_ptr += src_pitch; src_ptr += src_pitch;
dst_ptr += dst_pitch; dst_ptr += dst_pitch;
} while (height > 0); } while (height > 0);
} }
static void aom_highbd_filter_block1d16_h8_avx2( // -----------------------------------------------------------------------------
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_horiz(src, src_pitch, write_16x1_pixels, dst, dst_pitch,
height, filter, bd);
}
// 2-tap horizontal filtering // 2-tap horizontal filtering
static INLINE void pack_2t_filter(const int16_t *filter, __m256i *f) { static INLINE void pack_2t_filter(const int16_t *filter, __m256i *f) {
@ -493,16 +450,6 @@ static INLINE void filter_16_2t_pixels(const __m256i *sig, const __m256i *f,
*y1 = _mm256_srai_epi32(x1, CONV8_ROUNDING_BITS); *y1 = _mm256_srai_epi32(x1, CONV8_ROUNDING_BITS);
} }
static INLINE void filter_8x2_2t_pixels(const __m256i *sig, const __m256i *f,
__m256i *y0, __m256i *y1) {
filter_16_2t_pixels(sig, f, y0, y1);
}
static INLINE void filter_16x1_2t_pixels(const __m256i *sig, const __m256i *f,
__m256i *y0, __m256i *y1) {
filter_16_2t_pixels(sig, f, y0, y1);
}
static INLINE void filter_8x1_2t_pixels(const __m256i *sig, const __m256i *f, static INLINE void filter_8x1_2t_pixels(const __m256i *sig, const __m256i *f,
__m256i *y0) { __m256i *y0) {
const __m256i rounding = _mm256_set1_epi32(1 << (CONV8_ROUNDING_BITS - 1)); const __m256i rounding = _mm256_set1_epi32(1 << (CONV8_ROUNDING_BITS - 1));
@ -511,10 +458,9 @@ static INLINE void filter_8x1_2t_pixels(const __m256i *sig, const __m256i *f,
*y0 = _mm256_srai_epi32(x0, CONV8_ROUNDING_BITS); *y0 = _mm256_srai_epi32(x0, CONV8_ROUNDING_BITS);
} }
static void filter_block_width8_2t_horiz( static void aom_highbd_filter_block1d8_h2_avx2(
const uint16_t *src_ptr, ptrdiff_t src_pitch, const WritePixels write_8x1, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
const WritePixels write_8x2, uint16_t *dst_ptr, ptrdiff_t dst_pitch, ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
uint32_t height, const int16_t *filter, int bd) {
__m256i signal[2], res0, res1; __m256i signal[2], res0, res1;
const __m256i max = _mm256_set1_epi16((1 << bd) - 1); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
@ -524,8 +470,8 @@ static void filter_block_width8_2t_horiz(
src_ptr -= 3; src_ptr -= 3;
do { do {
pack_8x2_2t_pixels(src_ptr, src_pitch, signal); pack_8x2_2t_pixels(src_ptr, src_pitch, signal);
filter_8x2_2t_pixels(signal, &ff, &res0, &res1); filter_16_2t_pixels(signal, &ff, &res0, &res1);
write_8x2(&res0, &res1, &max, dst_ptr, dst_pitch); store_8x2_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
height -= 2; height -= 2;
src_ptr += src_pitch << 1; src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1; dst_ptr += dst_pitch << 1;
@ -534,24 +480,13 @@ static void filter_block_width8_2t_horiz(
if (height > 0) { if (height > 0) {
pack_8x1_2t_pixels(src_ptr, signal); pack_8x1_2t_pixels(src_ptr, signal);
filter_8x1_2t_pixels(signal, &ff, &res0); filter_8x1_2t_pixels(signal, &ff, &res0);
write_8x1(&res0, &res1, &max, dst_ptr, dst_pitch); store_8x1_pixels(&res0, &max, dst_ptr);
} }
} }
static void aom_highbd_filter_block1d8_h2_avx2( static void aom_highbd_filter_block1d16_h2_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_2t_horiz(src, src_pitch, write_8x1_pixels,
write_8x2_pixels, dst, dst_pitch, height, filter,
bd);
}
static void filter_block_width16_2t_horiz(const uint16_t *src_ptr,
ptrdiff_t src_pitch,
const WritePixels write_16x1,
uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height,
const int16_t *filter, int bd) {
__m256i signal[2], res0, res1; __m256i signal[2], res0, res1;
const __m256i max = _mm256_set1_epi16((1 << bd) - 1); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
@ -561,21 +496,15 @@ static void filter_block_width16_2t_horiz(const uint16_t *src_ptr,
src_ptr -= 3; src_ptr -= 3;
do { do {
pack_16x1_2t_pixels(src_ptr, signal); pack_16x1_2t_pixels(src_ptr, signal);
filter_16x1_2t_pixels(signal, &ff, &res0, &res1); filter_16_2t_pixels(signal, &ff, &res0, &res1);
write_16x1(&res0, &res1, &max, dst_ptr, dst_pitch); store_16x1_pixels(&res0, &res1, &max, dst_ptr);
height -= 1; height -= 1;
src_ptr += src_pitch; src_ptr += src_pitch;
dst_ptr += dst_pitch; dst_ptr += dst_pitch;
} while (height > 0); } while (height > 0);
} }
static void aom_highbd_filter_block1d16_h2_avx2( // -----------------------------------------------------------------------------
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_2t_horiz(src, src_pitch, write_16x1_pixels, dst,
dst_pitch, height, filter, bd);
}
// Vertical Filtering // Vertical Filtering
static void pack_8x9_init(const uint16_t *src, ptrdiff_t pitch, __m256i *sig) { static void pack_8x9_init(const uint16_t *src, ptrdiff_t pitch, __m256i *sig) {
@ -638,22 +567,9 @@ static INLINE void update_pixels(__m256i *sig) {
} }
} }
static INLINE void write_8x1_pixels_ver(const __m256i *y0, const __m256i *y1, static void aom_highbd_filter_block1d8_v8_avx2(
const __m256i *mask, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t pitch) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
(void)pitch;
const __m128i v0 = _mm256_castsi256_si128(*y0);
const __m128i v1 = _mm256_castsi256_si128(*y1);
__m128i p = _mm_packus_epi32(v0, v1);
p = _mm_min_epi16(p, _mm256_castsi256_si128(*mask));
_mm_storeu_si128((__m128i *)dst, p);
}
static void filter_block_width8_vert(const uint16_t *src_ptr,
ptrdiff_t src_pitch, WritePixels write_8x1,
WritePixels write_8x2, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height,
const int16_t *filter, int bd) {
__m256i signal[9], res0, res1; __m256i signal[9], res0, res1;
const __m256i max = _mm256_set1_epi16((1 << bd) - 1); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
@ -666,27 +582,13 @@ static void filter_block_width8_vert(const uint16_t *src_ptr,
pack_8x9_pixels(src_ptr, src_pitch, signal); pack_8x9_pixels(src_ptr, src_pitch, signal);
filter_8x9_pixels(signal, ff, &res0, &res1); filter_8x9_pixels(signal, ff, &res0, &res1);
write_8x2(&res0, &res1, &max, dst_ptr, dst_pitch); store_8x2_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
update_pixels(signal); update_pixels(signal);
src_ptr += src_pitch << 1; src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1; dst_ptr += dst_pitch << 1;
height -= 2; height -= 2;
} while (height > 1); } while (height > 0);
if (height > 0) {
pack_8x9_pixels(src_ptr, src_pitch, signal);
filter_8x9_pixels(signal, ff, &res0, &res1);
write_8x1(&res0, &res1, &max, dst_ptr, dst_pitch);
}
}
static void aom_highbd_filter_block1d8_v8_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_vert(src, src_pitch, write_8x1_pixels_ver,
write_8x2_pixels, dst, dst_pitch, height, filter,
bd);
} }
static void pack_16x9_init(const uint16_t *src, ptrdiff_t pitch, __m256i *sig) { static void pack_16x9_init(const uint16_t *src, ptrdiff_t pitch, __m256i *sig) {
@ -770,13 +672,15 @@ static INLINE void filter_16x9_pixels(const __m256i *sig, const __m256i *f,
filter_8x1_pixels(&sig[i << 2], f, &res[i]); filter_8x1_pixels(&sig[i << 2], f, &res[i]);
} }
const __m256i l0l1 = _mm256_packus_epi32(res[0], res[1]); {
const __m256i h0h1 = _mm256_packus_epi32(res[2], res[3]); const __m256i l0l1 = _mm256_packus_epi32(res[0], res[1]);
*y0 = _mm256_permute2x128_si256(l0l1, h0h1, 0x20); const __m256i h0h1 = _mm256_packus_epi32(res[2], res[3]);
*y1 = _mm256_permute2x128_si256(l0l1, h0h1, 0x31); *y0 = _mm256_permute2x128_si256(l0l1, h0h1, 0x20);
*y1 = _mm256_permute2x128_si256(l0l1, h0h1, 0x31);
}
} }
static INLINE void write_16x2_pixels(const __m256i *y0, const __m256i *y1, static INLINE void store_16x2_pixels(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst, const __m256i *mask, uint16_t *dst,
ptrdiff_t pitch) { ptrdiff_t pitch) {
__m256i p = _mm256_min_epi16(*y0, *mask); __m256i p = _mm256_min_epi16(*y0, *mask);
@ -785,26 +689,14 @@ static INLINE void write_16x2_pixels(const __m256i *y0, const __m256i *y1,
_mm256_storeu_si256((__m256i *)(dst + pitch), p); _mm256_storeu_si256((__m256i *)(dst + pitch), p);
} }
static INLINE void write_16x1_pixels_ver(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst,
ptrdiff_t pitch) {
(void)y1;
(void)pitch;
const __m256i p = _mm256_min_epi16(*y0, *mask);
_mm256_storeu_si256((__m256i *)dst, p);
}
static void update_16x9_pixels(__m256i *sig) { static void update_16x9_pixels(__m256i *sig) {
update_pixels(&sig[0]); update_pixels(&sig[0]);
update_pixels(&sig[8]); update_pixels(&sig[8]);
} }
static void filter_block_width16_vert(const uint16_t *src_ptr, static void aom_highbd_filter_block1d16_v8_avx2(
ptrdiff_t src_pitch, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
WritePixels write_16x1, ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
WritePixels write_16x2, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height,
const int16_t *filter, int bd) {
__m256i signal[17], res0, res1; __m256i signal[17], res0, res1;
const __m256i max = _mm256_set1_epi16((1 << bd) - 1); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
@ -816,29 +708,16 @@ static void filter_block_width16_vert(const uint16_t *src_ptr,
do { do {
pack_16x9_pixels(src_ptr, src_pitch, signal); pack_16x9_pixels(src_ptr, src_pitch, signal);
filter_16x9_pixels(signal, ff, &res0, &res1); filter_16x9_pixels(signal, ff, &res0, &res1);
write_16x2(&res0, &res1, &max, dst_ptr, dst_pitch); store_16x2_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
update_16x9_pixels(signal); update_16x9_pixels(signal);
src_ptr += src_pitch << 1; src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1; dst_ptr += dst_pitch << 1;
height -= 2; height -= 2;
} while (height > 1); } while (height > 0);
if (height > 0) {
pack_16x9_pixels(src_ptr, src_pitch, signal);
filter_16x9_pixels(signal, ff, &res0, &res1);
write_16x1(&res0, &res1, &max, dst_ptr, dst_pitch);
}
}
static void aom_highbd_filter_block1d16_v8_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_vert(src, src_pitch, write_16x1_pixels_ver,
write_16x2_pixels, dst, dst_pitch, height, filter,
bd);
} }
// -----------------------------------------------------------------------------
// 2-tap vertical filtering // 2-tap vertical filtering
static void pack_16x2_init(const uint16_t *src, __m256i *sig) { static void pack_16x2_init(const uint16_t *src, __m256i *sig) {
@ -859,12 +738,9 @@ static INLINE void filter_16x2_2t_pixels(const __m256i *sig, const __m256i *f,
filter_16_2t_pixels(sig, f, y0, y1); filter_16_2t_pixels(sig, f, y0, y1);
} }
static void filter_block_width16_2t_vert(const uint16_t *src_ptr, static void aom_highbd_filter_block1d16_v2_avx2(
ptrdiff_t src_pitch, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
WritePixels write_16x1, ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
uint16_t *dst_ptr, ptrdiff_t dst_pitch,
uint32_t height, const int16_t *filter,
int bd) {
__m256i signal[3], res0, res1; __m256i signal[3], res0, res1;
const __m256i max = _mm256_set1_epi16((1 << bd) - 1); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
__m256i ff; __m256i ff;
@ -875,7 +751,7 @@ static void filter_block_width16_2t_vert(const uint16_t *src_ptr,
do { do {
pack_16x2_2t_pixels(src_ptr, src_pitch, signal); pack_16x2_2t_pixels(src_ptr, src_pitch, signal);
filter_16x2_2t_pixels(signal, &ff, &res0, &res1); filter_16x2_2t_pixels(signal, &ff, &res0, &res1);
write_16x1(&res0, &res1, &max, dst_ptr, dst_pitch); store_16x1_pixels(&res0, &res1, &max, dst_ptr);
src_ptr += src_pitch; src_ptr += src_pitch;
dst_ptr += dst_pitch; dst_ptr += dst_pitch;
@ -883,13 +759,6 @@ static void filter_block_width16_2t_vert(const uint16_t *src_ptr,
} while (height > 0); } while (height > 0);
} }
static void aom_highbd_filter_block1d16_v2_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_2t_vert(src, src_pitch, write_16x1_pixels, dst,
dst_pitch, height, filter, bd);
}
static INLINE void pack_8x1_2t_filter(const int16_t *filter, __m128i *f) { static INLINE void pack_8x1_2t_filter(const int16_t *filter, __m128i *f) {
const __m128i h = _mm_loadu_si128((const __m128i *)filter); const __m128i h = _mm_loadu_si128((const __m128i *)filter);
const __m128i p = _mm_set1_epi32(0x09080706); const __m128i p = _mm_set1_epi32(0x09080706);
@ -920,22 +789,16 @@ static INLINE void filter_8_2t_pixels(const __m128i *sig, const __m128i *f,
*y1 = _mm_srai_epi32(x1, CONV8_ROUNDING_BITS); *y1 = _mm_srai_epi32(x1, CONV8_ROUNDING_BITS);
} }
static void write_8x1_2t_pixels_ver(const __m128i *y0, const __m128i *y1, static INLINE void store_8x1_2t_pixels_ver(const __m128i *y0, const __m128i *y1,
const __m128i *mask, uint16_t *dst) { const __m128i *mask, uint16_t *dst) {
__m128i res = _mm_packus_epi32(*y0, *y1); __m128i res = _mm_packus_epi32(*y0, *y1);
res = _mm_min_epi16(res, *mask); res = _mm_min_epi16(res, *mask);
_mm_storeu_si128((__m128i *)dst, res); _mm_storeu_si128((__m128i *)dst, res);
} }
typedef void (*Write8Pixels)(const __m128i *y0, const __m128i *y1, static void aom_highbd_filter_block1d8_v2_avx2(
const __m128i *mask, uint16_t *dst); const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
static void filter_block_width8_2t_vert(const uint16_t *src_ptr,
ptrdiff_t src_pitch,
Write8Pixels write_8x1,
uint16_t *dst_ptr, ptrdiff_t dst_pitch,
uint32_t height, const int16_t *filter,
int bd) {
__m128i signal[3], res0, res1; __m128i signal[3], res0, res1;
const __m128i max = _mm_set1_epi16((1 << bd) - 1); const __m128i max = _mm_set1_epi16((1 << bd) - 1);
__m128i ff; __m128i ff;
@ -946,7 +809,7 @@ static void filter_block_width8_2t_vert(const uint16_t *src_ptr,
do { do {
pack_8x2_2t_pixels_ver(src_ptr, src_pitch, signal); pack_8x2_2t_pixels_ver(src_ptr, src_pitch, signal);
filter_8_2t_pixels(signal, &ff, &res0, &res1); filter_8_2t_pixels(signal, &ff, &res0, &res1);
write_8x1(&res0, &res1, &max, dst_ptr); store_8x1_2t_pixels_ver(&res0, &res1, &max, dst_ptr);
src_ptr += src_pitch; src_ptr += src_pitch;
dst_ptr += dst_pitch; dst_ptr += dst_pitch;
