mirror of
https://chromium.googlesource.com/libyuv/libyuv
synced 2026-10-28 18:07:12 +08:00
20b0a91bc1
- TransposeWx16_AVX2 and TransposeUVWx16_AVX2 using 2 lanes of 8 ymm registers, with Any for tail. - Move TransposeUVWx16_C to rotate_common.cc and pass the C function to TUVANY, so the LSX 16 row Any version transposes all 16 rows. I420Rotate90_Opt (LIBYUV_WIDTH=1279 LIBYUV_HEIGHT=719 LIBYUV_REPEAT=1000): CPU SSSE3 AVX2 AVX512 AVX2 x 512 x Intel Haswell 1812 916 945 1.98x 1.92x AMD Milan (Zen 3) 177 118 118 1.50x 1.50x Intel Granite Rapids 245 186 186 1.32x 1.32x Intel Emerald Rapids 246 193 176 1.27x 1.40x Intel Sapphire Rapids 278 233 225 1.19x 1.24x Intel Ice Lake 308 213 221 1.45x 1.39x Intel Skylake-X 365 267 246 1.37x 1.48x AMD Turin (Zen 5) 106 75 75 1.41x 1.41x AMD Genoa (Zen 4) 149 98 93 1.52x 1.60x AMD Venice (Zen 6) 106 74 73 1.43x 1.45x NV12Rotate90_Opt (LIBYUV_WIDTH=1279 LIBYUV_HEIGHT=719 LIBYUV_REPEAT=1000): CPU SSSE3 AVX2 AVX512 AVX2 x 512 x Intel Haswell 1771 963 935 1.84x 1.89x AMD Milan (Zen 3) 173 118 118 1.47x 1.47x Intel Granite Rapids 275 189 186 1.46x 1.48x Intel Emerald Rapids 256 188 179 1.36x 1.43x Intel Sapphire Rapids 286 232 218 1.23x 1.31x Intel Ice Lake 309 208 218 1.49x 1.42x Intel Skylake-X 380 269 251 1.41x 1.51x AMD Turin (Zen 5) 112 74 73 1.51x 1.53x AMD Genoa (Zen 4) 146 97 93 1.51x 1.57x AMD Venice (Zen 6) 114 73 72 1.56x 1.58x Bug: None Change-Id: Ia89219e3336f39a817aa662e1452824d781664a3 TAG=agy CONV=bea53b37-b2bf-46fa-9e44-a708b438fabc Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/8489456 Reviewed-by: richard winterton <rrwinterton@gmail.com>
221 lines
8.7 KiB
C++
221 lines
8.7 KiB
C++
/*
|
|
* Copyright 2022 The LibYuv Project Authors. All rights reserved.
|
|
*
|
|
* Copyright (c) 2022 Loongson Technology Corporation Limited
|
|
*
|
|
* 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_row.h"
|
|
|
|
#if !defined(LIBYUV_DISABLE_LSX) && defined(__loongarch_sx)
|
|
#include "libyuv/loongson_intrinsics.h"
|
|
|
|
#ifdef __cplusplus
|
|
namespace libyuv {
|
|
extern "C" {
|
|
#endif
|
|
|
|
#define ILVLH_B(in0, in1, in2, in3, out0, out1, out2, out3) \
|
|
{ \
|
|
DUP2_ARG2(__lsx_vilvl_b, in1, in0, in3, in2, out0, out2); \
|
|
DUP2_ARG2(__lsx_vilvh_b, in1, in0, in3, in2, out1, out3); \
|
|
}
|
|
|
|
#define ILVLH_H(in0, in1, in2, in3, out0, out1, out2, out3) \
|
|
{ \
|
|
DUP2_ARG2(__lsx_vilvl_h, in1, in0, in3, in2, out0, out2); \
|
|
DUP2_ARG2(__lsx_vilvh_h, in1, in0, in3, in2, out1, out3); \
|
|
}
|
|
|
|
#define ILVLH_W(in0, in1, in2, in3, out0, out1, out2, out3) \
|
|
{ \
|
|
DUP2_ARG2(__lsx_vilvl_w, in1, in0, in3, in2, out0, out2); \
|
|
DUP2_ARG2(__lsx_vilvh_w, in1, in0, in3, in2, out1, out3); \
