diff options
author | Frank Barchard <fbarchard@google.com> | 2023-02-13 10:52:58 -0800 |
---|---|---|
committer | libyuv LUCI CQ <libyuv-scoped@luci-project-accounts.iam.gserviceaccount.com> | 2023-02-13 20:14:57 +0000 |
commit | 2bdc210be9eb11ded16bf3ef1f6cadb0d4dcb0c2 (patch) | |
tree | c446d71538c965d0e5391ef77cd49b45ba51463d /source/convert_from_argb.cc | |
parent | b2528b0be934de1918e20c85fc170d809eeb49ab (diff) | |
download | libyuv-2bdc210be9eb11ded16bf3ef1f6cadb0d4dcb0c2.tar.gz |
MergeUV_AVX512BW for I420ToNV12
On Skylake Xeon 640x360 100000 iterations
AVX512 MergeUVPlane_Opt (1196 ms)
AVX2 MergeUVPlane_Opt (1565 ms)
SSE2 MergeUVPlane_Opt (1780 ms)
Pixel 7 MergeUVPlane_Opt (1177 ms)
Bug: None
Change-Id: If47d4fa957cf27781bba5fd6a2f0bf554101a5c6
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/4242247
Commit-Queue: Frank Barchard <fbarchard@chromium.org>
Reviewed-by: richard winterton <rrwinterton@gmail.com>
Diffstat (limited to 'source/convert_from_argb.cc')
-rw-r--r-- | source/convert_from_argb.cc | 40 |
1 files changed, 40 insertions, 0 deletions
diff --git a/source/convert_from_argb.cc b/source/convert_from_argb.cc index d548aec2..1b8572a0 100644 --- a/source/convert_from_argb.cc +++ b/source/convert_from_argb.cc @@ -389,6 +389,14 @@ int ARGBToNV12(const uint8_t* src_argb, } } #endif +#if defined(HAS_MERGEUVROW_AVX512BW) + if (TestCpuFlag(kCpuHasAVX512BW)) { + MergeUVRow_ = MergeUVRow_Any_AVX512BW; + if (IS_ALIGNED(halfwidth, 32)) { + MergeUVRow_ = MergeUVRow_AVX512BW; + } + } +#endif #if defined(HAS_MERGEUVROW_NEON) if (TestCpuFlag(kCpuHasNEON)) { MergeUVRow_ = MergeUVRow_Any_NEON; @@ -559,6 +567,14 @@ int ARGBToNV21(const uint8_t* src_argb, } } #endif +#if defined(HAS_MERGEUVROW_AVX512BW) + if (TestCpuFlag(kCpuHasAVX512BW)) { + MergeUVRow_ = MergeUVRow_Any_AVX512BW; + if (IS_ALIGNED(halfwidth, 64)) { + MergeUVRow_ = MergeUVRow_AVX512BW; + } + } +#endif #if defined(HAS_MERGEUVROW_NEON) if (TestCpuFlag(kCpuHasNEON)) { MergeUVRow_ = MergeUVRow_Any_NEON; @@ -726,6 +742,14 @@ int ABGRToNV12(const uint8_t* src_abgr, } } #endif +#if defined(HAS_MERGEUVROW_AVX512BW) + if (TestCpuFlag(kCpuHasAVX512BW)) { + MergeUVRow_ = MergeUVRow_Any_AVX512BW; + if (IS_ALIGNED(halfwidth, 64)) { + MergeUVRow_ = MergeUVRow_AVX512BW; + } + } +#endif #if defined(HAS_MERGEUVROW_NEON) if (TestCpuFlag(kCpuHasNEON)) { MergeUVRow_ = MergeUVRow_Any_NEON; @@ -894,6 +918,14 @@ int ABGRToNV21(const uint8_t* src_abgr, } } #endif +#if defined(HAS_MERGEUVROW_AVX512BW) + if (TestCpuFlag(kCpuHasAVX512BW)) { + MergeUVRow_ = MergeUVRow_Any_AVX512BW; + if (IS_ALIGNED(halfwidth, 64)) { + MergeUVRow_ = MergeUVRow_AVX512BW; + } + } +#endif #if defined(HAS_MERGEUVROW_NEON) if (TestCpuFlag(kCpuHasNEON)) { MergeUVRow_ = MergeUVRow_Any_NEON; @@ -2921,6 +2953,14 @@ int RAWToJNV21(const uint8_t* src_raw, } } #endif +#if defined(HAS_MERGEUVROW_AVX512BW) + if (TestCpuFlag(kCpuHasAVX512BW)) { + MergeUVRow_ = MergeUVRow_Any_AVX512BW; + if (IS_ALIGNED(halfwidth, 64)) { + MergeUVRow_ = MergeUVRow_AVX512BW; + } + } +#endif #if defined(HAS_MERGEUVROW_NEON) if (TestCpuFlag(kCpuHasNEON)) { MergeUVRow_ = MergeUVRow_Any_NEON; |