@ -954,20 +817,10 @@ static void filter_block_width8_2t_vert(const uint16_t *src_ptr,
} while (height > 0); } while (height > 0);
} }
static void aom_highbd_filter_block1d8_v2_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_2t_vert(src, src_pitch, write_8x1_2t_pixels_ver, dst,
dst_pitch, height, filter, bd);
}
// Calculation with averaging the input pixels // Calculation with averaging the input pixels
static void write_8x1_avg_pixels(const __m256i *y0, const __m256i *y1, static INLINE void store_8x1_avg_pixels(const __m256i *y0, const __m256i *mask,
const __m256i *mask, uint16_t *dst, uint16_t *dst) {
ptrdiff_t pitch) {
(void)y1;
(void)pitch;
const __m128i a0 = _mm256_castsi256_si128(*y0); const __m128i a0 = _mm256_castsi256_si128(*y0);
const __m128i a1 = _mm256_extractf128_si256(*y0, 1); const __m128i a1 = _mm256_extractf128_si256(*y0, 1);
__m128i res = _mm_packus_epi32(a0, a1); __m128i res = _mm_packus_epi32(a0, a1);
@ -977,9 +830,9 @@ static void write_8x1_avg_pixels(const __m256i *y0, const __m256i *y1,
_mm_storeu_si128((__m128i *)dst, res); _mm_storeu_si128((__m128i *)dst, res);
} }
static void write_8x2_avg_pixels(const __m256i *y0, const __m256i *y1, static INLINE void store_8x2_avg_pixels(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst, const __m256i *mask, uint16_t *dst,
ptrdiff_t pitch) { ptrdiff_t pitch) {
__m256i a = _mm256_packus_epi32(*y0, *y1); __m256i a = _mm256_packus_epi32(*y0, *y1);
const __m128i pix0 = _mm_loadu_si128((const __m128i *)dst); const __m128i pix0 = _mm_loadu_si128((const __m128i *)dst);
const __m128i pix1 = _mm_loadu_si128((const __m128i *)(dst + pitch)); const __m128i pix1 = _mm_loadu_si128((const __m128i *)(dst + pitch));
@ -991,10 +844,8 @@ static void write_8x2_avg_pixels(const __m256i *y0, const __m256i *y1,
_mm_storeu_si128((__m128i *)(dst + pitch), _mm256_extractf128_si256(a, 1)); _mm_storeu_si128((__m128i *)(dst + pitch), _mm256_extractf128_si256(a, 1));
} }
static void write_16x1_avg_pixels(const __m256i *y0, const __m256i *y1, static INLINE void store_16x1_avg_pixels(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst, const __m256i *mask, uint16_t *dst) {
ptrdiff_t pitch) {
(void)pitch;
__m256i a = _mm256_packus_epi32(*y0, *y1); __m256i a = _mm256_packus_epi32(*y0, *y1);
const __m256i pix = _mm256_loadu_si256((const __m256i *)dst); const __m256i pix = _mm256_loadu_si256((const __m256i *)dst);
a = _mm256_min_epi16(a, *mask); a = _mm256_min_epi16(a, *mask);
@ -1002,21 +853,7 @@ static void write_16x1_avg_pixels(const __m256i *y0, const __m256i *y1,
_mm256_storeu_si256((__m256i *)dst, a); _mm256_storeu_si256((__m256i *)dst, a);
} }
static INLINE void write_8x1_avg_pixels_ver(const __m256i *y0, static INLINE void store_16x2_avg_pixels(const __m256i *y0, const __m256i *y1,
const __m256i *y1,
const __m256i *mask, uint16_t *dst,
ptrdiff_t pitch) {
(void)pitch;
const __m128i v0 = _mm256_castsi256_si128(*y0);
const __m128i v1 = _mm256_castsi256_si128(*y1);
__m128i p = _mm_packus_epi32(v0, v1);
const __m128i pix = _mm_loadu_si128((const __m128i *)dst);
p = _mm_min_epi16(p, _mm256_castsi256_si128(*mask));
p = _mm_avg_epu16(p, pix);
_mm_storeu_si128((__m128i *)dst, p);
}
static INLINE void write_16x2_avg_pixels(const __m256i *y0, const __m256i *y1,
const __m256i *mask, uint16_t *dst, const __m256i *mask, uint16_t *dst,
ptrdiff_t pitch) { ptrdiff_t pitch) {
const __m256i pix0 = _mm256_loadu_si256((const __m256i *)dst); const __m256i pix0 = _mm256_loadu_si256((const __m256i *)dst);
@ -1030,20 +867,10 @@ static INLINE void write_16x2_avg_pixels(const __m256i *y0, const __m256i *y1,
_mm256_storeu_si256((__m256i *)(dst + pitch), p); _mm256_storeu_si256((__m256i *)(dst + pitch), p);
} }
static INLINE void write_16x1_avg_pixels_ver(const __m256i *y0, static INLINE void store_8x1_2t_avg_pixels_ver(const __m128i *y0,
const __m256i *y1, const __m128i *y1,
const __m256i *mask, uint16_t *dst, const __m128i *mask,
ptrdiff_t pitch) { uint16_t *dst) {
(void)y1;
(void)pitch;
__m256i p = _mm256_min_epi16(*y0, *mask);
const __m256i pix = _mm256_loadu_si256((const __m256i *)dst);
p = _mm256_avg_epu16(p, pix);
_mm256_storeu_si256((__m256i *)dst, p);
}
static void write_8x1_2t_avg_pixels_ver(const __m128i *y0, const __m128i *y1,
const __m128i *mask, uint16_t *dst) {
__m128i res = _mm_packus_epi32(*y0, *y1); __m128i res = _mm_packus_epi32(*y0, *y1);
const __m128i pix = _mm_loadu_si128((const __m128i *)dst); const __m128i pix = _mm_loadu_si128((const __m128i *)dst);
res = _mm_min_epi16(res, *mask); res = _mm_min_epi16(res, *mask);
@ -1052,96 +879,229 @@ static void write_8x1_2t_avg_pixels_ver(const __m128i *y0, const __m128i *y1,
} }
static void aom_highbd_filter_block1d8_h8_avg_avx2( static void aom_highbd_filter_block1d8_h8_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_horiz(src, src_pitch, write_8x1_avg_pixels, __m256i signal[8], res0, res1;
write_8x2_avg_pixels, dst, dst_pitch, height, const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
filter, bd);
__m256i ff[4];
pack_filters(filter, ff);
src_ptr -= 3;
do {
pack_8x2_pixels(src_ptr, src_pitch, signal);
filter_8x1_pixels(signal, ff, &res0);
filter_8x1_pixels(&signal[4], ff, &res1);
store_8x2_avg_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
height -= 2;
src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1;
} while (height > 1);
if (height > 0) {
pack_8x1_pixels(src_ptr, signal);
filter_8x1_pixels(signal, ff, &res0);
store_8x1_avg_pixels(&res0, &max, dst_ptr);
}
} }
static void aom_highbd_filter_block1d16_h8_avg_avx2( static void aom_highbd_filter_block1d16_h8_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_horiz(src, src_pitch, write_16x1_avg_pixels, dst, __m256i signal[8], res0, res1;
dst_pitch, height, filter, bd); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
__m256i ff[4];
pack_filters(filter, ff);
src_ptr -= 3;
do {
pack_16x1_pixels(src_ptr, signal);
filter_8x1_pixels(signal, ff, &res0);
filter_8x1_pixels(&signal[4], ff, &res1);
store_16x1_avg_pixels(&res0, &res1, &max, dst_ptr);
height -= 1;
src_ptr += src_pitch;
dst_ptr += dst_pitch;
} while (height > 0);
} }
static void aom_highbd_filter_block1d8_v8_avg_avx2( static void aom_highbd_filter_block1d8_v8_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_vert(src, src_pitch, write_8x1_avg_pixels_ver, __m256i signal[9], res0, res1;
write_8x2_avg_pixels, dst, dst_pitch, height, filter, const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
bd);
__m256i ff[4];
pack_filters(filter, ff);
pack_8x9_init(src_ptr, src_pitch, signal);
do {
pack_8x9_pixels(src_ptr, src_pitch, signal);
filter_8x9_pixels(signal, ff, &res0, &res1);
store_8x2_avg_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
update_pixels(signal);
src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1;
height -= 2;
} while (height > 0);
} }
static void aom_highbd_filter_block1d16_v8_avg_avx2( static void aom_highbd_filter_block1d16_v8_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_vert(src, src_pitch, write_16x1_avg_pixels_ver, __m256i signal[17], res0, res1;
write_16x2_avg_pixels, dst, dst_pitch, height, const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
filter, bd);
__m256i ff[4];
pack_filters(filter, ff);
pack_16x9_init(src_ptr, src_pitch, signal);
do {
pack_16x9_pixels(src_ptr, src_pitch, signal);
filter_16x9_pixels(signal, ff, &res0, &res1);
store_16x2_avg_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
update_16x9_pixels(signal);
src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1;
height -= 2;
} while (height > 0);
} }
// 2-tap averaging
static void aom_highbd_filter_block1d8_h2_avg_avx2( static void aom_highbd_filter_block1d8_h2_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_2t_horiz(src, src_pitch, write_8x1_avg_pixels, __m256i signal[2], res0, res1;
write_8x2_avg_pixels, dst, dst_pitch, height, const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
filter, bd);
__m256i ff;
pack_2t_filter(filter, &ff);
src_ptr -= 3;
do {
pack_8x2_2t_pixels(src_ptr, src_pitch, signal);
filter_16_2t_pixels(signal, &ff, &res0, &res1);
store_8x2_avg_pixels(&res0, &res1, &max, dst_ptr, dst_pitch);
height -= 2;
src_ptr += src_pitch << 1;
dst_ptr += dst_pitch << 1;
} while (height > 1);
if (height > 0) {
pack_8x1_2t_pixels(src_ptr, signal);
filter_8x1_2t_pixels(signal, &ff, &res0);
store_8x1_avg_pixels(&res0, &max, dst_ptr);
}
} }
static void aom_highbd_filter_block1d16_h2_avg_avx2( static void aom_highbd_filter_block1d16_h2_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_2t_horiz(src, src_pitch, write_16x1_avg_pixels, dst, __m256i signal[2], res0, res1;
dst_pitch, height, filter, bd); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
__m256i ff;
pack_2t_filter(filter, &ff);
src_ptr -= 3;
do {
pack_16x1_2t_pixels(src_ptr, signal);
filter_16_2t_pixels(signal, &ff, &res0, &res1);
store_16x1_avg_pixels(&res0, &res1, &max, dst_ptr);
height -= 1;
src_ptr += src_pitch;
dst_ptr += dst_pitch;
} while (height > 0);
} }
static void aom_highbd_filter_block1d16_v2_avg_avx2( static void aom_highbd_filter_block1d16_v2_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width16_2t_vert(src, src_pitch, write_16x1_avg_pixels, dst, __m256i signal[3], res0, res1;
dst_pitch, height, filter, bd); const __m256i max = _mm256_set1_epi16((1 << bd) - 1);
__m256i ff;
pack_2t_filter(filter, &ff);
pack_16x2_init(src_ptr, signal);
do {
pack_16x2_2t_pixels(src_ptr, src_pitch, signal);
filter_16x2_2t_pixels(signal, &ff, &res0, &res1);
store_16x1_avg_pixels(&res0, &res1, &max, dst_ptr);
src_ptr += src_pitch;
dst_ptr += dst_pitch;
height -= 1;
} while (height > 0);
} }
static void aom_highbd_filter_block1d8_v2_avg_avx2( static void aom_highbd_filter_block1d8_v2_avg_avx2(
const uint16_t *src, ptrdiff_t src_pitch, uint16_t *dst, const uint16_t *src_ptr, ptrdiff_t src_pitch, uint16_t *dst_ptr,
ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) { ptrdiff_t dst_pitch, uint32_t height, const int16_t *filter, int bd) {
filter_block_width8_2t_vert(src, src_pitch, write_8x1_2t_avg_pixels_ver, dst, __m128i signal[3], res0, res1;
dst_pitch, height, filter, bd); const __m128i max = _mm_set1_epi16((1 << bd) - 1);
__m128i ff;
pack_8x1_2t_filter(filter, &ff);
pack_8x2_init(src_ptr, signal);
do {
pack_8x2_2t_pixels_ver(src_ptr, src_pitch, signal);
filter_8_2t_pixels(signal, &ff, &res0, &res1);
store_8x1_2t_avg_pixels_ver(&res0, &res1, &max, dst_ptr);
src_ptr += src_pitch;
dst_ptr += dst_pitch;
height -= 1;
} while (height > 0);
} }
typedef void HbdFilter1dFunc(const uint16_t *, ptrdiff_t, uint16_t *, ptrdiff_t, void aom_highbd_filter_block1d4_h8_sse2(const uint16_t *, ptrdiff_t, uint16_t *,
uint32_t, const int16_t *, int); ptrdiff_t, uint32_t, const int16_t *,
int);
#define HIGHBD_FUNC(width, dir, avg, opt) \ void aom_highbd_filter_block1d4_h2_sse2(const uint16_t *, ptrdiff_t, uint16_t *,
aom_highbd_filter_block1d##width##_##dir##_##avg##opt ptrdiff_t, uint32_t, const int16_t *,
int);
HbdFilter1dFunc HIGHBD_FUNC(4, h8, , sse2); void aom_highbd_filter_block1d4_v8_sse2(const uint16_t *, ptrdiff_t, uint16_t *,
HbdFilter1dFunc HIGHBD_FUNC(4, h2, , sse2); ptrdiff_t, uint32_t, const int16_t *,
HbdFilter1dFunc HIGHBD_FUNC(4, v8, , sse2); int);
HbdFilter1dFunc HIGHBD_FUNC(4, v2, , sse2); void aom_highbd_filter_block1d4_v2_sse2(const uint16_t *, ptrdiff_t, uint16_t *,
ptrdiff_t, uint32_t, const int16_t *,
#define aom_highbd_filter_block1d4_h8_avx2 HIGHBD_FUNC(4, h8, , sse2) int);
#define aom_highbd_filter_block1d4_h2_avx2 HIGHBD_FUNC(4, h2, , sse2) #define aom_highbd_filter_block1d4_h8_avx2 aom_highbd_filter_block1d4_h8_sse2
#define aom_highbd_filter_block1d4_v8_avx2 HIGHBD_FUNC(4, v8, , sse2) #define aom_highbd_filter_block1d4_h2_avx2 aom_highbd_filter_block1d4_h2_sse2
#define aom_highbd_filter_block1d4_v2_avx2 HIGHBD_FUNC(4, v2, , sse2) #define aom_highbd_filter_block1d4_v8_avx2 aom_highbd_filter_block1d4_v8_sse2
#define aom_highbd_filter_block1d4_v2_avx2 aom_highbd_filter_block1d4_v2_sse2
HIGH_FUN_CONV_1D(horiz, x_step_q4, filter_x, h, src, , avx2); HIGH_FUN_CONV_1D(horiz, x_step_q4, filter_x, h, src, , avx2);
HIGH_FUN_CONV_1D(vert, y_step_q4, filter_y, v, src - src_stride * 3, , avx2); HIGH_FUN_CONV_1D(vert, y_step_q4, filter_y, v, src - src_stride * 3, , avx2);
HIGH_FUN_CONV_2D(, avx2); HIGH_FUN_CONV_2D(, avx2);
HbdFilter1dFunc HIGHBD_FUNC(4, h8, avg_, sse2); void aom_highbd_filter_block1d4_h8_avg_sse2(const uint16_t *, ptrdiff_t,
HbdFilter1dFunc HIGHBD_FUNC(4, h2, avg_, sse2); uint16_t *, ptrdiff_t, uint32_t,
HbdFilter1dFunc HIGHBD_FUNC(4, v8, avg_, sse2); const int16_t *, int);
HbdFilter1dFunc HIGHBD_FUNC(4, v2, avg_, sse2); void aom_highbd_filter_block1d4_h2_avg_sse2(const uint16_t *, ptrdiff_t,
uint16_t *, ptrdiff_t, uint32_t,
#define aom_highbd_filter_block1d4_h8_avg_avx2 HIGHBD_FUNC(4, h8, avg_, sse2) const int16_t *, int);
#define aom_highbd_filter_block1d4_h2_avg_avx2 HIGHBD_FUNC(4, h2, avg_, sse2) void aom_highbd_filter_block1d4_v8_avg_sse2(const uint16_t *, ptrdiff_t,
#define aom_highbd_filter_block1d4_v8_avg_avx2 HIGHBD_FUNC(4, v8, avg_, sse2) uint16_t *, ptrdiff_t, uint32_t,
#define aom_highbd_filter_block1d4_v2_avg_avx2 HIGHBD_FUNC(4, v2, avg_, sse2) const int16_t *, int);
void aom_highbd_filter_block1d4_v2_avg_sse2(const uint16_t *, ptrdiff_t,
uint16_t *, ptrdiff_t, uint32_t,
const int16_t *, int);
#define aom_highbd_filter_block1d4_h8_avg_avx2 \
aom_highbd_filter_block1d4_h8_avg_sse2
#define aom_highbd_filter_block1d4_h2_avg_avx2 \
aom_highbd_filter_block1d4_h2_avg_sse2
#define aom_highbd_filter_block1d4_v8_avg_avx2 \
aom_highbd_filter_block1d4_v8_avg_sse2
#define aom_highbd_filter_block1d4_v2_avg_avx2 \
aom_highbd_filter_block1d4_v2_avg_sse2
HIGH_FUN_CONV_1D(avg_horiz, x_step_q4, filter_x, h, src, avg_, avx2); HIGH_FUN_CONV_1D(avg_horiz, x_step_q4, filter_x, h, src, avg_, avx2);
HIGH_FUN_CONV_1D(avg_vert, y_step_q4, filter_y, v, src - src_stride * 3, avg_, HIGH_FUN_CONV_1D(avg_vert, y_step_q4, filter_y, v, src - src_stride * 3, avg_,

