| /external/libgav1/libgav1/src/dsp/ |
| D | loop_restoration.cc | 280 uint32_t* const* square_sum5) { in BoxSum() argument
301 (*square_sum5)[x] = square_sum + source0 * source0; in BoxSum()
307 ++square_sum5; in BoxSum()
365 const uint16_t* const sum5[5], const uint32_t* const square_sum5[5], in BoxFilterPreProcess5()
373 a += square_sum5[dy][x]; in BoxFilterPreProcess5()
474 uint32_t* const square_sum5[5], const int width, in BoxFilterPass1()
479 BoxFilterPreProcess5<bitdepth>(sum5, square_sum5, width, scale, sgr_buffer, in BoxFilterPass1()
516 uint32_t* const square_sum5[5], const int width, in BoxFilter()
523 BoxFilterPreProcess5<bitdepth>(sum5, square_sum5, width, scales[0], in BoxFilter()
563 uint32_t *square_sum3[4], *square_sum5[5], *b343[4], *b444[3], *b565[2]; in BoxFilterProcess() local
[all …]
|
| D | common.h | 53 alignas(kMaxAlignment) uint32_t square_sum5[5 * kSgrStride]; member
|
| /external/libgav1/libgav1/src/dsp/arm/ |
| D | loop_restoration_neon.cc | 908 uint32_t* square_sum3, uint32_t* square_sum5) { in BoxSum() argument
927 vst1q_u32(square_sum5 + 0, row_sq5.val[0]); in BoxSum()
928 vst1q_u32(square_sum5 + 4, row_sq5.val[1]); in BoxSum()
934 square_sum5 += 8; in BoxSum()
1147 uint8x16_t s[2][2], uint16_t* const sum5[5], uint32_t* const square_sum5[5], in BoxFilterPreProcess5Lo()
1163 vst1q_u32(square_sum5[3] + 0, sq5[3].val[0]); in BoxFilterPreProcess5Lo()
1164 vst1q_u32(square_sum5[3] + 4, sq5[3].val[1]); in BoxFilterPreProcess5Lo()
1165 vst1q_u32(square_sum5[4] + 0, sq5[4].val[0]); in BoxFilterPreProcess5Lo()
1166 vst1q_u32(square_sum5[4] + 4, sq5[4].val[1]); in BoxFilterPreProcess5Lo()
1170 sq5[0].val[0] = vld1q_u32(square_sum5[0] + 0); in BoxFilterPreProcess5Lo()
[all …]
|
| /external/libgav1/libgav1/src/dsp/x86/ |
| D | loop_restoration_10bit_sse4.cc | 897 uint32_t* square_sum3, uint32_t* square_sum5) { in BoxSum() argument
922 StoreAligned32U32(square_sum5 + 0, row_sq5); in BoxSum()
926 StoreAligned32U32(square_sum5 + 8, row_sq5); in BoxSum()
933 square_sum5 += 16; in BoxSum()
939 square_sum5 += sum_stride - sum_width; in BoxSum()
1299 uint32_t* const square_sum5[5], __m128i sq[2][8], __m128i* const ma, in BoxFilterPreProcess5Lo()
1309 StoreAligned32U32(square_sum5[3], sq5[3]); in BoxFilterPreProcess5Lo()
1311 StoreAligned32U32(square_sum5[4], sq5[4]); in BoxFilterPreProcess5Lo()
1313 LoadAligned32x3U32(square_sum5, 0, sq5); in BoxFilterPreProcess5Lo()
1320 uint32_t* const square_sum5[5], __m128i sq[2][8], __m128i ma[2], in BoxFilterPreProcess5()
[all …]
|
| D | loop_restoration_sse4.cc | 1053 uint32_t* square_sum3, uint32_t* square_sum5) { in BoxSum() argument
1073 StoreAligned32U32(square_sum5 + 0, row_sq5); in BoxSum()
1076 StoreAligned32U32(square_sum5 + 8, row_sq5); in BoxSum()
1082 square_sum5 += 16; in BoxSum()
1088 square_sum5 += sum_stride - sum_width; in BoxSum()
1437 uint32_t* const square_sum5[5], __m128i sq[2][4], __m128i* const ma, in BoxFilterPreProcess5Lo()
1447 StoreAligned32U32(square_sum5[3], sq5[3]); in BoxFilterPreProcess5Lo()
1449 StoreAligned32U32(square_sum5[4], sq5[4]); in BoxFilterPreProcess5Lo()
1451 LoadAligned32x3U32(square_sum5, 0, sq5); in BoxFilterPreProcess5Lo()
1458 uint32_t* const square_sum5[5], __m128i sq[2][4], __m128i ma[2], in BoxFilterPreProcess5()
[all …]
|
| D | loop_restoration_10bit_avx2.cc | 1137 uint32_t* square_sum3, uint32_t* square_sum5) { in BoxSum() argument
1155 StoreAligned32U32(square_sum5, sq5); in BoxSum()
1160 square_sum5 += 8; in BoxSum()
1185 StoreAligned64(square_sum5 + 0, row_sq5); in BoxSum()
1189 StoreAligned64(square_sum5 + 16, row_sq5); in BoxSum()
1196 square_sum5 += 32; in BoxSum()
1203 square_sum5 += sum_stride - sum_width - 8; in BoxSum()
1829 uint32_t* const square_sum5[5], __m128i sq[2][4], __m128i* const ma, in BoxFilterPreProcess5Lo()
1839 StoreAligned32U32(square_sum5[3], sq5[3]); in BoxFilterPreProcess5Lo()
1841 StoreAligned32U32(square_sum5[4], sq5[4]); in BoxFilterPreProcess5Lo()
[all …]
|
| D | loop_restoration_avx2.cc | 1245 uint32_t* square_sum3, uint32_t* square_sum5) { in BoxSum() argument
1259 StoreAligned32U32(square_sum5, sq5); in BoxSum()
1264 square_sum5 += 8; in BoxSum()
1280 StoreAligned64(square_sum5 + 0, row_sq5); in BoxSum()
1283 StoreAligned64(square_sum5 + 16, row_sq5); in BoxSum()
1289 square_sum5 += 32; in BoxSum()
1296 square_sum5 += sum_stride - sum_width - 8; in BoxSum()
1748 uint32_t* const square_sum5[5], __m128i sq[2][2], __m128i* const ma, in BoxFilterPreProcess5Lo()
1758 StoreAligned32U32(square_sum5[3], sq5[3]); in BoxFilterPreProcess5Lo()
1760 StoreAligned32U32(square_sum5[4], sq5[4]); in BoxFilterPreProcess5Lo()
[all …]
|