mirror of
https://chromium.googlesource.com/libyuv/libyuv
synced 2026-11-09 14:49:57 +08:00
bad060c3a5
- Port rotate_gcc.cc to intrinsics in rotate_win.cc: TransposeWx8_SSSE3,
TransposeUVWx8_SSE2, TransposeWx16_AVX2, TransposeUVWx16_AVX2,
Transpose4x4_32_SSE2, Transpose4x4_32_AVX2, and for 64 bit,
TransposeWx16_AVX512BW and TransposeUVWx16_AVX512BW.
- Remove the 32 bit Visual C inline assembly.
- Enable rotate_win.cc for Visual C 32/64 bit, LIBYUV_ENABLE_ROWWIN,
which MemorySanitizer also enables, and all 32 bit x86 compilers.
Visual C 64 bit previously used C.
- rotate_gcc.cc is now 64 bit only. 32 bit gcc and clang only had an
8x8 TransposeWx8_SSSE3 and Transpose4x4_32, because the UV and AVX2
inline assembly uses xmm8..xmm15, so 32 bit NV12 rotate used C.
Remove the 32 bit TransposeWx8_SSSE3 inline assembly.
- Add a target("sse2") attribute to the SSE2 intrinsics so gcc and
clang build them when the baseline lacks SSE2, such as
-march=pentium3.
- Disable the generic x86 scale_row.h and compare_row.h HAS_ macros for
32 bit LIBYUV_ENABLE_ROWWIN, as is done for 64 bit. scale_gcc.cc and
compare_gcc.cc are disabled for ROWWIN and scale_win.cc and
compare_win.cc require _MSC_VER, so out/rowwin_x86 failed to link with
undefined symbols such as ScaleARGBCols_SSE2 and HashDjb2_SSE41.
- TransposeWx8_SSSE3 and TransposeUVWx8_SSE2 share a 16x8 helper.
- TransposeWx8_SSSE3 requires a multiple of 16 on all platforms.
- Byte transposes use a shared helper with dst_a / dst_b outputs so
2, 3 and 4 channel variants can reuse the same shuffles.
Functions available:
Function rotate_gcc rotate_win rotate_win
64 bit 64 bit 32 bit
TransposeWx8_SSSE3 asm intrinsics intrinsics
TransposeUVWx8_SSE2 asm intrinsics intrinsics
TransposeWx16_AVX2 asm intrinsics intrinsics
TransposeUVWx16_AVX2 asm intrinsics intrinsics
Transpose4x4_32_SSE2 asm intrinsics intrinsics
Transpose4x4_32_AVX2 asm intrinsics intrinsics
TransposeWx16_AVX512BW asm intrinsics -
TransposeUVWx16_AVX512BW asm intrinsics -
32 bit x86 Linux clang on AMD Milan (Zen 3), ms
(LIBYUV_WIDTH=1280 LIBYUV_HEIGHT=720 LIBYUV_REPEAT=2000):
C Before After SSE2 After AVX2
I420Rotate90_Opt 944 321 322 220
NV12Rotate90_Opt 1412 935 315 232
Testing:
for w in 1 2 3 7 8 9 15 16 17 23 24 31 32 33 64 65 1279 1280; do
for h in 1 7 8 15 16 17 24 31 32 33 719; do
LIBYUV_WIDTH=$w LIBYUV_HEIGHT=$h out/rowwin/libyuv_unittest \
--gtest_filter='*Rotate*:*Transpose*'
done
done
Repeated with LIBYUV_DISABLE_AVX2=1, LIBYUV_DISABLE_AVX512BW=1, with
out/Release and out/Release_x86, under sde64 -skx for AVX512BW, and
with out/msan (is_msan = true) and out/rowwin_x86
(target_cpu = "x86", libyuv_enable_rowwin = true). Built out/android_x86
(target_os = "android", target_cpu = "x86"). Compiled rotate_win.cc
with g++ and clang++ -m32 -march=pentium3.
Bug: None
Change-Id: Ic70c4f0b7eb46b1b8522a1eba851dac5f506ec3e
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/8499394
Reviewed-by: richard winterton <rrwinterton@gmail.com>
Commit-Queue: Frank Barchard <fbarchard@google.com>
309 lines
11 KiB
C++
309 lines
11 KiB
C++
/*
|
|
* Copyright 2013 The LibYuv Project Authors. All rights reserved.
|
|
*
|
|
* Use of this source code is governed by a BSD-style license
|
|
* that can be found in the LICENSE file in the root of the source
|
|
* tree. An additional intellectual property rights grant can be found
|
|
* in the file PATENTS. All contributing project authors may
|
|
* be found in the AUTHORS file in the root of the source tree.
|
|
*/
|
|
|
|
#ifndef INCLUDE_LIBYUV_ROTATE_ROW_H_
|
|
#define INCLUDE_LIBYUV_ROTATE_ROW_H_
|
|
|
|
#include <stddef.h>
|
|
#include <stdint.h>
|
|
|
|
#include "libyuv/cpu_support.h"
|
|
|
|
#ifdef __cplusplus
|
|
namespace libyuv {
|
|
extern "C" {
|
|
#endif
|
|
|
|
// The following are available for 32 bit x86, Visual C 64 bit, and for clang
|
|
// with LIBYUV_ENABLE_ROWWIN, which is also enabled for MemorySanitizer. These
|
|
// are intrinsics in rotate_win.cc.