1238
third_party/aom/aom_dsp/x86/inv_txfm_avx2.c vendored Normal file
View file

File diff suppressed because it is too large Load diff

80
third_party/aom/aom_dsp/x86/inv_txfm_common_avx2.h vendored Normal file
View file

@ -0,0 +1,80 @@
/*
* Copyright (c) 2017, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#ifndef AOM_DSP_X86_INV_TXFM_COMMON_AVX2_H
#define AOM_DSP_X86_INV_TXFM_COMMON_AVX2_H
#include <immintrin.h>
#include "aom_dsp/txfm_common.h"
#include "aom_dsp/x86/txfm_common_avx2.h"
static INLINE void load_coeff(const tran_low_t *coeff, __m256i *in) {
#if CONFIG_HIGHBITDEPTH
*in = _mm256_setr_epi16(
(int16_t)coeff[0], (int16_t)coeff[1], (int16_t)coeff[2],
(int16_t)coeff[3], (int16_t)coeff[4], (int16_t)coeff[5],
(int16_t)coeff[6], (int16_t)coeff[7], (int16_t)coeff[8],
(int16_t)coeff[9], (int16_t)coeff[10], (int16_t)coeff[11],
(int16_t)coeff[12], (int16_t)coeff[13], (int16_t)coeff[14],
(int16_t)coeff[15]);
#else
*in = _mm256_loadu_si256((const __m256i *)coeff);
#endif
}
static INLINE void load_buffer_16x16(const tran_low_t *coeff, __m256i *in) {
int i = 0;
while (i < 16) {
load_coeff(coeff + (i << 4), &in[i]);
i += 1;
}
}
static INLINE void recon_and_store(const __m256i *res, uint8_t *output) {
const __m128i zero = _mm_setzero_si128();
__m128i x = _mm_loadu_si128((__m128i const *)output);
__m128i p0 = _mm_unpacklo_epi8(x, zero);
__m128i p1 = _mm_unpackhi_epi8(x, zero);
p0 = _mm_add_epi16(p0, _mm256_castsi256_si128(*res));
p1 = _mm_add_epi16(p1, _mm256_extractf128_si256(*res, 1));
x = _mm_packus_epi16(p0, p1);
_mm_storeu_si128((__m128i *)output, x);
}
#define IDCT_ROUNDING_POS (6)
static INLINE void store_buffer_16xN(__m256i *in, const int stride,
uint8_t *output, int num) {
const __m256i rounding = _mm256_set1_epi16(1 << (IDCT_ROUNDING_POS - 1));
int i = 0;
while (i < num) {
in[i] = _mm256_adds_epi16(in[i], rounding);
in[i] = _mm256_srai_epi16(in[i], IDCT_ROUNDING_POS);
recon_and_store(&in[i], output + i * stride);
i += 1;
}
}
static INLINE void unpack_butter_fly(const __m256i *a0, const __m256i *a1,
const __m256i *c0, const __m256i *c1,
__m256i *b0, __m256i *b1) {
__m256i x0, x1;
x0 = _mm256_unpacklo_epi16(*a0, *a1);
x1 = _mm256_unpackhi_epi16(*a0, *a1);
*b0 = butter_fly(&x0, &x1, c0);
*b1 = butter_fly(&x0, &x1, c1);
}
void av1_idct16_avx2(__m256i *in);
#endif // AOM_DSP_X86_INV_TXFM_COMMON_AVX2_H

103
third_party/aom/aom_dsp/x86/inv_txfm_sse2.c vendored
View file

@ -3628,4 +3628,107 @@ void aom_highbd_idct4x4_16_add_sse2(const tran_low_t *input, uint8_t *dest8,
} }
} }
void aom_highbd_idct8x8_10_add_sse2(const tran_low_t *input, uint8_t *dest8,
int stride, int bd) {
tran_low_t out[8 * 8] = { 0 };
tran_low_t *outptr = out;
int i, j, test;
__m128i inptr[8];
__m128i min_input, max_input, temp1, temp2, sign_bits;
uint16_t *dest = CONVERT_TO_SHORTPTR(dest8);
const __m128i zero = _mm_set1_epi16(0);
const __m128i sixteen = _mm_set1_epi16(16);
const __m128i max = _mm_set1_epi16(6201);
const __m128i min = _mm_set1_epi16(-6201);
int optimised_cols = 0;
// Load input into __m128i & pack to 16 bits
for (i = 0; i < 8; i++) {
temp1 = _mm_loadu_si128((const __m128i *)(input + 8 * i));
temp2 = _mm_loadu_si128((const __m128i *)(input + 8 * i + 4));
inptr[i] = _mm_packs_epi32(temp1, temp2);
}
// Find the min & max for the row transform
// only first 4 row has non-zero coefs
max_input = _mm_max_epi16(inptr[0], inptr[1]);
min_input = _mm_min_epi16(inptr[0], inptr[1]);
for (i = 2; i < 4; i++) {
max_input = _mm_max_epi16(max_input, inptr[i]);
min_input = _mm_min_epi16(min_input, inptr[i]);
}
max_input = _mm_cmpgt_epi16(max_input, max);
min_input = _mm_cmplt_epi16(min_input, min);
temp1 = _mm_or_si128(max_input, min_input);
test = _mm_movemask_epi8(temp1);
if (!test) {
// Do the row transform
aom_idct8_sse2(inptr);
// Find the min & max for the column transform
// N.B. Only first 4 cols contain non-zero coeffs
max_input = _mm_max_epi16(inptr[0], inptr[1]);
min_input = _mm_min_epi16(inptr[0], inptr[1]);
for (i = 2; i < 8; i++) {
max_input = _mm_max_epi16(max_input, inptr[i]);
min_input = _mm_min_epi16(min_input, inptr[i]);
}
max_input = _mm_cmpgt_epi16(max_input, max);
min_input = _mm_cmplt_epi16(min_input, min);
temp1 = _mm_or_si128(max_input, min_input);
test = _mm_movemask_epi8(temp1);
if (test) {
// Use fact only first 4 rows contain non-zero coeffs
array_transpose_4X8(inptr, inptr);
for (i = 0; i < 4; i++) {
sign_bits = _mm_cmplt_epi16(inptr[i], zero);
temp1 = _mm_unpackhi_epi16(inptr[i], sign_bits);
temp2 = _mm_unpacklo_epi16(inptr[i], sign_bits);
_mm_storeu_si128((__m128i *)(outptr + 4 * (2 * i + 1)), temp1);
_mm_storeu_si128((__m128i *)(outptr + 4 * (2 * i)), temp2);
}
} else {
// Set to use the optimised transform for the column
optimised_cols = 1;
}
} else {
// Run the un-optimised row transform
for (i = 0; i < 4; ++i) {
aom_highbd_idct8_c(input, outptr, bd);
input += 8;
outptr += 8;
}
}
if (optimised_cols) {
aom_idct8_sse2(inptr);
// Final round & shift and Reconstruction and Store
{
__m128i d[8];
for (i = 0; i < 8; i++) {
inptr[i] = _mm_add_epi16(inptr[i], sixteen);
d[i] = _mm_loadu_si128((const __m128i *)(dest + stride * i));
inptr[i] = _mm_srai_epi16(inptr[i], 5);
d[i] = clamp_high_sse2(_mm_adds_epi16(d[i], inptr[i]), bd);
// Store
_mm_storeu_si128((__m128i *)(dest + stride * i), d[i]);
}
}
} else {
// Run the un-optimised column transform
tran_low_t temp_in[8], temp_out[8];
for (i = 0; i < 8; ++i) {
for (j = 0; j < 8; ++j) temp_in[j] = out[j * 8 + i];
aom_highbd_idct8_c(temp_in, temp_out, bd);
for (j = 0; j < 8; ++j) {
dest[j * stride + i] = highbd_clip_pixel_add(
dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 5), bd);
}
}
}
}
#endif // CONFIG_HIGHBITDEPTH #endif // CONFIG_HIGHBITDEPTH

