diff options
| author | Justin Green <greenjustin@google.com> | 2022-02-03 11:46:44 -0500 |
|---|---|---|
| committer | libyuv LUCI CQ <libyuv-scoped@luci-project-accounts.iam.gserviceaccount.com> | 2022-02-03 17:01:49 +0000 |
| commit | b4ddbaf549a1bf5572bf703fd2862d1eb7380c6a (patch) | |
| tree | 9e0a90646de7b1c50f40e3aeb452f749d297561b /source/row_neon64.cc | |
| parent | 804980bbab748fd0e180cd6e7d9292ff49baf704 (diff) | |
| download | libyuv-b4ddbaf549a1bf5572bf703fd2862d1eb7380c6a.tar.gz | |
Add support for MM21.
Add support for MM21 to NV12 and I420 conversion, and add SIMD
optimizations for arm, aarch64, SSE2, and SSSE3 machines.
Bug: libyuv:915, b/215425056
Change-Id: Iecb0c33287f35766a6169d4adf3b7397f1ba8b5d
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/3433269
Reviewed-by: Frank Barchard <fbarchard@chromium.org>
Commit-Queue: Justin Green <greenjustin@google.com>
Diffstat (limited to 'source/row_neon64.cc')
| -rw-r--r-- | source/row_neon64.cc | 23 |
1 files changed, 23 insertions, 0 deletions
diff --git a/source/row_neon64.cc b/source/row_neon64.cc index 1d1f9bb1..442e60cd 100644 --- a/source/row_neon64.cc +++ b/source/row_neon64.cc @@ -627,6 +627,29 @@ void DetileRow_NEON(const uint8_t* src, ); } +// Read 16 bytes of UV, detile, and write 8 bytes of U and 8 bytes of V. +void DetileSplitUVRow_NEON(const uint8_t* src_uv, + ptrdiff_t src_tile_stride, + uint8_t* dst_u, + uint8_t* dst_v, + int width) { + asm volatile( + "1: \n" + "ld2 {v0.8b,v1.8b}, [%0], %4 \n" + "subs %w3, %w3, #16 \n" + "prfm pldl1keep, [%0, 1792] \n" + "st1 {v0.8b}, [%1], #8 \n" + "st1 {v1.8b}, [%2], #8 \n" + "b.gt 1b \n" + : "+r"(src_uv), // %0 + "+r"(dst_u), // %1 + "+r"(dst_v), // %2 + "+r"(width) // %3 + : "r"(src_tile_stride) // %4 + : "cc", "memory", "v0", "v1" // Clobber List + ); +} + #if LIBYUV_USE_ST2 // Reads 16 U's and V's and writes out 16 pairs of UV. void MergeUVRow_NEON(const uint8_t* src_u, |