coffee/libyuv
1
0
Fork 0
mirror of https://chromium.googlesource.com/libyuv/libyuv synced 2026-01-01 03:12:16 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add optimization functions in row_lasx.cc file.

Browse Source 
Optimize 32 functions in source/row_lasx.cc file.
All test cases passed on loongarch platform.

Bug: libyuv:912
Signed-off-by: Hao Chen <chenhao@loongson.cn>
Change-Id: I7d3f649f753f72ca9bd052d5e0562dbc6f6ccfed
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/3351466
Reviewed-by: Frank Barchard <fbarchard@chromium.org>
Commit-Queue: Frank Barchard <fbarchard@chromium.org>
This commit is contained in:
Hao Chen 2021-12-20 19:57:26 +08:00 committed by Frank Barchard
parent 51de1e16f2
commit de8ae8c679
9 changed files with 1720 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

243
include/libyuv/row.h
View File

@ -689,6 +689,38 @@ extern "C" {
#define HAS_I422TOARGBROW_LASX
#define HAS_I422TORGBAROW_LASX
#define HAS_I422ALPHATOARGBROW_LASX
#define HAS_I422TOYUY2ROW_LASX
#define HAS_I422TOUYVYROW_LASX
#define HAS_MIRRORROW_LASX
#define HAS_MIRRORUVROW_LASX
#define HAS_ARGBMIRRORROW_LASX
#define HAS_I422TORGB24ROW_LASX
#define HAS_I422TORGB565ROW_LASX
#define HAS_I422TOARGB4444ROW_LASX
#define HAS_I422TOARGB1555ROW_LASX
#define HAS_YUY2TOUVROW_LASX
#define HAS_YUY2TOYROW_LASX
#define HAS_YUY2TOUV422ROW_LASX
#define HAS_UYVYTOYROW_LASX
#define HAS_UYVYTOUVROW_LASX
#define HAS_UYVYTOUV422ROW_LASX
#define HAS_ARGBTOYROW_LASX
#define HAS_ARGBTOUVROW_LASX
#define HAS_ARGBTORGB24ROW_LASX
#define HAS_ARGBTORAWROW_LASX
#define HAS_ARGBTORGB565ROW_LASX
#define HAS_ARGBTOARGB1555ROW_LASX
#define HAS_ARGBTOARGB4444ROW_LASX
#define HAS_ARGBTOUV444ROW_LASX
#define HAS_ARGBMULTIPLYROW_LASX
#define HAS_ARGBADDROW_LASX
#define HAS_ARGBSUBTRACTROW_LASX
#define HAS_ARGBATTENUATEROW_LASX
#define HAS_ARGBTORGB565DITHERROW_LASX
#define HAS_ARGBSHUFFLEROW_LASX
#define HAS_ARGBSHADEROW_LASX
#define HAS_ARGBGRAYROW_LASX
#define HAS_ARGBSEPIAROW_LASX
#endif
#if defined(_MSC_VER) && !defined(__CLR_VER) && !defined(__clang__)
@ -1005,24 +1037,48 @@ void I422ToRGB24Row_MSA(const uint8_t* src_y,
uint8_t* dst_argb,
const struct YuvConstants* yuvconstants,
int width);
void I422ToRGB24Row_LASX(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_argb,
const struct YuvConstants* yuvconstants,
int width);
void I422ToRGB565Row_MSA(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_rgb565,
const struct YuvConstants* yuvconstants,
int width);
void I422ToRGB565Row_LASX(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_rgb565,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB4444Row_MSA(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_argb4444,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB4444Row_LASX(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_argb4444,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB1555Row_MSA(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_argb1555,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB1555Row_LASX(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_argb1555,
const struct YuvConstants* yuvconstants,
int width);
void NV12ToARGBRow_MSA(const uint8_t* src_y,
const uint8_t* src_uv,
uint8_t* dst_argb,
@ -1074,6 +1130,7 @@ void ARGBToYRow_MSA(const uint8_t* src_argb0, uint8_t* dst_y, int width);
void ARGBToYJRow_MSA(const uint8_t* src_argb0, uint8_t* dst_y, int width);
void ARGBToYRow_MMI(const uint8_t* src_argb0, uint8_t* dst_y, int width);