466
third_party/aom/aom_dsp/x86/masked_sad_intrin_ssse3.c vendored
View file

@ -1,5 +1,5 @@
/* /*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved * Copyright (c) 2017, Alliance for Open Media. All rights reserved
* *
* This source code is subject to the terms of the BSD 2 Clause License and * This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
@ -9,49 +9,70 @@
* PATENTS file, you can obtain it at www.aomedia.org/license/patent. * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/ */
#include <stdlib.h> #include <stdio.h>
#include <emmintrin.h>
#include <tmmintrin.h> #include <tmmintrin.h>
#include "aom_ports/mem.h"
#include "./aom_config.h" #include "./aom_config.h"
#include "./aom_dsp_rtcd.h"
#include "aom_dsp/blend.h"
#include "aom/aom_integer.h" #include "aom/aom_integer.h"
#include "aom_dsp/x86/synonyms.h"
static INLINE __m128i width8_load_2rows(const uint8_t *ptr, int stride) { // For width a multiple of 16
__m128i temp1 = _mm_loadl_epi64((const __m128i *)ptr); static INLINE unsigned int masked_sad_ssse3(const uint8_t *src_ptr,
__m128i temp2 = _mm_loadl_epi64((const __m128i *)(ptr + stride)); int src_stride,
return _mm_unpacklo_epi64(temp1, temp2); const uint8_t *a_ptr, int a_stride,
}
static INLINE __m128i width4_load_4rows(const uint8_t *ptr, int stride) {
__m128i temp1 = _mm_cvtsi32_si128(*(const uint32_t *)ptr);
__m128i temp2 = _mm_cvtsi32_si128(*(const uint32_t *)(ptr + stride));
__m128i temp3 = _mm_unpacklo_epi32(temp1, temp2);
temp1 = _mm_cvtsi32_si128(*(const uint32_t *)(ptr + stride * 2));
temp2 = _mm_cvtsi32_si128(*(const uint32_t *)(ptr + stride * 3));
temp1 = _mm_unpacklo_epi32(temp1, temp2);
return _mm_unpacklo_epi64(temp3, temp1);
}
static INLINE unsigned int masked_sad_ssse3(const uint8_t *a_ptr, int a_stride,
const uint8_t *b_ptr, int b_stride, const uint8_t *b_ptr, int b_stride,
const uint8_t *m_ptr, int m_stride, const uint8_t *m_ptr, int m_stride,
int width, int height); int width, int height);
static INLINE unsigned int masked_sad8xh_ssse3( static INLINE unsigned int masked_sad8xh_ssse3(
const uint8_t *a_ptr, int a_stride, const uint8_t *b_ptr, int b_stride, const uint8_t *src_ptr, int src_stride, const uint8_t *a_ptr, int a_stride,
const uint8_t *m_ptr, int m_stride, int height); const uint8_t *b_ptr, int b_stride, const uint8_t *m_ptr, int m_stride,
int height);
static INLINE unsigned int masked_sad4xh_ssse3( static INLINE unsigned int masked_sad4xh_ssse3(
const uint8_t *a_ptr, int a_stride, const uint8_t *b_ptr, int b_stride, const uint8_t *src_ptr, int src_stride, const uint8_t *a_ptr, int a_stride,
const uint8_t *m_ptr, int m_stride, int height); const uint8_t *b_ptr, int b_stride, const uint8_t *m_ptr, int m_stride,
int height);
#define MASKSADMXN_SSSE3(m, n) \ #define MASKSADMXN_SSSE3(m, n) \
unsigned int aom_masked_sad##m##x##n##_ssse3( \ unsigned int aom_masked_sad##m##x##n##_ssse3( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \ const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
const uint8_t *msk, int msk_stride) { \ const uint8_t *second_pred, const uint8_t *msk, int msk_stride, \
return masked_sad_ssse3(src, src_stride, ref, ref_stride, msk, msk_stride, \ int invert_mask) { \
m, n); \ if (!invert_mask) \
return masked_sad_ssse3(src, src_stride, ref, ref_stride, second_pred, \
m, msk, msk_stride, m, n); \
else \
return masked_sad_ssse3(src, src_stride, second_pred, m, ref, \
ref_stride, msk, msk_stride, m, n); \
}
#define MASKSAD8XN_SSSE3(n) \
unsigned int aom_masked_sad8x##n##_ssse3( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
const uint8_t *second_pred, const uint8_t *msk, int msk_stride, \
int invert_mask) { \
if (!invert_mask) \
return masked_sad8xh_ssse3(src, src_stride, ref, ref_stride, \
second_pred, 8, msk, msk_stride, n); \
else \
return masked_sad8xh_ssse3(src, src_stride, second_pred, 8, ref, \
ref_stride, msk, msk_stride, n); \
}
#define MASKSAD4XN_SSSE3(n) \
unsigned int aom_masked_sad4x##n##_ssse3( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
const uint8_t *second_pred, const uint8_t *msk, int msk_stride, \
int invert_mask) { \
if (!invert_mask) \
return masked_sad4xh_ssse3(src, src_stride, ref, ref_stride, \
second_pred, 4, msk, msk_stride, n); \
else \
return masked_sad4xh_ssse3(src, src_stride, second_pred, 4, ref, \
ref_stride, msk, msk_stride, n); \
} }
#if CONFIG_EXT_PARTITION #if CONFIG_EXT_PARTITION
@ -67,165 +88,181 @@ MASKSADMXN_SSSE3(32, 16)
MASKSADMXN_SSSE3(16, 32) MASKSADMXN_SSSE3(16, 32)
MASKSADMXN_SSSE3(16, 16) MASKSADMXN_SSSE3(16, 16)
MASKSADMXN_SSSE3(16, 8) MASKSADMXN_SSSE3(16, 8)
#define MASKSAD8XN_SSSE3(n) \
unsigned int aom_masked_sad8x##n##_ssse3( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
const uint8_t *msk, int msk_stride) { \
return masked_sad8xh_ssse3(src, src_stride, ref, ref_stride, msk, \
msk_stride, n); \
}
MASKSAD8XN_SSSE3(16) MASKSAD8XN_SSSE3(16)
MASKSAD8XN_SSSE3(8) MASKSAD8XN_SSSE3(8)
MASKSAD8XN_SSSE3(4) MASKSAD8XN_SSSE3(4)
#define MASKSAD4XN_SSSE3(n) \
unsigned int aom_masked_sad4x##n##_ssse3( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
const uint8_t *msk, int msk_stride) { \
return masked_sad4xh_ssse3(src, src_stride, ref, ref_stride, msk, \
msk_stride, n); \
}
MASKSAD4XN_SSSE3(8) MASKSAD4XN_SSSE3(8)
MASKSAD4XN_SSSE3(4) MASKSAD4XN_SSSE3(4)
// For width a multiple of 16 static INLINE unsigned int masked_sad_ssse3(const uint8_t *src_ptr,
// Assumes values in m are <=64 int src_stride,
static INLINE unsigned int masked_sad_ssse3(const uint8_t *a_ptr, int a_stride, const uint8_t *a_ptr, int a_stride,
const uint8_t *b_ptr, int b_stride, const uint8_t *b_ptr, int b_stride,
const uint8_t *m_ptr, int m_stride, const uint8_t *m_ptr, int m_stride,
int width, int height) { int width, int height) {
int y, x; int x, y;
__m128i a, b, m, temp1, temp2;
__m128i res = _mm_setzero_si128(); __m128i res = _mm_setzero_si128();
__m128i one = _mm_set1_epi16(1); const __m128i mask_max = _mm_set1_epi8((1 << AOM_BLEND_A64_ROUND_BITS));
// For each row
for (y = 0; y < height; y++) { for (y = 0; y < height; y++) {
// Covering the full width
for (x = 0; x < width; x += 16) { for (x = 0; x < width; x += 16) {
// Load a, b, m in xmm registers const __m128i src = _mm_loadu_si128((const __m128i *)&src_ptr[x]);
a = _mm_loadu_si128((const __m128i *)(a_ptr + x)); const __m128i a = _mm_loadu_si128((const __m128i *)&a_ptr[x]);
b = _mm_loadu_si128((const __m128i *)(b_ptr + x)); const __m128i b = _mm_loadu_si128((const __m128i *)&b_ptr[x]);
m = _mm_loadu_si128((const __m128i *)(m_ptr + x)); const __m128i m = _mm_loadu_si128((const __m128i *)&m_ptr[x]);
const __m128i m_inv = _mm_sub_epi8(mask_max, m);
// Calculate the difference between a & b // Calculate 16 predicted pixels.
temp1 = _mm_subs_epu8(a, b); // Note that the maximum value of any entry of 'pred_l' or 'pred_r'
temp2 = _mm_subs_epu8(b, a); // is 64 * 255, so we have plenty of space to add rounding constants.
temp1 = _mm_or_si128(temp1, temp2); const __m128i data_l = _mm_unpacklo_epi8(a, b);
const __m128i mask_l = _mm_unpacklo_epi8(m, m_inv);
__m128i pred_l = _mm_maddubs_epi16(data_l, mask_l);
pred_l = xx_roundn_epu16(pred_l, AOM_BLEND_A64_ROUND_BITS);
// Multiply by m and add together const __m128i data_r = _mm_unpackhi_epi8(a, b);
temp2 = _mm_maddubs_epi16(temp1, m); const __m128i mask_r = _mm_unpackhi_epi8(m, m_inv);
// Pad out row result to 32 bit integers & add to running total __m128i pred_r = _mm_maddubs_epi16(data_r, mask_r);
res = _mm_add_epi32(res, _mm_madd_epi16(temp2, one)); pred_r = xx_roundn_epu16(pred_r, AOM_BLEND_A64_ROUND_BITS);
const __m128i pred = _mm_packus_epi16(pred_l, pred_r);
res = _mm_add_epi32(res, _mm_sad_epu8(pred, src));
} }
// Move onto the next row
src_ptr += src_stride;
a_ptr += a_stride; a_ptr += a_stride;
b_ptr += b_stride; b_ptr += b_stride;
m_ptr += m_stride; m_ptr += m_stride;
} }
res = _mm_hadd_epi32(res, _mm_setzero_si128()); // At this point, we have two 32-bit partial SADs in lanes 0 and 2 of 'res'.
res = _mm_hadd_epi32(res, _mm_setzero_si128()); int32_t sad =
// sad = (sad + 31) >> 6; _mm_cvtsi128_si32(res) + _mm_cvtsi128_si32(_mm_srli_si128(res, 8));
return (_mm_cvtsi128_si32(res) + 31) >> 6; return (sad + 31) >> 6;
} }
static INLINE unsigned int masked_sad8xh_ssse3( static INLINE unsigned int masked_sad8xh_ssse3(
const uint8_t *a_ptr, int a_stride, const uint8_t *b_ptr, int b_stride, const uint8_t *src_ptr, int src_stride, const uint8_t *a_ptr, int a_stride,
const uint8_t *m_ptr, int m_stride, int height) { const uint8_t *b_ptr, int b_stride, const uint8_t *m_ptr, int m_stride,
int height) {
int y; int y;
__m128i a, b, m, temp1, temp2, row_res;
__m128i res = _mm_setzero_si128(); __m128i res = _mm_setzero_si128();
__m128i one = _mm_set1_epi16(1); const __m128i mask_max = _mm_set1_epi8((1 << AOM_BLEND_A64_ROUND_BITS));
// Add the masked SAD for 2 rows at a time
for (y = 0; y < height; y += 2) { for (y = 0; y < height; y += 2) {
// Load a, b, m in xmm registers const __m128i src = _mm_unpacklo_epi64(
a = width8_load_2rows(a_ptr, a_stride); _mm_loadl_epi64((const __m128i *)src_ptr),
b = width8_load_2rows(b_ptr, b_stride); _mm_loadl_epi64((const __m128i *)&src_ptr[src_stride]));
m = width8_load_2rows(m_ptr, m_stride); const __m128i a0 = _mm_loadl_epi64((const __m128i *)a_ptr);
const __m128i a1 = _mm_loadl_epi64((const __m128i *)&a_ptr[a_stride]);
const __m128i b0 = _mm_loadl_epi64((const __m128i *)b_ptr);
const __m128i b1 = _mm_loadl_epi64((const __m128i *)&b_ptr[b_stride]);
const __m128i m =
_mm_unpacklo_epi64(_mm_loadl_epi64((const __m128i *)m_ptr),
_mm_loadl_epi64((const __m128i *)&m_ptr[m_stride]));
const __m128i m_inv = _mm_sub_epi8(mask_max, m);
// Calculate the difference between a & b const __m128i data_l = _mm_unpacklo_epi8(a0, b0);
temp1 = _mm_subs_epu8(a, b); const __m128i mask_l = _mm_unpacklo_epi8(m, m_inv);
temp2 = _mm_subs_epu8(b, a); __m128i pred_l = _mm_maddubs_epi16(data_l, mask_l);
temp1 = _mm_or_si128(temp1, temp2); pred_l = xx_roundn_epu16(pred_l, AOM_BLEND_A64_ROUND_BITS);
// Multiply by m and add together const __m128i data_r = _mm_unpacklo_epi8(a1, b1);
row_res = _mm_maddubs_epi16(temp1, m); const __m128i mask_r = _mm_unpackhi_epi8(m, m_inv);
__m128i pred_r = _mm_maddubs_epi16(data_r, mask_r);
pred_r = xx_roundn_epu16(pred_r, AOM_BLEND_A64_ROUND_BITS);
// Pad out row result to 32 bit integers & add to running total const __m128i pred = _mm_packus_epi16(pred_l, pred_r);
res = _mm_add_epi32(res, _mm_madd_epi16(row_res, one)); res = _mm_add_epi32(res, _mm_sad_epu8(pred, src));
// Move onto the next rows src_ptr += src_stride * 2;
a_ptr += a_stride * 2; a_ptr += a_stride * 2;
b_ptr += b_stride * 2; b_ptr += b_stride * 2;
m_ptr += m_stride * 2; m_ptr += m_stride * 2;
} }
res = _mm_hadd_epi32(res, _mm_setzero_si128()); int32_t sad =
res = _mm_hadd_epi32(res, _mm_setzero_si128()); _mm_cvtsi128_si32(res) + _mm_cvtsi128_si32(_mm_srli_si128(res, 8));
// sad = (sad + 31) >> 6; return (sad + 31) >> 6;
return (_mm_cvtsi128_si32(res) + 31) >> 6;
} }
static INLINE unsigned int masked_sad4xh_ssse3( static INLINE unsigned int masked_sad4xh_ssse3(
const uint8_t *a_ptr, int a_stride, const uint8_t *b_ptr, int b_stride, const uint8_t *src_ptr, int src_stride, const uint8_t *a_ptr, int a_stride,
const uint8_t *m_ptr, int m_stride, int height) { const uint8_t *b_ptr, int b_stride, const uint8_t *m_ptr, int m_stride,
int height) {
int y; int y;
__m128i a, b, m, temp1, temp2, row_res;
__m128i res = _mm_setzero_si128(); __m128i res = _mm_setzero_si128();
__m128i one = _mm_set1_epi16(1); const __m128i mask_max = _mm_set1_epi8((1 << AOM_BLEND_A64_ROUND_BITS));
// Add the masked SAD for 4 rows at a time
for (y = 0; y < height; y += 4) {
// Load a, b, m in xmm registers
a = width4_load_4rows(a_ptr, a_stride);
b = width4_load_4rows(b_ptr, b_stride);
m = width4_load_4rows(m_ptr, m_stride);
// Calculate the difference between a & b for (y = 0; y < height; y += 2) {
