wuggy/Dactyloidae
wuggy/Dactyloidae
1
0
Fork 0
mirror of https://repo.dactyloidae.xyz/Dactyloidae/UXP.git synced 2026-10-24 16:17:29 +09:00
Code Issues Projects Releases Packages Wiki Activity

Update libaom to rev b25610052a1398032320008d69b51d2da94f5928

Browse source
This commit is contained in:
trav90 2018-10-19 23:00:02 -05:00 • committed by Roy Tam
commit 3d109dd4fc
240 changed files with 16977 additions and 6139 deletions
Download patch file Download diff file Expand all files Collapse all files

84
third_party/aom/av1/common/arm/mem_neon.h vendored
View file

@ -22,6 +22,14 @@ static INLINE void store_row2_u8_8x8(uint8_t *s, int p, const uint8x8_t s0,
s += p;
}
/* These intrinsics require immediate values, so we must use #defines
to enforce that. */
#define load_u8_4x1(s, s0, lane) \
do { \
*(s0) = vreinterpret_u8_u32( \
vld1_lane_u32((uint32_t *)(s), vreinterpret_u32_u8(*(s0)), lane)); \
} while (0)
static INLINE void load_u8_8x8(const uint8_t *s, ptrdiff_t p,
uint8x8_t *const s0, uint8x8_t *const s1,
uint8x8_t *const s2, uint8x8_t *const s3,
@ -128,6 +136,13 @@ static INLINE void load_s16_4x4(const int16_t *s, ptrdiff_t p,
*s3 = vld1_s16(s);
}
/* These intrinsics require immediate values, so we must use #defines
to enforce that. */
#define store_u8_4x1(s, s0, lane) \
do { \
vst1_lane_u32((uint32_t *)(s), vreinterpret_u32_u8(s0), lane); \
} while (0)
static INLINE void store_u8_8x8(uint8_t *s, ptrdiff_t p, const uint8x8_t s0,
const uint8x8_t s1, const uint8x8_t s2,
const uint8x8_t s3, const uint8x8_t s4,
@ -242,6 +257,30 @@ static INLINE void store_s16_8x8(int16_t *s, ptrdiff_t dst_stride,
vst1q_s16(s, s7);
}
static INLINE void store_s16_4x4(int16_t *s, ptrdiff_t dst_stride,
const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3) {
vst1_s16(s, s0);
s += dst_stride;
vst1_s16(s, s1);
s += dst_stride;
vst1_s16(s, s2);
s += dst_stride;
vst1_s16(s, s3);
}
static INLINE void store_s16_8x4(int16_t *s, ptrdiff_t dst_stride,
const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3) {
vst1q_s16(s, s0);
s += dst_stride;
vst1q_s16(s, s1);
s += dst_stride;
vst1q_s16(s, s2);
s += dst_stride;
vst1q_s16(s, s3);
}
static INLINE void load_s16_8x8(const int16_t *s, ptrdiff_t p,
int16x8_t *const s0, int16x8_t *const s1,
int16x8_t *const s2, int16x8_t *const s3,
@ -398,4 +437,49 @@ static INLINE void load_unaligned_u16_4x4(const uint16_t *buf, uint32_t stride,
*tu1 = vsetq_lane_u64(a, *tu1, 1);
}
static INLINE void load_s32_4x4(int32_t *s, int32_t p, int32x4_t *s1,
int32x4_t *s2, int32x4_t *s3, int32x4_t *s4) {
*s1 = vld1q_s32(s);
s += p;
*s2 = vld1q_s32(s);
s += p;
*s3 = vld1q_s32(s);
s += p;
*s4 = vld1q_s32(s);
}
static INLINE void store_s32_4x4(int32_t *s, int32_t p, int32x4_t s1,
int32x4_t s2, int32x4_t s3, int32x4_t s4) {
vst1q_s32(s, s1);
s += p;
vst1q_s32(s, s2);
s += p;
vst1q_s32(s, s3);
s += p;
vst1q_s32(s, s4);
}
static INLINE void load_u32_4x4(uint32_t *s, int32_t p, uint32x4_t *s1,
uint32x4_t *s2, uint32x4_t *s3,
uint32x4_t *s4) {
*s1 = vld1q_u32(s);
s += p;
*s2 = vld1q_u32(s);
s += p;
*s3 = vld1q_u32(s);
s += p;
*s4 = vld1q_u32(s);
}
static INLINE void store_u32_4x4(uint32_t *s, int32_t p, uint32x4_t s1,
uint32x4_t s2, uint32x4_t s3, uint32x4_t s4) {
vst1q_u32(s, s1);
s += p;
vst1q_u32(s, s2);
s += p;
vst1q_u32(s, s3);
s += p;
vst1q_u32(s, s4);
}
#endif // AV1_COMMON_ARM_MEM_NEON_H_