Searched refs:sum_low (Results 1 – 3 of 3) sorted by relevance
| /external/libgav1/libgav1/src/dsp/x86/ |
| D | warp_sse4.cc | 107 __m128i sum_low = _mm_set1_epi32(kOffsetRemoval); in WriteVerticalFilter() local
108 __m128i sum_high = sum_low; in WriteVerticalFilter()
123 sum_low = _mm_add_epi32(sum_low, product_low); in WriteVerticalFilter()
126 sum_low = RightShiftWithRounding_S32(sum_low, kRoundBitsVertical); in WriteVerticalFilter()
129 const __m128i sum = _mm_packs_epi32(sum_low, sum_high); in WriteVerticalFilter()
132 const __m128i sum = _mm_packus_epi32(sum_low, sum_high); in WriteVerticalFilter()
143 __m128i sum_low = _mm_setzero_si128(); in WriteVerticalFilter() local
154 sum_low = _mm_add_epi32(sum_low, product_low); in WriteVerticalFilter()
157 sum_low = RightShiftWithRounding_S32(sum_low, kRoundBitsVertical); in WriteVerticalFilter()
160 const __m128i sum = _mm_packs_epi32(sum_low, sum_high); in WriteVerticalFilter()
[all …]
|
| /external/libgav1/libgav1/src/dsp/arm/ |
| D | warp_neon.cc | 302 int32x4_t sum_low = vdupq_n_s32(0); in Warp_NEON() local
303 int32x4_t sum_high = sum_low; in Warp_NEON()
306 sum_low = in Warp_NEON()
307 vmlal_n_s16(sum_low, vget_low_s16(filter[k]), intermediate); in Warp_NEON()
312 vcombine_s16(vrshrn_n_s32(sum_low, kRoundBitsVertical), in Warp_NEON()
404 int32x4_t sum_low = vdupq_n_s32(-kOffsetRemoval); in Warp_NEON() local
405 int32x4_t sum_high = sum_low; in Warp_NEON()
408 sum_low = vmlal_s16(sum_low, vget_low_s16(filter[k]), in Warp_NEON()
414 vcombine_s16(vrshrn_n_s32(sum_low, kRoundBitsVertical), in Warp_NEON()
|
| /external/libvpx/libvpx/vpx_dsp/arm/ |
| D | deblock_neon.c | 289 static uint8x8_t combine_mask(const int16x4_t sum_low, const int16x4_t sum_high, in combine_mask() argument
293 const uint16x4_t mask16_low = calculate_mask(sum_low, sumsq_low, f, fifteen); in combine_mask()
321 int16x4_t sum_low, sum_high; in vpx_mbpost_proc_across_ip_neon() local
359 sum_low = vdup_lane_s16(sum_high, 3); in vpx_mbpost_proc_across_ip_neon()
362 accumulate_sum_sumsq(vget_low_s16(x), xy_low, &sum_low, &sumsq_low); in vpx_mbpost_proc_across_ip_neon()
366 sum_high = vdup_lane_s16(sum_low, 3); in vpx_mbpost_proc_across_ip_neon()
371 mask = combine_mask(sum_low, sum_high, sumsq_low, sumsq_high, f); in vpx_mbpost_proc_across_ip_neon()
373 output = filter_pixels(vcombine_s16(sum_low, sum_high), s); in vpx_mbpost_proc_across_ip_neon()
|