temp1 = _mm_subs_epu8(a, b); // Load two rows at a time, this seems to be a bit faster
temp2 = _mm_subs_epu8(b, a); // than four rows at a time in this case.
temp1 = _mm_or_si128(temp1, temp2); const __m128i src = _mm_unpacklo_epi32(
_mm_cvtsi32_si128(*(uint32_t *)src_ptr),
_mm_cvtsi32_si128(*(uint32_t *)&src_ptr[src_stride]));
const __m128i a =
_mm_unpacklo_epi32(_mm_cvtsi32_si128(*(uint32_t *)a_ptr),
_mm_cvtsi32_si128(*(uint32_t *)&a_ptr[a_stride]));
const __m128i b =
_mm_unpacklo_epi32(_mm_cvtsi32_si128(*(uint32_t *)b_ptr),
_mm_cvtsi32_si128(*(uint32_t *)&b_ptr[b_stride]));
const __m128i m =
_mm_unpacklo_epi32(_mm_cvtsi32_si128(*(uint32_t *)m_ptr),
_mm_cvtsi32_si128(*(uint32_t *)&m_ptr[m_stride]));
const __m128i m_inv = _mm_sub_epi8(mask_max, m);
// Multiply by m and add together const __m128i data = _mm_unpacklo_epi8(a, b);
row_res = _mm_maddubs_epi16(temp1, m); const __m128i mask = _mm_unpacklo_epi8(m, m_inv);
__m128i pred_16bit = _mm_maddubs_epi16(data, mask);
pred_16bit = xx_roundn_epu16(pred_16bit, AOM_BLEND_A64_ROUND_BITS);
// Pad out row result to 32 bit integers & add to running total const __m128i pred = _mm_packus_epi16(pred_16bit, _mm_setzero_si128());
res = _mm_add_epi32(res, _mm_madd_epi16(row_res, one)); res = _mm_add_epi32(res, _mm_sad_epu8(pred, src));
// Move onto the next rows src_ptr += src_stride * 2;
a_ptr += a_stride * 4; a_ptr += a_stride * 2;
b_ptr += b_stride * 4; b_ptr += b_stride * 2;
m_ptr += m_stride * 4; m_ptr += m_stride * 2;
} }
// Pad out row result to 32 bit integers & add to running total // At this point, the SAD is stored in lane 0 of 'res'
res = _mm_hadd_epi32(res, _mm_setzero_si128()); int32_t sad = _mm_cvtsi128_si32(res);
res = _mm_hadd_epi32(res, _mm_setzero_si128()); return (sad + 31) >> 6;
// sad = (sad + 31) >> 6;
return (_mm_cvtsi128_si32(res) + 31) >> 6;
} }
#if CONFIG_HIGHBITDEPTH #if CONFIG_HIGHBITDEPTH
static INLINE __m128i highbd_width4_load_2rows(const uint16_t *ptr, // For width a multiple of 8
int stride) {
__m128i temp1 = _mm_loadl_epi64((const __m128i *)ptr);
__m128i temp2 = _mm_loadl_epi64((const __m128i *)(ptr + stride));
return _mm_unpacklo_epi64(temp1, temp2);
}
static INLINE unsigned int highbd_masked_sad_ssse3( static INLINE unsigned int highbd_masked_sad_ssse3(
const uint8_t *a8_ptr, int a_stride, const uint8_t *b8_ptr, int b_stride, const uint8_t *src8, int src_stride, const uint8_t *a8, int a_stride,
const uint8_t *m_ptr, int m_stride, int width, int height); const uint8_t *b8, int b_stride, const uint8_t *m_ptr, int m_stride,
int width, int height);
static INLINE unsigned int highbd_masked_sad4xh_ssse3( static INLINE unsigned int highbd_masked_sad4xh_ssse3(
const uint8_t *a8_ptr, int a_stride, const uint8_t *b8_ptr, int b_stride, const uint8_t *src8, int src_stride, const uint8_t *a8, int a_stride,
const uint8_t *m_ptr, int m_stride, int height); const uint8_t *b8, int b_stride, const uint8_t *m_ptr, int m_stride,
int height);
#define HIGHBD_MASKSADMXN_SSSE3(m, n) \ #define HIGHBD_MASKSADMXN_SSSE3(m, n) \
unsigned int aom_highbd_masked_sad##m##x##n##_ssse3( \ unsigned int aom_highbd_masked_sad##m##x##n##_ssse3( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \ const uint8_t *src8, int src_stride, const uint8_t *ref8, \
const uint8_t *msk, int msk_stride) { \ int ref_stride, const uint8_t *second_pred8, const uint8_t *msk, \
return highbd_masked_sad_ssse3(src, src_stride, ref, ref_stride, msk, \ int msk_stride, int invert_mask) { \
msk_stride, m, n); \ if (!invert_mask) \
return highbd_masked_sad_ssse3(src8, src_stride, ref8, ref_stride, \
second_pred8, m, msk, msk_stride, m, n); \
else \
return highbd_masked_sad_ssse3(src8, src_stride, second_pred8, m, ref8, \
ref_stride, msk, msk_stride, m, n); \
}
#define HIGHBD_MASKSAD4XN_SSSE3(n) \
unsigned int aom_highbd_masked_sad4x##n##_ssse3( \
const uint8_t *src8, int src_stride, const uint8_t *ref8, \
int ref_stride, const uint8_t *second_pred8, const uint8_t *msk, \
int msk_stride, int invert_mask) { \
if (!invert_mask) \
return highbd_masked_sad4xh_ssse3(src8, src_stride, ref8, ref_stride, \
second_pred8, 4, msk, msk_stride, n); \
else \
return highbd_masked_sad4xh_ssse3(src8, src_stride, second_pred8, 4, \
ref8, ref_stride, msk, msk_stride, n); \
} }
#if CONFIG_EXT_PARTITION #if CONFIG_EXT_PARTITION
@ -244,91 +281,124 @@ HIGHBD_MASKSADMXN_SSSE3(16, 8)
HIGHBD_MASKSADMXN_SSSE3(8, 16) HIGHBD_MASKSADMXN_SSSE3(8, 16)
HIGHBD_MASKSADMXN_SSSE3(8, 8) HIGHBD_MASKSADMXN_SSSE3(8, 8)
HIGHBD_MASKSADMXN_SSSE3(8, 4) HIGHBD_MASKSADMXN_SSSE3(8, 4)
#define HIGHBD_MASKSAD4XN_SSSE3(n) \
unsigned int aom_highbd_masked_sad4x##n##_ssse3( \
const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
const uint8_t *msk, int msk_stride) { \
return highbd_masked_sad4xh_ssse3(src, src_stride, ref, ref_stride, msk, \
msk_stride, n); \
}
HIGHBD_MASKSAD4XN_SSSE3(8) HIGHBD_MASKSAD4XN_SSSE3(8)
HIGHBD_MASKSAD4XN_SSSE3(4) HIGHBD_MASKSAD4XN_SSSE3(4)
// For width a multiple of 8
// Assumes values in m are <=64
static INLINE unsigned int highbd_masked_sad_ssse3( static INLINE unsigned int highbd_masked_sad_ssse3(
const uint8_t *a8_ptr, int a_stride, const uint8_t *b8_ptr, int b_stride, const uint8_t *src8, int src_stride, const uint8_t *a8, int a_stride,
const uint8_t *m_ptr, int m_stride, int width, int height) { const uint8_t *b8, int b_stride, const uint8_t *m_ptr, int m_stride,
int y, x; int width, int height) {
__m128i a, b, m, temp1, temp2; const uint16_t *src_ptr = CONVERT_TO_SHORTPTR(src8);
const uint16_t *a_ptr = CONVERT_TO_SHORTPTR(a8_ptr); const uint16_t *a_ptr = CONVERT_TO_SHORTPTR(a8);
const uint16_t *b_ptr = CONVERT_TO_SHORTPTR(b8_ptr); const uint16_t *b_ptr = CONVERT_TO_SHORTPTR(b8);
int x, y;
__m128i res = _mm_setzero_si128(); __m128i res = _mm_setzero_si128();
// For each row const __m128i mask_max = _mm_set1_epi16((1 << AOM_BLEND_A64_ROUND_BITS));
const __m128i round_const =
_mm_set1_epi32((1 << AOM_BLEND_A64_ROUND_BITS) >> 1);
const __m128i one = _mm_set1_epi16(1);
for (y = 0; y < height; y++) { for (y = 0; y < height; y++) {
// Covering the full width
for (x = 0; x < width; x += 8) { for (x = 0; x < width; x += 8) {
// Load a, b, m in xmm registers const __m128i src = _mm_loadu_si128((const __m128i *)&src_ptr[x]);
a = _mm_loadu_si128((const __m128i *)(a_ptr + x)); const __m128i a = _mm_loadu_si128((const __m128i *)&a_ptr[x]);
b = _mm_loadu_si128((const __m128i *)(b_ptr + x)); const __m128i b = _mm_loadu_si128((const __m128i *)&b_ptr[x]);
m = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i *)(m_ptr + x)), // Zero-extend mask to 16 bits
_mm_setzero_si128()); const __m128i m = _mm_unpacklo_epi8(
_mm_loadl_epi64((const __m128i *)&m_ptr[x]), _mm_setzero_si128());
const __m128i m_inv = _mm_sub_epi16(mask_max, m);
// Calculate the difference between a & b const __m128i data_l = _mm_unpacklo_epi16(a, b);
temp1 = _mm_subs_epu16(a, b); const __m128i mask_l = _mm_unpacklo_epi16(m, m_inv);
temp2 = _mm_subs_epu16(b, a); __m128i pred_l = _mm_madd_epi16(data_l, mask_l);
temp1 = _mm_or_si128(temp1, temp2); pred_l = _mm_srai_epi32(_mm_add_epi32(pred_l, round_const),
AOM_BLEND_A64_ROUND_BITS);
// Add result of multiplying by m and add pairs together to running total const __m128i data_r = _mm_unpackhi_epi16(a, b);
res = _mm_add_epi32(res, _mm_madd_epi16(temp1, m)); const __m128i mask_r = _mm_unpackhi_epi16(m, m_inv);
__m128i pred_r = _mm_madd_epi16(data_r, mask_r);
pred_r = _mm_srai_epi32(_mm_add_epi32(pred_r, round_const),
AOM_BLEND_A64_ROUND_BITS);
// Note: the maximum value in pred_l/r is (2^bd)-1 < 2^15,
// so it is safe to do signed saturation here.
const __m128i pred = _mm_packs_epi32(pred_l, pred_r);
// There is no 16-bit SAD instruction, so we have to synthesize
// an 8-element SAD. We do this by storing 4 32-bit partial SADs,
// and accumulating them at the end
const __m128i diff = _mm_abs_epi16(_mm_sub_epi16(pred, src));
res = _mm_add_epi32(res, _mm_madd_epi16(diff, one));
} }
// Move onto the next row
src_ptr += src_stride;
a_ptr += a_stride; a_ptr += a_stride;
b_ptr += b_stride; b_ptr += b_stride;
m_ptr += m_stride; m_ptr += m_stride;
} }
res = _mm_hadd_epi32(res, _mm_setzero_si128()); // At this point, we have four 32-bit partial SADs stored in 'res'.
res = _mm_hadd_epi32(res, _mm_setzero_si128()); res = _mm_hadd_epi32(res, res);
// sad = (sad + 31) >> 6; res = _mm_hadd_epi32(res, res);
return (_mm_cvtsi128_si32(res) + 31) >> 6; int sad = _mm_cvtsi128_si32(res);
return (sad + 31) >> 6;
} }
static INLINE unsigned int highbd_masked_sad4xh_ssse3( static INLINE unsigned int highbd_masked_sad4xh_ssse3(
const uint8_t *a8_ptr, int a_stride, const uint8_t *b8_ptr, int b_stride, const uint8_t *src8, int src_stride, const uint8_t *a8, int a_stride,
const uint8_t *m_ptr, int m_stride, int height) { const uint8_t *b8, int b_stride, const uint8_t *m_ptr, int m_stride,
int height) {
const uint16_t *src_ptr = CONVERT_TO_SHORTPTR(src8);
const uint16_t *a_ptr = CONVERT_TO_SHORTPTR(a8);
const uint16_t *b_ptr = CONVERT_TO_SHORTPTR(b8);
int y; int y;
__m128i a, b, m, temp1, temp2;
const uint16_t *a_ptr = CONVERT_TO_SHORTPTR(a8_ptr);
const uint16_t *b_ptr = CONVERT_TO_SHORTPTR(b8_ptr);
__m128i res = _mm_setzero_si128(); __m128i res = _mm_setzero_si128();
// Add the masked SAD for 2 rows at a time const __m128i mask_max = _mm_set1_epi16((1 << AOM_BLEND_A64_ROUND_BITS));
const __m128i round_const =
_mm_set1_epi32((1 << AOM_BLEND_A64_ROUND_BITS) >> 1);
const __m128i one = _mm_set1_epi16(1);
for (y = 0; y < height; y += 2) { for (y = 0; y < height; y += 2) {
// Load a, b, m in xmm registers const __m128i src = _mm_unpacklo_epi64(
a = highbd_width4_load_2rows(a_ptr, a_stride); _mm_loadl_epi64((const __m128i *)src_ptr),
b = highbd_width4_load_2rows(b_ptr, b_stride); _mm_loadl_epi64((const __m128i *)&src_ptr[src_stride]));
temp1 = _mm_loadl_epi64((const __m128i *)m_ptr); const __m128i a =
temp2 = _mm_loadl_epi64((const __m128i *)(m_ptr + m_stride)); _mm_unpacklo_epi64(_mm_loadl_epi64((const __m128i *)a_ptr),
m = _mm_unpacklo_epi8(_mm_unpacklo_epi32(temp1, temp2), _mm_loadl_epi64((const __m128i *)&a_ptr[a_stride]));
_mm_setzero_si128()); const __m128i b =
_mm_unpacklo_epi64(_mm_loadl_epi64((const __m128i *)b_ptr),
_mm_loadl_epi64((const __m128i *)&b_ptr[b_stride]));
// Zero-extend mask to 16 bits
const __m128i m = _mm_unpacklo_epi8(
_mm_unpacklo_epi32(
_mm_cvtsi32_si128(*(const uint32_t *)m_ptr),
_mm_cvtsi32_si128(*(const uint32_t *)&m_ptr[m_stride])),
_mm_setzero_si128());
const __m128i m_inv = _mm_sub_epi16(mask_max, m);
// Calculate the difference between a & b const __m128i data_l = _mm_unpacklo_epi16(a, b);
temp1 = _mm_subs_epu16(a, b); const __m128i mask_l = _mm_unpacklo_epi16(m, m_inv);
temp2 = _mm_subs_epu16(b, a); __m128i pred_l = _mm_madd_epi16(data_l, mask_l);
temp1 = _mm_or_si128(temp1, temp2); pred_l = _mm_srai_epi32(_mm_add_epi32(pred_l, round_const),
AOM_BLEND_A64_ROUND_BITS);
// Multiply by m and add together const __m128i data_r = _mm_unpackhi_epi16(a, b);
res = _mm_add_epi32(res, _mm_madd_epi16(temp1, m)); const __m128i mask_r = _mm_unpackhi_epi16(m, m_inv);
__m128i pred_r = _mm_madd_epi16(data_r, mask_r);
pred_r = _mm_srai_epi32(_mm_add_epi32(pred_r, round_const),
AOM_BLEND_A64_ROUND_BITS);
// Move onto the next rows const __m128i pred = _mm_packs_epi32(pred_l, pred_r);
const __m128i diff = _mm_abs_epi16(_mm_sub_epi16(pred, src));
res = _mm_add_epi32(res, _mm_madd_epi16(diff, one));
src_ptr += src_stride * 2;
a_ptr += a_stride * 2; a_ptr += a_stride * 2;
b_ptr += b_stride * 2; b_ptr += b_stride * 2;
m_ptr += m_stride * 2; m_ptr += m_stride * 2;
} }
res = _mm_hadd_epi32(res, _mm_setzero_si128()); res = _mm_hadd_epi32(res, res);
res = _mm_hadd_epi32(res, _mm_setzero_si128()); res = _mm_hadd_epi32(res, res);
// sad = (sad + 31) >> 6; int sad = _mm_cvtsi128_si32(res);
return (_mm_cvtsi128_si32(res) + 31) >> 6; return (sad + 31) >> 6;
} }
#endif // CONFIG_HIGHBITDEPTH
#endif