|
|
}
|
|
|
|
#define ILVLH_D(in0, in1, in2, in3, out0, out1, out2, out3) \
|
|
{ \
|
|
DUP2_ARG2(__lsx_vilvl_d, in1, in0, in3, in2, out0, out2); \
|
|
DUP2_ARG2(__lsx_vilvh_d, in1, in0, in3, in2, out1, out3); \
|
|
}
|
|
|
|
#define LSX_ST_4(_dst0, _dst1, _dst2, _dst3, _dst, _stride, _stride2, \
|
|
_stride3, _stride4) \
|
|
{ \
|
|
__lsx_vst(_dst0, _dst, 0); \
|
|
__lsx_vstx(_dst1, _dst, _stride); \
|
|
__lsx_vstx(_dst2, _dst, _stride2); \
|
|
__lsx_vstx(_dst3, _dst, _stride3); \
|
|
_dst += _stride4; \
|
|
}
|
|
|
|
#define LSX_ST_2(_dst0, _dst1, _dst, _stride, _stride2) \
|
|
{ \
|
|
__lsx_vst(_dst0, _dst, 0); \
|
|
__lsx_vstx(_dst1, _dst, _stride); \
|
|
_dst += _stride2; \
|
|
}
|
|
|
|
void TransposeWx16_LSX(const uint8_t* src,
|
|
int src_stride,
|
|
uint8_t* dst,
|
|
int dst_stride,
|
|
int width) {
|
|
int x;
|
|
int len = width / 16;
|
|
uint8_t* s;
|
|
int src_stride2 = src_stride << 1;
|
|
int src_stride3 = src_stride + src_stride2;
|
|
int src_stride4 = src_stride2 << 1;
|
|
int dst_stride2 = dst_stride << 1;
|
|
int dst_stride3 = dst_stride + dst_stride2;
|
|
int dst_stride4 = dst_stride2 << 1;
|
|
__m128i src0, src1, src2, src3, dst0, dst1, dst2, dst3;
|
|
__m128i tmp0, tmp1, tmp2, tmp3;
|
|
__m128i reg0, reg1, reg2, reg3, reg4, reg5, reg6, reg7;
|
|
__m128i res0, res1, res2, res3, res4, res5, res6, res7, res8, res9;
|
|
|
|
for (x = 0; x < len; x++) {
|
|
s = (uint8_t*)src;
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg0, reg1, reg2, reg3);
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg4, reg5, reg6, reg7);
|
|
ILVLH_W(reg0, reg4, reg1, reg5, res0, res1, res2, res3);
|
|
ILVLH_W(reg2, reg6, reg3, reg7, res4, res5, res6, res7);
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg0, reg1, reg2, reg3);
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg4, reg5, reg6, reg7);
|
|
res8 = __lsx_vilvl_w(reg4, reg0);
|
|
res9 = __lsx_vilvh_w(reg4, reg0);
|
|
ILVLH_D(res0, res8, res1, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_4(dst0, dst1, dst2, dst3, dst, dst_stride, dst_stride2, dst_stride3,
|
|
dst_stride4);
|
|
res8 = __lsx_vilvl_w(reg5, reg1);
|
|
res9 = __lsx_vilvh_w(reg5, reg1);
|
|
ILVLH_D(res2, res8, res3, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_4(dst0, dst1, dst2, dst3, dst, dst_stride, dst_stride2, dst_stride3,
|
|
dst_stride4);
|
|
res8 = __lsx_vilvl_w(reg6, reg2);
|
|
res9 = __lsx_vilvh_w(reg6, reg2);
|
|
ILVLH_D(res4, res8, res5, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_4(dst0, dst1, dst2, dst3, dst, dst_stride, dst_stride2, dst_stride3,
|
|
dst_stride4);
|
|
res8 = __lsx_vilvl_w(reg7, reg3);
|
|
res9 = __lsx_vilvh_w(reg7, reg3);
|
|