|
|
#if !defined(LIBYUV_DISABLE_X86) && \
|
|
(defined(__i386__) || defined(_M_IX86) || \
|
|
((defined(__x86_64__) || defined(_M_X64)) && \
|
|
((defined(_MSC_VER) && !defined(__clang__)) || \
|
|
defined(LIBYUV_ENABLE_ROWWIN))))
|
|
#define HAS_TRANSPOSEWX8_SSSE3
|
|
#define HAS_TRANSPOSEUVWX8_SSE2
|
|
#define HAS_TRANSPOSE4X4_32_SSE2
|
|
#define HAS_TRANSPOSE4X4_32_AVX2
|
|
#define HAS_TRANSPOSEWX16_AVX2
|
|
#define HAS_TRANSPOSEUVWX16_AVX2
|
|
#if defined(__x86_64__) || defined(_M_X64)
|
|
#define HAS_TRANSPOSEWX16_AVX512BW
|
|
#define HAS_TRANSPOSEUVWX16_AVX512BW
|
|
#endif
|
|
#endif
|
|
|
|
// The following are available for 64 bit GCC and clang, as inline assembly in
|
|
// rotate_gcc.cc:
|
|
#if !defined(LIBYUV_DISABLE_X86) && defined(__x86_64__) && \
|
|
!defined(LIBYUV_ENABLE_ROWWIN)
|
|
#define HAS_TRANSPOSEWX8_SSSE3
|
|
#define HAS_TRANSPOSEUVWX8_SSE2
|
|
#define HAS_TRANSPOSE4X4_32_SSE2
|
|
#define HAS_TRANSPOSE4X4_32_AVX2
|
|
#define HAS_TRANSPOSEWX16_AVX2
|
|
#define HAS_TRANSPOSEUVWX16_AVX2
|
|
#endif
|
|
|
|
// The following are available for AVX512 clang x64 platforms:
|
|
#if !defined(LIBYUV_DISABLE_X86) && defined(__x86_64__) && \
|
|
defined(CLANG_HAS_AVX512) && !defined(LIBYUV_ENABLE_ROWWIN)
|
|
#define HAS_TRANSPOSEWX16_AVX512BW
|
|
#define HAS_TRANSPOSEUVWX16_AVX512BW
|
|
#endif
|
|
|
|
#if !defined(LIBYUV_DISABLE_NEON) && \
|
|
(defined(__ARM_NEON__) || defined(LIBYUV_NEON) || defined(__aarch64__))
|
|
#if defined(__aarch64__)
|
|
#define HAS_TRANSPOSEWX16_NEON
|
|
#else
|
|
#define HAS_TRANSPOSEWX8_NEON
|
|
#endif
|
|
#define HAS_TRANSPOSEUVWX8_NEON
|
|
#define HAS_TRANSPOSE4X4_32_NEON
|
|
#endif
|
|
|
|
#if !defined(LIBYUV_DISABLE_SME) && defined(CLANG_HAS_SME) && \
|
|
defined(__aarch64__)
|
|
#define HAS_TRANSPOSEWXH_SME
|
|
#define HAS_TRANSPOSEUVWXH_SME
|
|
#endif
|
|
|
|
#if !defined(LIBYUV_DISABLE_LSX) && defined(__loongarch_sx)
|
|
#define HAS_TRANSPOSEWX16_LSX
|
|
#define HAS_TRANSPOSEUVWX16_LSX
|
|
#endif
|
|
|
|
void TransposeWxH_C(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width,
|
|
int height);
|
|
|
|
void TransposeWx8_C(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_C(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx8_NEON(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_NEON(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWxH_SME(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width,
|
|
int height);
|
|
void TransposeWx8_SSSE3(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_AVX2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_AVX512BW(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_LSX(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
|
|
void TransposeWx8_Any_NEON(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_Any_NEON(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx8_Any_SSSE3(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_Any_AVX2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx16_Any_LSX(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
|
|
void TransposeUVWxH_C(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width,
|
|
int height);
|
|
|
|
void TransposeUVWx8_C(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx16_C(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx8_SSE2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx16_AVX2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx16_AVX512BW(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx8_NEON(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWxH_SME(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width,
|
|
int height);
|
|
void TransposeUVWx16_LSX(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
|
|
void TransposeUVWx8_Any_SSE2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx8_Any_NEON(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx16_Any_AVX2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeUVWx16_Any_LSX(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
uint8_t* dst_b,
|
|
int dst_stride_b,
|
|
int width);
|
|
void TransposeWxH_16_C(const uint16_t* src,
|
|
int src_stride,
|
|
uint16_t* dst,
|
|
int dst_stride,
|
|
int width,
|
|
int height);
|
|
|
|
void TransposeWx8_16_C(const uint16_t* src,
|
|
int src_stride,
|
|
uint16_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
void TransposeWx1_16_C(const uint16_t* src,
|
|
int src_stride,
|
|
uint16_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
|
|
// Transpose 32 bit values (ARGB)
|
|
void Transpose4x4_32_NEON(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
|
|
void Transpose4x4_32_SSE2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
|
|
void Transpose4x4_32_AVX2(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
|
|
void Transpose4x4_32_C(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width);
|
|
|
|
#ifdef __cplusplus
|
|
} // extern "C"
|
|
} // namespace libyuv
|
|
#endif
|
|
|
|
#endif // INCLUDE_LIBYUV_ROTATE_ROW_H_
|