2851
third_party/aom/aom_dsp/x86/masked_variance_intrin_ssse3.c vendored
View file

File diff suppressed because it is too large Load diff

45
third_party/aom/aom_dsp/x86/obmc_intrinsic_ssse3.h vendored Normal file
View file

@ -0,0 +1,45 @@
/*
* Copyright (c) 2017, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#ifndef AOM_DSP_X86_OBMC_INTRINSIC_SSSE3_H_
#define AOM_DSP_X86_OBMC_INTRINSIC_SSSE3_H_
#include <immintrin.h>
#include "./aom_config.h"
static INLINE int32_t xx_hsum_epi32_si32(__m128i v_d) {
v_d = _mm_hadd_epi32(v_d, v_d);
v_d = _mm_hadd_epi32(v_d, v_d);
return _mm_cvtsi128_si32(v_d);
}
static INLINE int64_t xx_hsum_epi64_si64(__m128i v_q) {
v_q = _mm_add_epi64(v_q, _mm_srli_si128(v_q, 8));
#if ARCH_X86_64
return _mm_cvtsi128_si64(v_q);
#else
{
int64_t tmp;
_mm_storel_epi64((__m128i *)&tmp, v_q);
return tmp;
}
#endif
}
static INLINE int64_t xx_hsum_epi32_si64(__m128i v_d) {
const __m128i v_sign_d = _mm_cmplt_epi32(v_d, _mm_setzero_si128());
const __m128i v_0_q = _mm_unpacklo_epi32(v_d, v_sign_d);
const __m128i v_1_q = _mm_unpackhi_epi32(v_d, v_sign_d);
return xx_hsum_epi64_si64(_mm_add_epi64(v_0_q, v_1_q));
}
#endif // AOM_DSP_X86_OBMC_INTRINSIC_SSSE3_H_

1
third_party/aom/aom_dsp/x86/obmc_sad_sse4.c vendored
View file

@ -17,6 +17,7 @@
#include "aom/aom_integer.h" #include "aom/aom_integer.h"
#include "aom_dsp/aom_dsp_common.h" #include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/x86/obmc_intrinsic_ssse3.h"
#include "aom_dsp/x86/synonyms.h" #include "aom_dsp/x86/synonyms.h"
//////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////

3
third_party/aom/aom_dsp/x86/obmc_variance_sse4.c vendored
View file

@ -17,8 +17,9 @@
#include "aom/aom_integer.h" #include "aom/aom_integer.h"
#include "aom_dsp/aom_dsp_common.h" #include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/x86/synonyms.h"
#include "aom_dsp/aom_filter.h" #include "aom_dsp/aom_filter.h"
#include "aom_dsp/x86/obmc_intrinsic_ssse3.h"
#include "aom_dsp/x86/synonyms.h"
//////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////
// 8 bit // 8 bit

28
third_party/aom/aom_dsp/x86/synonyms.h vendored
View file

@ -89,32 +89,4 @@ static INLINE __m128i xx_roundn_epi32(__m128i v_val_d, int bits) {
return _mm_srai_epi32(v_tmp_d, bits); return _mm_srai_epi32(v_tmp_d, bits);
} }
#ifdef __SSSE3__
static INLINE int32_t xx_hsum_epi32_si32(__m128i v_d) {
v_d = _mm_hadd_epi32(v_d, v_d);
v_d = _mm_hadd_epi32(v_d, v_d);
return _mm_cvtsi128_si32(v_d);
}
static INLINE int64_t xx_hsum_epi64_si64(__m128i v_q) {
v_q = _mm_add_epi64(v_q, _mm_srli_si128(v_q, 8));
#if ARCH_X86_64
return _mm_cvtsi128_si64(v_q);
#else
{
int64_t tmp;
_mm_storel_epi64((__m128i *)&tmp, v_q);
return tmp;
}
#endif
}
static INLINE int64_t xx_hsum_epi32_si64(__m128i v_d) {
const __m128i v_sign_d = _mm_cmplt_epi32(v_d, _mm_setzero_si128());
const __m128i v_0_q = _mm_unpacklo_epi32(v_d, v_sign_d);
const __m128i v_1_q = _mm_unpackhi_epi32(v_d, v_sign_d);
return xx_hsum_epi64_si64(_mm_add_epi64(v_0_q, v_1_q));
}
#endif // __SSSE3__
#endif // AOM_DSP_X86_SYNONYMS_H_ #endif // AOM_DSP_X86_SYNONYMS_H_

42
third_party/aom/aom_dsp/x86/txfm_common_avx2.h vendored
View file

@ -34,7 +34,8 @@ static INLINE void mm256_reverse_epi16(__m256i *u) {
*u = _mm256_permute2x128_si256(v, v, 1); *u = _mm256_permute2x128_si256(v, v, 1);
} }
static INLINE void mm256_transpose_16x16(__m256i *in) { // Note: in and out could have the same value
static INLINE void mm256_transpose_16x16(const __m256i *in, __m256i *out) {
__m256i tr0_0 = _mm256_unpacklo_epi16(in[0], in[1]); __m256i tr0_0 = _mm256_unpacklo_epi16(in[0], in[1]);
__m256i tr0_1 = _mm256_unpackhi_epi16(in[0], in[1]); __m256i tr0_1 = _mm256_unpackhi_epi16(in[0], in[1]);
__m256i tr0_2 = _mm256_unpacklo_epi16(in[2], in[3]); __m256i tr0_2 = _mm256_unpacklo_epi16(in[2], in[3]);
@ -143,29 +144,30 @@ static INLINE void mm256_transpose_16x16(__m256i *in) {
// 86 96 a6 b6 c6 d6 e6 f6 8e ae 9e be ce de ee fe // 86 96 a6 b6 c6 d6 e6 f6 8e ae 9e be ce de ee fe
// 87 97 a7 b7 c7 d7 e7 f7 8f 9f af bf cf df ef ff // 87 97 a7 b7 c7 d7 e7 f7 8f 9f af bf cf df ef ff
in[0] = _mm256_permute2x128_si256(tr0_0, tr0_8, 0x20); // 0010 0000 out[0] = _mm256_permute2x128_si256(tr0_0, tr0_8, 0x20); // 0010 0000
in[8] = _mm256_permute2x128_si256(tr0_0, tr0_8, 0x31); // 0011 0001 out[8] = _mm256_permute2x128_si256(tr0_0, tr0_8, 0x31); // 0011 0001
in[1] = _mm256_permute2x128_si256(tr0_1, tr0_9, 0x20); out[1] = _mm256_permute2x128_si256(tr0_1, tr0_9, 0x20);
in[9] = _mm256_permute2x128_si256(tr0_1, tr0_9, 0x31); out[9] = _mm256_permute2x128_si256(tr0_1, tr0_9, 0x31);
in[2] = _mm256_permute2x128_si256(tr0_2, tr0_a, 0x20); out[2] = _mm256_permute2x128_si256(tr0_2, tr0_a, 0x20);
in[10] = _mm256_permute2x128_si256(tr0_2, tr0_a, 0x31); out[10] = _mm256_permute2x128_si256(tr0_2, tr0_a, 0x31);
in[3] = _mm256_permute2x128_si256(tr0_3, tr0_b, 0x20); out[3] = _mm256_permute2x128_si256(tr0_3, tr0_b, 0x20);
in[11] = _mm256_permute2x128_si256(tr0_3, tr0_b, 0x31); out[11] = _mm256_permute2x128_si256(tr0_3, tr0_b, 0x31);
in[4] = _mm256_permute2x128_si256(tr0_4, tr0_c, 0x20); out[4] = _mm256_permute2x128_si256(tr0_4, tr0_c, 0x20);
in[12] = _mm256_permute2x128_si256(tr0_4, tr0_c, 0x31); out[12] = _mm256_permute2x128_si256(tr0_4, tr0_c, 0x31);
in[5] = _mm256_permute2x128_si256(tr0_5, tr0_d, 0x20); out[5] = _mm256_permute2x128_si256(tr0_5, tr0_d, 0x20);
in[13] = _mm256_permute2x128_si256(tr0_5, tr0_d, 0x31); out[13] = _mm256_permute2x128_si256(tr0_5, tr0_d, 0x31);
in[6] = _mm256_permute2x128_si256(tr0_6, tr0_e, 0x20); out[6] = _mm256_permute2x128_si256(tr0_6, tr0_e, 0x20);
in[14] = _mm256_permute2x128_si256(tr0_6, tr0_e, 0x31); out[14] = _mm256_permute2x128_si256(tr0_6, tr0_e, 0x31);
in[7] = _mm256_permute2x128_si256(tr0_7, tr0_f, 0x20); out[7] = _mm256_permute2x128_si256(tr0_7, tr0_f, 0x20);
in[15] = _mm256_permute2x128_si256(tr0_7, tr0_f, 0x31); out[15] = _mm256_permute2x128_si256(tr0_7, tr0_f, 0x31);
} }
static INLINE __m256i butter_fly(__m256i a0, __m256i a1, const __m256i cospi) { static INLINE __m256i butter_fly(const __m256i *a0, const __m256i *a1,
const __m256i *cospi) {
const __m256i dct_rounding = _mm256_set1_epi32(DCT_CONST_ROUNDING); const __m256i dct_rounding = _mm256_set1_epi32(DCT_CONST_ROUNDING);
__m256i y0 = _mm256_madd_epi16(a0, cospi); __m256i y0 = _mm256_madd_epi16(*a0, *cospi);
__m256i y1 = _mm256_madd_epi16(a1, cospi); __m256i y1 = _mm256_madd_epi16(*a1, *cospi);
y0 = _mm256_add_epi32(y0, dct_rounding); y0 = _mm256_add_epi32(y0, dct_rounding);
y1 = _mm256_add_epi32(y1, dct_rounding); y1 = _mm256_add_epi32(y1, dct_rounding);

5
third_party/aom/aom_mem/aom_mem.cmake vendored
View file

@ -8,6 +8,9 @@
## Media Patent License 1.0 was not distributed with this source code in the ## Media Patent License 1.0 was not distributed with this source code in the
## PATENTS file, you can obtain it at www.aomedia.org/license/patent. ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
## ##
if (NOT AOM_AOM_MEM_AOM_MEM_CMAKE_)
set(AOM_AOM_MEM_AOM_MEM_CMAKE_ 1)
set(AOM_MEM_SOURCES set(AOM_MEM_SOURCES
"${AOM_ROOT}/aom_mem/aom_mem.c" "${AOM_ROOT}/aom_mem/aom_mem.c"
"${AOM_ROOT}/aom_mem/aom_mem.h" "${AOM_ROOT}/aom_mem/aom_mem.h"
@ -20,3 +23,5 @@ function (setup_aom_mem_targets)
set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_mem PARENT_SCOPE) set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_mem PARENT_SCOPE)
target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_mem>) target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_mem>)
endfunction () endfunction ()
endif () # AOM_AOM_MEM_AOM_MEM_CMAKE_

19
third_party/aom/aom_mem/aom_mem.h vendored
View file

@ -39,6 +39,25 @@ void *aom_memset16(void *dest, int val, size_t length);
#include AOM_MEM_PLTFRM #include AOM_MEM_PLTFRM
#endif #endif
#if CONFIG_DEBUG
#define AOM_CHECK_MEM_ERROR(error_info, lval, expr) \
do { \
lval = (expr); \
if (!lval) \
aom_internal_error(error_info, AOM_CODEC_MEM_ERROR, \
"Failed to allocate " #lval " at %s:%d", __FILE__, \
__LINE__); \
} while (0)
#else
#define AOM_CHECK_MEM_ERROR(error_info, lval, expr) \
do { \
lval = (expr); \
if (!lval) \
aom_internal_error(error_info, AOM_CODEC_MEM_ERROR, \
"Failed to allocate " #lval); \
} while (0)
#endif
#if defined(__cplusplus) #if defined(__cplusplus)
} }
#endif #endif

5
third_party/aom/aom_ports/aom_ports.cmake vendored
View file

@ -8,6 +8,9 @@
## Media Patent License 1.0 was not distributed with this source code in the ## Media Patent License 1.0 was not distributed with this source code in the
## PATENTS file, you can obtain it at www.aomedia.org/license/patent. ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
## ##
if (NOT AOM_AOM_PORTS_AOM_PORTS_CMAKE_)
set(AOM_AOM_PORTS_AOM_PORTS_CMAKE_ 1)
set(AOM_PORTS_INCLUDES set(AOM_PORTS_INCLUDES
"${AOM_ROOT}/aom_ports/aom_once.h" "${AOM_ROOT}/aom_ports/aom_once.h"
"${AOM_ROOT}/aom_ports/aom_timer.h" "${AOM_ROOT}/aom_ports/aom_timer.h"
@ -64,3 +67,5 @@ function (setup_aom_ports_targets)
endif () endif ()
endif () endif ()
endfunction () endfunction ()
endif () # AOM_AOM_PORTS_AOM_PORTS_CMAKE_

26
third_party/aom/aom_ports/arm_cpudetect.c vendored
View file

@ -50,9 +50,6 @@ int arm_cpu_caps(void) {
return flags; return flags;
} }
mask = arm_cpu_env_mask(); mask = arm_cpu_env_mask();
#if HAVE_MEDIA
flags |= HAS_MEDIA;
#endif /* HAVE_MEDIA */
#if HAVE_NEON || HAVE_NEON_ASM #if HAVE_NEON || HAVE_NEON_ASM
flags |= HAS_NEON; flags |= HAS_NEON;
#endif /* HAVE_NEON || HAVE_NEON_ASM */ #endif /* HAVE_NEON || HAVE_NEON_ASM */
@ -76,17 +73,6 @@ int arm_cpu_caps(void) {
* instructions via their assembled hex code. * instructions via their assembled hex code.
* All of these instructions should be essentially nops. * All of these instructions should be essentially nops.
*/ */
#if HAVE_MEDIA
if (mask & HAS_MEDIA) {
__try {
/*SHADD8 r3,r3,r3*/
__emit(0xE6333F93);
flags |= HAS_MEDIA;
} __except (GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION) {
/*Ignore exception.*/
}
}
#endif /* HAVE_MEDIA */
#if HAVE_NEON || HAVE_NEON_ASM #if HAVE_NEON || HAVE_NEON_ASM
if (mask & HAS_NEON) { if (mask & HAS_NEON) {
__try { __try {
@ -114,9 +100,6 @@ int arm_cpu_caps(void) {
mask = arm_cpu_env_mask(); mask = arm_cpu_env_mask();
features = android_getCpuFeatures(); features = android_getCpuFeatures();
#if HAVE_MEDIA
flags |= HAS_MEDIA;
#endif /* HAVE_MEDIA */
#if HAVE_NEON || HAVE_NEON_ASM #if HAVE_NEON || HAVE_NEON_ASM
if (features & ANDROID_CPU_ARM_FEATURE_NEON) flags |= HAS_NEON; if (features & ANDROID_CPU_ARM_FEATURE_NEON) flags |= HAS_NEON;
#endif /* HAVE_NEON || HAVE_NEON_ASM */ #endif /* HAVE_NEON || HAVE_NEON_ASM */
@ -155,15 +138,6 @@ int arm_cpu_caps(void) {
} }
} }
#endif /* HAVE_NEON || HAVE_NEON_ASM */ #endif /* HAVE_NEON || HAVE_NEON_ASM */
#if HAVE_MEDIA
if (memcmp(buf, "CPU architecture:", 17) == 0) {
int version;
version = atoi(buf + 17);
if (version >= 6) {
flags |= HAS_MEDIA;
}
}
#endif /* HAVE_MEDIA */
} }
fclose(fin); fclose(fin);
} }

