mirror of
https://chromium.googlesource.com/libyuv/libyuv
synced 2025-12-06 16:56:55 +08:00
4f7fd808b7
The existing Neon code only makes use of 64-bit vectors throughout which limits the performance on larger cores. To avoid this, swap the Neon code from a Wx8 implementation to a Wx16 implementation and process blocks of 16 full vectors at a time. The original code also handled widths that were not exact multiples of 16, however this should already be handled by the "any" kernel so it is removed. Finally, avoid duplicating the TransposeWx16_C fallback kernel definition in all architectures that need it, and just put it once in rotate_common.cc instead. Observed speedups for TransposePlane across a range of micro-architectures: Cortex-A53: -40.0% Cortex-A55: -20.7% Cortex-A57: -43.9% Cortex-A510: -43.5% Cortex-A520: -43.9% Cortex-A720: -31.1% Cortex-X2: -38.3% Cortex-X4: -43.6% Change-Id: Ic7c4d5f24eb27091d743ddc00cd95ef178b6984e Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/5545459 Reviewed-by: Frank Barchard <fbarchard@chromium.org>
83 lines
3.1 KiB
C++
83 lines
3.1 KiB
C++
/*
|
|
* Copyright 2015 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.
|
|
*/
|
|
|
|
#include "libyuv/rotate.h"
|
|
#include "libyuv/rotate_row.h"
|
|
|
|
#include "libyuv/basic_types.h"
|
|
|
|
#ifdef __cplusplus
|
|
namespace libyuv {
|
|
extern "C" {
|
|
#endif
|
|
|
|
#define TANY(NAMEANY, TPOS_SIMD, MASK) \
|
|
void NAMEANY(const uint8_t* src, int src_stride, uint8_t* dst, \
|
|
int dst_stride, int width) { \
|
|
int r = width & MASK; \
|
|
int n = width - r; \
|
|
if (n > 0) { \
|
|
TPOS_SIMD(src, src_stride, dst, dst_stride, n); \
|
|
} \
|
|
TransposeWx8_C(src + n, src_stride, dst + n * dst_stride, dst_stride, r); \
|
|
}
|
|
|
|
#ifdef HAS_TRANSPOSEWX8_NEON
|
|
TANY(TransposeWx8_Any_NEON, TransposeWx8_NEON, 7)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEWX16_NEON
|
|
TANY(TransposeWx16_Any_NEON, TransposeWx16_NEON, 15)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEWX8_SSSE3
|
|
TANY(TransposeWx8_Any_SSSE3, TransposeWx8_SSSE3, 7)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEWX8_FAST_SSSE3
|
|
TANY(TransposeWx8_Fast_Any_SSSE3, TransposeWx8_Fast_SSSE3, 15)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEWX16_MSA
|
|
TANY(TransposeWx16_Any_MSA, TransposeWx16_MSA, 15)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEWX16_LSX
|
|
TANY(TransposeWx16_Any_LSX, TransposeWx16_LSX, 15)
|
|
#endif
|
|
#undef TANY
|
|
|
|
#define TUVANY(NAMEANY, TPOS_SIMD, MASK) \
|
|
void NAMEANY(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 r = width & MASK; \
|
|
int n = width - r; \
|
|
if (n > 0) { \
|
|
TPOS_SIMD(src, src_stride, dst_a, dst_stride_a, dst_b, dst_stride_b, n); \
|
|
} \
|
|
TransposeUVWx8_C(src + n * 2, src_stride, dst_a + n * dst_stride_a, \
|
|
dst_stride_a, dst_b + n * dst_stride_b, dst_stride_b, r); \
|
|
}
|
|
|
|
#ifdef HAS_TRANSPOSEUVWX8_NEON
|
|
TUVANY(TransposeUVWx8_Any_NEON, TransposeUVWx8_NEON, 7)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEUVWX8_SSE2
|
|
TUVANY(TransposeUVWx8_Any_SSE2, TransposeUVWx8_SSE2, 7)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEUVWX16_MSA
|
|
TUVANY(TransposeUVWx16_Any_MSA, TransposeUVWx16_MSA, 7)
|
|
#endif
|
|
#ifdef HAS_TRANSPOSEUVWX16_LSX
|
|
TUVANY(TransposeUVWx16_Any_LSX, TransposeUVWx16_LSX, 7)
|
|
#endif
|
|
#undef TUVANY
|
|
|
|
#ifdef __cplusplus
|
|
} // extern "C"
|
|
} // namespace libyuv
|
|
#endif
|