ILVLH_D(res6, res8, res7, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_4(dst0, dst1, dst2, dst3, dst, dst_stride, dst_stride2, dst_stride3,
|
|
dst_stride4);
|
|
src += 16;
|
|
}
|
|
}
|
|
|
|
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) {
|
|
int x;
|
|
int len = width / 8;
|
|
uint8_t* s;
|
|
int src_stride2 = src_stride << 1;
|
|
int src_stride3 = src_stride + src_stride2;
|
|
int src_stride4 = src_stride2 << 1;
|
|
int dst_stride_a2 = dst_stride_a << 1;
|
|
int dst_stride_b2 = dst_stride_b << 1;
|
|
__m128i src0, src1, src2, src3, dst0, dst1, dst2, dst3;
|
|
__m128i tmp0, tmp1, tmp2, tmp3;
|
|
__m128i reg0, reg1, reg2, reg3, reg4, reg5, reg6, reg7;
|
|
__m128i res0, res1, res2, res3, res4, res5, res6, res7, res8, res9;
|
|
|
|
for (x = 0; x < len; x++) {
|
|
s = (uint8_t*)src;
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg0, reg1, reg2, reg3);
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg4, reg5, reg6, reg7);
|
|
ILVLH_W(reg0, reg4, reg1, reg5, res0, res1, res2, res3);
|
|
ILVLH_W(reg2, reg6, reg3, reg7, res4, res5, res6, res7);
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg0, reg1, reg2, reg3);
|
|
src0 = __lsx_vld(s, 0);
|
|
src1 = __lsx_vldx(s, src_stride);
|
|
src2 = __lsx_vldx(s, src_stride2);
|
|
src3 = __lsx_vldx(s, src_stride3);
|
|
s += src_stride4;
|
|
ILVLH_B(src0, src1, src2, src3, tmp0, tmp1, tmp2, tmp3);
|
|
ILVLH_H(tmp0, tmp2, tmp1, tmp3, reg4, reg5, reg6, reg7);
|
|
res8 = __lsx_vilvl_w(reg4, reg0);
|
|
res9 = __lsx_vilvh_w(reg4, reg0);
|
|
ILVLH_D(res0, res8, res1, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_2(dst0, dst2, dst_a, dst_stride_a, dst_stride_a2);
|
|
LSX_ST_2(dst1, dst3, dst_b, dst_stride_b, dst_stride_b2);
|
|
res8 = __lsx_vilvl_w(reg5, reg1);
|
|
res9 = __lsx_vilvh_w(reg5, reg1);
|
|
ILVLH_D(res2, res8, res3, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_2(dst0, dst2, dst_a, dst_stride_a, dst_stride_a2);
|
|
LSX_ST_2(dst1, dst3, dst_b, dst_stride_b, dst_stride_b2);
|
|
res8 = __lsx_vilvl_w(reg6, reg2);
|
|
res9 = __lsx_vilvh_w(reg6, reg2);
|
|
ILVLH_D(res4, res8, res5, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_2(dst0, dst2, dst_a, dst_stride_a, dst_stride_a2);
|
|
LSX_ST_2(dst1, dst3, dst_b, dst_stride_b, dst_stride_b2);
|
|
res8 = __lsx_vilvl_w(reg7, reg3);
|
|
res9 = __lsx_vilvh_w(reg7, reg3);
|
|
ILVLH_D(res6, res8, res7, res9, dst0, dst1, dst2, dst3);
|
|
LSX_ST_2(dst0, dst2, dst_a, dst_stride_a, dst_stride_a2);
|
|
LSX_ST_2(dst1, dst3, dst_b, dst_stride_b, dst_stride_b2);
|
|
src += 16;
|
|
}
|
|
}
|
|
|
|
#ifdef __cplusplus
|
|
} // extern "C"
|
|
} // namespace libyuv
|
|
#endif
|
|
|
|
#endif // !defined(LIBYUV_DISABLE_LSX) && defined(__loongarch_sx)
|