5
third_party/aom/aom_scale/aom_scale.cmake vendored
View file

@ -8,6 +8,9 @@
## Media Patent License 1.0 was not distributed with this source code in the ## Media Patent License 1.0 was not distributed with this source code in the
## PATENTS file, you can obtain it at www.aomedia.org/license/patent. ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
## ##
if (NOT AOM_AOM_SCALE_AOM_SCALE_CMAKE_)
set(AOM_AOM_SCALE_AOM_SCALE_CMAKE_ 1)
set(AOM_SCALE_SOURCES set(AOM_SCALE_SOURCES
"${AOM_ROOT}/aom_scale/aom_scale.h" "${AOM_ROOT}/aom_scale/aom_scale.h"
"${AOM_ROOT}/aom_scale/generic/aom_scale.c" "${AOM_ROOT}/aom_scale/generic/aom_scale.c"
@ -32,3 +35,5 @@ function (setup_aom_scale_targets)
set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_scale PARENT_SCOPE) set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_scale PARENT_SCOPE)
endfunction () endfunction ()
endif () # AOM_AOM_SCALE_AOM_SCALE_CMAKE_

5
third_party/aom/aom_util/aom_util.cmake vendored
View file

@ -8,6 +8,9 @@
## Media Patent License 1.0 was not distributed with this source code in the ## Media Patent License 1.0 was not distributed with this source code in the
## PATENTS file, you can obtain it at www.aomedia.org/license/patent. ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
## ##
if (NOT AOM_AOM_UTIL_AOM_UTIL_CMAKE_)
set(AOM_AOM_UTIL_AOM_UTIL_CMAKE_ 1)
set(AOM_UTIL_SOURCES set(AOM_UTIL_SOURCES
"${AOM_ROOT}/aom_util/aom_thread.c" "${AOM_ROOT}/aom_util/aom_thread.c"
"${AOM_ROOT}/aom_util/aom_thread.h" "${AOM_ROOT}/aom_util/aom_thread.h"
@ -27,3 +30,5 @@ function (setup_aom_util_targets)
set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_util PARENT_SCOPE) set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} aom_util PARENT_SCOPE)
target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_util>) target_sources(aom PUBLIC $<TARGET_OBJECTS:aom_util>)
endfunction () endfunction ()
endif () # AOM_AOM_UTIL_AOM_UTIL_CMAKE_

26
third_party/aom/aom_util/debug_util.c vendored
View file

@ -16,12 +16,8 @@
#define QUEUE_MAX_SIZE 2000000 #define QUEUE_MAX_SIZE 2000000
static int result_queue[QUEUE_MAX_SIZE]; static int result_queue[QUEUE_MAX_SIZE];
#if CONFIG_DAALA_EC
static int nsymbs_queue[QUEUE_MAX_SIZE]; static int nsymbs_queue[QUEUE_MAX_SIZE];
static aom_cdf_prob cdf_queue[QUEUE_MAX_SIZE][16]; static aom_cdf_prob cdf_queue[QUEUE_MAX_SIZE][16];
#else
static int prob_queue[QUEUE_MAX_SIZE];
#endif
static int queue_r = 0; static int queue_r = 0;
static int queue_w = 0; static int queue_w = 0;
@ -53,42 +49,24 @@ int bitstream_queue_get_write(void) { return queue_w; }
int bitstream_queue_get_read(void) { return queue_r; } int bitstream_queue_get_read(void) { return queue_r; }
void bitstream_queue_pop(int *result, void bitstream_queue_pop(int *result, aom_cdf_prob *cdf, int *nsymbs) {
#if CONFIG_DAALA_EC
aom_cdf_prob *cdf, int *nsymbs) {
#else
int *prob) {
#endif // CONFIG_DAALA_EC
if (!skip_r) { if (!skip_r) {
if (queue_w == queue_r) { if (queue_w == queue_r) {
printf("buffer underflow queue_w %d queue_r %d\n", queue_w, queue_r); printf("buffer underflow queue_w %d queue_r %d\n", queue_w, queue_r);
assert(0); assert(0);
} }
*result = result_queue[queue_r]; *result = result_queue[queue_r];
#if CONFIG_DAALA_EC
*nsymbs = nsymbs_queue[queue_r]; *nsymbs = nsymbs_queue[queue_r];
memcpy(cdf, cdf_queue[queue_r], *nsymbs * sizeof(*cdf)); memcpy(cdf, cdf_queue[queue_r], *nsymbs * sizeof(*cdf));
#else
*prob = prob_queue[queue_r];
#endif // CONFIG_DAALA_EC
queue_r = (queue_r + 1) % QUEUE_MAX_SIZE; queue_r = (queue_r + 1) % QUEUE_MAX_SIZE;
} }
} }
void bitstream_queue_push(int result, void bitstream_queue_push(int result, const aom_cdf_prob *cdf, int nsymbs) {
#if CONFIG_DAALA_EC
const aom_cdf_prob *cdf, int nsymbs) {
#else
int prob) {
#endif // CONFIG_DAALA_EC
if (!skip_w) { if (!skip_w) {
result_queue[queue_w] = result; result_queue[queue_w] = result;
#if CONFIG_DAALA_EC
nsymbs_queue[queue_w] = nsymbs; nsymbs_queue[queue_w] = nsymbs;
memcpy(cdf_queue[queue_w], cdf, nsymbs * sizeof(*cdf)); memcpy(cdf_queue[queue_w], cdf, nsymbs * sizeof(*cdf));
#else
prob_queue[queue_w] = prob;
#endif // CONFIG_DAALA_EC
queue_w = (queue_w + 1) % QUEUE_MAX_SIZE; queue_w = (queue_w + 1) % QUEUE_MAX_SIZE;
if (queue_w == queue_r) { if (queue_w == queue_r) {
printf("buffer overflow queue_w %d queue_r %d\n", queue_w, queue_r); printf("buffer overflow queue_w %d queue_r %d\n", queue_w, queue_r);

7
third_party/aom/aom_util/debug_util.h vendored
View file

@ -13,9 +13,7 @@
#define AOM_UTIL_DEBUG_UTIL_H_ #define AOM_UTIL_DEBUG_UTIL_H_
#include "./aom_config.h" #include "./aom_config.h"
#if CONFIG_DAALA_EC
#include "aom_dsp/prob.h" #include "aom_dsp/prob.h"
#endif
#ifdef __cplusplus #ifdef __cplusplus
extern "C" { extern "C" {
@ -33,13 +31,8 @@ int bitstream_queue_get_write(void);
int bitstream_queue_get_read(void); int bitstream_queue_get_read(void);
void bitstream_queue_record_write(void); void bitstream_queue_record_write(void);
void bitstream_queue_reset_write(void); void bitstream_queue_reset_write(void);
#if CONFIG_DAALA_EC
void bitstream_queue_pop(int *result, aom_cdf_prob *cdf, int *nsymbs); void bitstream_queue_pop(int *result, aom_cdf_prob *cdf, int *nsymbs);
void bitstream_queue_push(int result, const aom_cdf_prob *cdf, int nsymbs); void bitstream_queue_push(int result, const aom_cdf_prob *cdf, int nsymbs);
#else
void bitstream_queue_pop(int *result, int *prob);
void bitstream_queue_push(int result, int prob);
#endif
void bitstream_queue_set_skip_write(int skip); void bitstream_queue_set_skip_write(int skip);
void bitstream_queue_set_skip_read(int skip); void bitstream_queue_set_skip_read(int skip);
void bitstream_queue_set_frame_write(int frame_idx); void bitstream_queue_set_frame_write(int frame_idx);

40
third_party/aom/aomdec.c vendored
View file

@ -312,8 +312,6 @@ static int file_is_raw(struct AvxInputContext *input) {
int is_raw = 0; int is_raw = 0;
aom_codec_stream_info_t si; aom_codec_stream_info_t si;
si.sz = sizeof(si);
if (fread(buf, 1, 32, input->file) == 32) { if (fread(buf, 1, 32, input->file) == 32) {
int i; int i;
@ -892,28 +890,32 @@ static int main_loop(int argc, const char **argv_) {
output_bit_depth = img->bit_depth; output_bit_depth = img->bit_depth;
} }
// Shift up or down if necessary // Shift up or down if necessary
if (output_bit_depth != 0 && output_bit_depth != img->bit_depth) { if (output_bit_depth != 0) {
const aom_img_fmt_t shifted_fmt = const aom_img_fmt_t shifted_fmt =
output_bit_depth == 8 output_bit_depth == 8
? img->fmt ^ (img->fmt & AOM_IMG_FMT_HIGHBITDEPTH) ? img->fmt ^ (img->fmt & AOM_IMG_FMT_HIGHBITDEPTH)
: img->fmt | AOM_IMG_FMT_HIGHBITDEPTH; : img->fmt | AOM_IMG_FMT_HIGHBITDEPTH;
if (img_shifted &&
img_shifted_realloc_required(img, img_shifted, shifted_fmt)) { if (shifted_fmt != img->fmt || output_bit_depth != img->bit_depth) {
aom_img_free(img_shifted); if (img_shifted &&
img_shifted = NULL; img_shifted_realloc_required(img, img_shifted, shifted_fmt)) {
aom_img_free(img_shifted);
img_shifted = NULL;
}
if (!img_shifted) {
img_shifted =
aom_img_alloc(NULL, shifted_fmt, img->d_w, img->d_h, 16);
img_shifted->bit_depth = output_bit_depth;
}
if (output_bit_depth > img->bit_depth) {
aom_img_upshift(img_shifted, img,
output_bit_depth - img->bit_depth);
} else {
aom_img_downshift(img_shifted, img,
img->bit_depth - output_bit_depth);
}
img = img_shifted;
} }
if (!img_shifted) {
img_shifted =
aom_img_alloc(NULL, shifted_fmt, img->d_w, img->d_h, 16);
img_shifted->bit_depth = output_bit_depth;
}
if (output_bit_depth > img->bit_depth) {
aom_img_upshift(img_shifted, img, output_bit_depth - img->bit_depth);
} else {
aom_img_downshift(img_shifted, img,
img->bit_depth - output_bit_depth);
}
img = img_shifted;
} }
#endif #endif

21
third_party/aom/aomenc.c vendored
View file

@ -25,7 +25,7 @@
#endif #endif
#include "aom/aom_encoder.h" #include "aom/aom_encoder.h"
#if CONFIG_DECODERS #if CONFIG_AV1_DECODER
#include "aom/aom_decoder.h" #include "aom/aom_decoder.h"
#endif #endif
@ -45,6 +45,7 @@
#include "./rate_hist.h" #include "./rate_hist.h"
#include "./warnings.h" #include "./warnings.h"
#include "aom/aom_integer.h" #include "aom/aom_integer.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_ports/aom_timer.h" #include "aom_ports/aom_timer.h"
#include "aom_ports/mem_ops.h" #include "aom_ports/mem_ops.h"
#if CONFIG_WEBM_IO #if CONFIG_WEBM_IO
@ -627,7 +628,6 @@ void usage_exit(void) {
exit(EXIT_FAILURE); exit(EXIT_FAILURE);
} }
#define NELEMENTS(x) (sizeof(x) / sizeof(x[0]))
#if CONFIG_AV1_ENCODER #if CONFIG_AV1_ENCODER
#define ARG_CTRL_CNT_MAX NELEMENTS(av1_arg_ctrl_map) #define ARG_CTRL_CNT_MAX NELEMENTS(av1_arg_ctrl_map)
#endif #endif
@ -1061,15 +1061,8 @@ static int parse_stream_params(struct AvxEncoderConfig *global,
} }
} }
#if CONFIG_HIGHBITDEPTH #if CONFIG_HIGHBITDEPTH
#if CONFIG_LOWBITDEPTH config->use_16bit_internal =
if (strcmp(global->codec->name, "av1") == 0 || test_16bit_internal || (config->cfg.g_profile > 1) || !CONFIG_LOWBITDEPTH;
strcmp(global->codec->name, "av1") == 0) {
config->use_16bit_internal =
test_16bit_internal | (config->cfg.g_profile > 1);
}
#else
config->use_16bit_internal = 1;
#endif
#endif #endif
return eos_mark_found; return eos_mark_found;
} }
@ -1344,13 +1337,13 @@ static void initialize_encoder(struct stream_state *stream,
ctx_exit_on_error(&stream->encoder, "Failed to control codec"); ctx_exit_on_error(&stream->encoder, "Failed to control codec");
} }
#if CONFIG_DECODERS #if CONFIG_AV1_DECODER
if (global->test_decode != TEST_DECODE_OFF) { if (global->test_decode != TEST_DECODE_OFF) {
const AvxInterface *decoder = get_aom_decoder_by_name(global->codec->name); const AvxInterface *decoder = get_aom_decoder_by_name(global->codec->name);
aom_codec_dec_cfg_t cfg = { 0, 0, 0 }; aom_codec_dec_cfg_t cfg = { 0, 0, 0 };
aom_codec_dec_init(&stream->decoder, decoder->codec_interface(), &cfg, 0); aom_codec_dec_init(&stream->decoder, decoder->codec_interface(), &cfg, 0);
#if CONFIG_AV1_DECODER && CONFIG_EXT_TILE #if CONFIG_EXT_TILE
if (strcmp(global->codec->name, "av1") == 0) { if (strcmp(global->codec->name, "av1") == 0) {
aom_codec_control(&stream->decoder, AV1_SET_DECODE_TILE_ROW, -1); aom_codec_control(&stream->decoder, AV1_SET_DECODE_TILE_ROW, -1);
ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_row"); ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_row");
@ -1510,7 +1503,7 @@ static void get_cx_data(struct stream_state *stream,
stream->nbytes += pkt->data.raw.sz; stream->nbytes += pkt->data.raw.sz;
*got_data = 1; *got_data = 1;
#if CONFIG_DECODERS #if CONFIG_AV1_DECODER
if (global->test_decode != TEST_DECODE_OFF && !stream->mismatch_seen) { if (global->test_decode != TEST_DECODE_OFF && !stream->mismatch_seen) {
aom_codec_decode(&stream->decoder, pkt->data.frame.buf, aom_codec_decode(&stream->decoder, pkt->data.frame.buf,
(unsigned int)pkt->data.frame.sz, NULL, 0); (unsigned int)pkt->data.frame.sz, NULL, 0);

65
third_party/aom/av1/av1.cmake vendored
View file

@ -8,6 +8,9 @@
## Media Patent License 1.0 was not distributed with this source code in the ## Media Patent License 1.0 was not distributed with this source code in the
## PATENTS file, you can obtain it at www.aomedia.org/license/patent. ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
## ##
if (NOT AOM_AV1_AV1_CMAKE_)
set(AOM_AV1_AV1_CMAKE_ 1)
set(AOM_AV1_COMMON_SOURCES set(AOM_AV1_COMMON_SOURCES
"${AOM_ROOT}/av1/av1_iface_common.h" "${AOM_ROOT}/av1/av1_iface_common.h"
"${AOM_ROOT}/av1/common/alloccommon.c" "${AOM_ROOT}/av1/common/alloccommon.c"
@ -16,11 +19,11 @@ set(AOM_AV1_COMMON_SOURCES
"${AOM_ROOT}/av1/common/av1_fwd_txfm1d.c" "${AOM_ROOT}/av1/common/av1_fwd_txfm1d.c"
"${AOM_ROOT}/av1/common/av1_fwd_txfm1d.h" "${AOM_ROOT}/av1/common/av1_fwd_txfm1d.h"
"${AOM_ROOT}/av1/common/av1_fwd_txfm2d.c" "${AOM_ROOT}/av1/common/av1_fwd_txfm2d.c"
"${AOM_ROOT}/av1/common/av1_fwd_txfm2d_cfg.h" "${AOM_ROOT}/av1/common/av1_fwd_txfm1d_cfg.h"
"${AOM_ROOT}/av1/common/av1_inv_txfm1d.c" "${AOM_ROOT}/av1/common/av1_inv_txfm1d.c"
"${AOM_ROOT}/av1/common/av1_inv_txfm1d.h" "${AOM_ROOT}/av1/common/av1_inv_txfm1d.h"
"${AOM_ROOT}/av1/common/av1_inv_txfm2d.c" "${AOM_ROOT}/av1/common/av1_inv_txfm2d.c"
"${AOM_ROOT}/av1/common/av1_inv_txfm2d_cfg.h" "${AOM_ROOT}/av1/common/av1_inv_txfm1d_cfg.h"
"${AOM_ROOT}/av1/common/av1_loopfilter.c" "${AOM_ROOT}/av1/common/av1_loopfilter.c"
"${AOM_ROOT}/av1/common/av1_loopfilter.h" "${AOM_ROOT}/av1/common/av1_loopfilter.h"
"${AOM_ROOT}/av1/common/av1_txfm.h" "${AOM_ROOT}/av1/common/av1_txfm.h"
@ -60,7 +63,6 @@ set(AOM_AV1_COMMON_SOURCES
"${AOM_ROOT}/av1/common/reconintra.h" "${AOM_ROOT}/av1/common/reconintra.h"
"${AOM_ROOT}/av1/common/resize.c" "${AOM_ROOT}/av1/common/resize.c"
"${AOM_ROOT}/av1/common/resize.h" "${AOM_ROOT}/av1/common/resize.h"
"${AOM_ROOT}/av1/common/restoration.h"
"${AOM_ROOT}/av1/common/scale.c" "${AOM_ROOT}/av1/common/scale.c"
"${AOM_ROOT}/av1/common/scale.h" "${AOM_ROOT}/av1/common/scale.h"
"${AOM_ROOT}/av1/common/scan.c" "${AOM_ROOT}/av1/common/scan.c"
@ -146,13 +148,9 @@ set(AOM_AV1_ENCODER_SOURCES
"${AOM_ROOT}/av1/encoder/tokenize.c" "${AOM_ROOT}/av1/encoder/tokenize.c"
"${AOM_ROOT}/av1/encoder/tokenize.h" "${AOM_ROOT}/av1/encoder/tokenize.h"
"${AOM_ROOT}/av1/encoder/treewriter.c" "${AOM_ROOT}/av1/encoder/treewriter.c"
"${AOM_ROOT}/av1/encoder/treewriter.h" "${AOM_ROOT}/av1/encoder/treewriter.h")
"${AOM_ROOT}/av1/encoder/variance_tree.c"
"${AOM_ROOT}/av1/encoder/variance_tree.h")
set(AOM_AV1_COMMON_INTRIN_SSE2 set(AOM_AV1_COMMON_INTRIN_SSE2
# Requires CONFIG_GLOBAL_MOTION or CONFIG_WARPED_MOTION
#"${AOM_ROOT}/av1/common/x86/warp_plane_sse2.c"
"${AOM_ROOT}/av1/common/x86/idct_intrin_sse2.c") "${AOM_ROOT}/av1/common/x86/idct_intrin_sse2.c")
set(AOM_AV1_COMMON_INTRIN_SSSE3 set(AOM_AV1_COMMON_INTRIN_SSSE3
@ -293,6 +291,27 @@ if (CONFIG_ACCOUNTING)
"${AOM_ROOT}/av1/decoder/accounting.h") "${AOM_ROOT}/av1/decoder/accounting.h")
endif () endif ()
if (CONFIG_GLOBAL_MOTION)
set(AOM_AV1_ENCODER_SOURCES
${AOM_AV1_ENCODER_SOURCES}
"${AOM_ROOT}/av1/encoder/corner_detect.c"
"${AOM_ROOT}/av1/encoder/corner_detect.h"
"${AOM_ROOT}/av1/encoder/corner_match.c"
"${AOM_ROOT}/av1/encoder/corner_match.h"
"${AOM_ROOT}/av1/encoder/global_motion.c"
"${AOM_ROOT}/av1/encoder/global_motion.h"
"${AOM_ROOT}/av1/encoder/ransac.c"
"${AOM_ROOT}/av1/encoder/ransac.h"
"${AOM_ROOT}/third_party/fastfeat/fast_9.c"
"${AOM_ROOT}/third_party/fastfeat/fast.c"
"${AOM_ROOT}/third_party/fastfeat/fast.h"
"${AOM_ROOT}/third_party/fastfeat/nonmax.c")
set(AOM_AV1_ENCODER_INTRIN_SSE4_1
${AOM_AV1_ENCODER_INTRIN_SSE4_1}
"${AOM_ROOT}/av1/encoder/x86/corner_match_sse4.c")
endif ()
if (CONFIG_INSPECTION) if (CONFIG_INSPECTION)
set(AOM_AV1_DECODER_SOURCES set(AOM_AV1_DECODER_SOURCES
${AOM_AV1_DECODER_SOURCES} ${AOM_AV1_DECODER_SOURCES}
@ -320,6 +339,22 @@ if (CONFIG_CFL)
"${AOM_ROOT}/av1/common/cfl.h") "${AOM_ROOT}/av1/common/cfl.h")
endif () endif ()
if (CONFIG_LOOP_RESTORATION)
set(AOM_AV1_COMMON_SOURCES
${AOM_AV1_COMMON_SOURCES}
"${AOM_ROOT}/av1/common/restoration.c"
"${AOM_ROOT}/av1/common/restoration.h")
set(AOM_AV1_COMMON_INTRIN_SSE4_1
${AOM_AV1_COMMON_INTRIN_SSE4_1}
"${AOM_ROOT}/av1/common/x86/selfguided_sse4.c")
set(AOM_AV1_ENCODER_SOURCES
${AOM_AV1_ENCODER_SOURCES}
"${AOM_ROOT}/av1/encoder/pickrst.c"
"${AOM_ROOT}/av1/encoder/pickrst.h")
endif ()
if (CONFIG_PVQ) if (CONFIG_PVQ)
set(AOM_AV1_COMMON_SOURCES set(AOM_AV1_COMMON_SOURCES
${AOM_AV1_COMMON_SOURCES} ${AOM_AV1_COMMON_SOURCES}
@ -382,7 +417,7 @@ if (CONFIG_PVQ)
endif () endif ()
endif () endif ()
if (CONFIG_WARPED_MOTION) if (CONFIG_WARPED_MOTION OR CONFIG_GLOBAL_MOTION)
set(AOM_AV1_COMMON_SOURCES set(AOM_AV1_COMMON_SOURCES
${AOM_AV1_COMMON_SOURCES} ${AOM_AV1_COMMON_SOURCES}
"${AOM_ROOT}/av1/common/warped_motion.c" "${AOM_ROOT}/av1/common/warped_motion.c"
@ -391,6 +426,16 @@ if (CONFIG_WARPED_MOTION)
set(AOM_AV1_COMMON_INTRIN_SSE2 set(AOM_AV1_COMMON_INTRIN_SSE2
${AOM_AV1_COMMON_INTRIN_SSE2} ${AOM_AV1_COMMON_INTRIN_SSE2}
"${AOM_ROOT}/av1/common/x86/warp_plane_sse2.c") "${AOM_ROOT}/av1/common/x86/warp_plane_sse2.c")
set(AOM_AV1_COMMON_INTRIN_SSSE3
${AOM_AV1_COMMON_INTRIN_SSSE3}
"${AOM_ROOT}/av1/common/x86/warp_plane_ssse3.c")
if (CONFIG_HIGHBITDEPTH)
set(AOM_AV1_COMMON_INTRIN_SSSE3
${AOM_AV1_COMMON_INTRIN_SSSE3}
"${AOM_ROOT}/av1/common/x86/highbd_warp_plane_ssse3.c")
endif ()
endif () endif ()
# Setup AV1 common/decoder/encoder targets. The libaom target must exist before # Setup AV1 common/decoder/encoder targets. The libaom target must exist before
@ -516,3 +561,5 @@ endfunction ()
function (setup_av1_test_targets) function (setup_av1_test_targets)
endfunction () endfunction ()
endif () # AOM_AV1_AV1_CMAKE_

6
third_party/aom/av1/av1_common.mk vendored
View file

@ -58,7 +58,6 @@ AV1_COMMON_SRCS-yes += common/reconinter.c
AV1_COMMON_SRCS-yes += common/reconintra.c AV1_COMMON_SRCS-yes += common/reconintra.c
AV1_COMMON_SRCS-yes += common/resize.c AV1_COMMON_SRCS-yes += common/resize.c
AV1_COMMON_SRCS-yes += common/resize.h AV1_COMMON_SRCS-yes += common/resize.h
AV1_COMMON_SRCS-yes += common/restoration.h
AV1_COMMON_SRCS-yes += common/common_data.h AV1_COMMON_SRCS-yes += common/common_data.h
AV1_COMMON_SRCS-yes += common/scan.c AV1_COMMON_SRCS-yes += common/scan.c
AV1_COMMON_SRCS-yes += common/scan.h AV1_COMMON_SRCS-yes += common/scan.h
@ -69,9 +68,9 @@ AV1_COMMON_SRCS-yes += common/av1_fwd_txfm1d.c
AV1_COMMON_SRCS-yes += common/av1_inv_txfm1d.h AV1_COMMON_SRCS-yes += common/av1_inv_txfm1d.h
AV1_COMMON_SRCS-yes += common/av1_inv_txfm1d.c AV1_COMMON_SRCS-yes += common/av1_inv_txfm1d.c
AV1_COMMON_SRCS-yes += common/av1_fwd_txfm2d.c AV1_COMMON_SRCS-yes += common/av1_fwd_txfm2d.c
AV1_COMMON_SRCS-yes += common/av1_fwd_txfm2d_cfg.h AV1_COMMON_SRCS-yes += common/av1_fwd_txfm1d_cfg.h
AV1_COMMON_SRCS-yes += common/av1_inv_txfm2d.c AV1_COMMON_SRCS-yes += common/av1_inv_txfm2d.c
AV1_COMMON_SRCS-yes += common/av1_inv_txfm2d_cfg.h AV1_COMMON_SRCS-yes += common/av1_inv_txfm1d_cfg.h
AV1_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/av1_convolve_ssse3.c AV1_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/av1_convolve_ssse3.c
ifeq ($(CONFIG_HIGHBITDEPTH),yes) ifeq ($(CONFIG_HIGHBITDEPTH),yes)
AV1_COMMON_SRCS-$(HAVE_SSE4_1) += common/x86/av1_highbd_convolve_sse4.c AV1_COMMON_SRCS-$(HAVE_SSE4_1) += common/x86/av1_highbd_convolve_sse4.c
@ -172,6 +171,7 @@ endif
ifneq ($(findstring yes,$(CONFIG_GLOBAL_MOTION) $(CONFIG_WARPED_MOTION)),) ifneq ($(findstring yes,$(CONFIG_GLOBAL_MOTION) $(CONFIG_WARPED_MOTION)),)
AV1_COMMON_SRCS-$(HAVE_SSE2) += common/x86/warp_plane_sse2.c AV1_COMMON_SRCS-$(HAVE_SSE2) += common/x86/warp_plane_sse2.c
AV1_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/warp_plane_ssse3.c
ifeq ($(CONFIG_HIGHBITDEPTH),yes) ifeq ($(CONFIG_HIGHBITDEPTH),yes)
AV1_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/highbd_warp_plane_ssse3.c AV1_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/highbd_warp_plane_ssse3.c
endif endif

7
third_party/aom/av1/av1_cx.mk vendored
View file

@ -23,8 +23,6 @@ AV1_CX_SRCS-yes += encoder/av1_quantize.h
AV1_CX_SRCS-yes += encoder/bitstream.c AV1_CX_SRCS-yes += encoder/bitstream.c
AV1_CX_SRCS-yes += encoder/context_tree.c AV1_CX_SRCS-yes += encoder/context_tree.c
AV1_CX_SRCS-yes += encoder/context_tree.h AV1_CX_SRCS-yes += encoder/context_tree.h
AV1_CX_SRCS-yes += encoder/variance_tree.c
AV1_CX_SRCS-yes += encoder/variance_tree.h
AV1_CX_SRCS-yes += encoder/cost.h AV1_CX_SRCS-yes += encoder/cost.h
AV1_CX_SRCS-yes += encoder/cost.c AV1_CX_SRCS-yes += encoder/cost.c
AV1_CX_SRCS-yes += encoder/dct.c AV1_CX_SRCS-yes += encoder/dct.c
@ -38,6 +36,7 @@ AV1_CX_SRCS-yes += encoder/ethread.h
AV1_CX_SRCS-yes += encoder/ethread.c AV1_CX_SRCS-yes += encoder/ethread.c
AV1_CX_SRCS-yes += encoder/extend.c AV1_CX_SRCS-yes += encoder/extend.c
AV1_CX_SRCS-yes += encoder/firstpass.c AV1_CX_SRCS-yes += encoder/firstpass.c
AV1_CX_SRCS-yes += encoder/mathutils.h
AV1_CX_SRCS-$(CONFIG_GLOBAL_MOTION) += ../third_party/fastfeat/fast.h AV1_CX_SRCS-$(CONFIG_GLOBAL_MOTION) += ../third_party/fastfeat/fast.h
AV1_CX_SRCS-$(CONFIG_GLOBAL_MOTION) += ../third_party/fastfeat/nonmax.c AV1_CX_SRCS-$(CONFIG_GLOBAL_MOTION) += ../third_party/fastfeat/nonmax.c
AV1_CX_SRCS-$(CONFIG_GLOBAL_MOTION) += ../third_party/fastfeat/fast_9.c AV1_CX_SRCS-$(CONFIG_GLOBAL_MOTION) += ../third_party/fastfeat/fast_9.c
@ -162,4 +161,8 @@ AV1_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/fdct16x16_msa.c
AV1_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/fdct_msa.h AV1_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/fdct_msa.h
AV1_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/temporal_filter_msa.c AV1_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/temporal_filter_msa.c
ifeq ($(CONFIG_GLOBAL_MOTION),yes)
AV1_CX_SRCS-$(HAVE_SSE4_1) += encoder/x86/corner_match_sse4.c
endif
AV1_CX_SRCS-yes := $(filter-out $(AV1_CX_SRCS_REMOVE-yes),$(AV1_CX_SRCS-yes)) AV1_CX_SRCS-yes := $(filter-out $(AV1_CX_SRCS_REMOVE-yes),$(AV1_CX_SRCS-yes))

Some files were not shown because too many files have changed in this diff Show more