void ARGBToYJRow_MMI(const uint8_t* src_argb0, uint8_t* dst_y, int width);
void ARGBToYRow_LASX(const uint8_t* src_argb0, uint8_t* dst_y, int width);
void ARGBToUV444Row_NEON(const uint8_t* src_argb,
uint8_t* dst_u,
uint8_t* dst_v,
@ -1092,6 +1149,15 @@ void ARGBToUVRow_MSA(const uint8_t* src_argb,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void ARGBToUVRow_LASX(const uint8_t* src_argb,
int src_stride_argb,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void ARGBToUV444Row_LASX(const uint8_t* src_argb,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void ARGBToUV444Row_MMI(const uint8_t* src_argb,
uint8_t* dst_u,
uint8_t* dst_v,
@ -1341,6 +1407,8 @@ void ARGB4444ToYRow_Any_MMI(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
void ARGBToYRow_Any_LASX(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void ARGBToUVRow_AVX2(const uint8_t* src_argb,
int src_stride_argb,
uint8_t* dst_u,
@ -1439,6 +1507,15 @@ void ARGBToUVRow_Any_MSA(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void ARGBToUVRow_Any_LASX(const uint8_t* src_ptr,
int src_stride_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void ARGBToUV444Row_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void ARGBToUV444Row_Any_MMI(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
@ -1678,6 +1755,7 @@ void MirrorRow_SSSE3(const uint8_t* src, uint8_t* dst, int width);
void MirrorRow_NEON(const uint8_t* src, uint8_t* dst, int width);
void MirrorRow_MSA(const uint8_t* src, uint8_t* dst, int width);
void MirrorRow_MMI(const uint8_t* src, uint8_t* dst, int width);
void MirrorRow_LASX(const uint8_t* src, uint8_t* dst, int width);
void MirrorRow_C(const uint8_t* src, uint8_t* dst, int width);
void MirrorRow_Any_AVX2(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorRow_Any_SSSE3(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
@ -1685,15 +1763,18 @@ void MirrorRow_Any_SSE2(const uint8_t* src, uint8_t* dst, int width);
void MirrorRow_Any_NEON(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorRow_Any_MSA(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorRow_Any_MMI(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorRow_Any_LASX(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorUVRow_AVX2(const uint8_t* src_uv, uint8_t* dst_uv, int width);
void MirrorUVRow_SSSE3(const uint8_t* src_uv, uint8_t* dst_uv, int width);
void MirrorUVRow_NEON(const uint8_t* src_uv, uint8_t* dst_uv, int width);
void MirrorUVRow_MSA(const uint8_t* src_uv, uint8_t* dst_uv, int width);
void MirrorUVRow_LASX(const uint8_t* src_uv, uint8_t* dst_uv, int width);
void MirrorUVRow_C(const uint8_t* src_uv, uint8_t* dst_uv, int width);
void MirrorUVRow_Any_AVX2(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorUVRow_Any_SSSE3(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorUVRow_Any_NEON(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorUVRow_Any_MSA(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorUVRow_Any_LASX(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void MirrorSplitUVRow_SSSE3(const uint8_t* src,
uint8_t* dst_u,
@ -1721,6 +1802,7 @@ void ARGBMirrorRow_SSE2(const uint8_t* src, uint8_t* dst, int width);
void ARGBMirrorRow_NEON(const uint8_t* src_argb, uint8_t* dst_argb, int width);
void ARGBMirrorRow_MSA(const uint8_t* src, uint8_t* dst, int width);
void ARGBMirrorRow_MMI(const uint8_t* src, uint8_t* dst, int width);
void ARGBMirrorRow_LASX(const uint8_t* src, uint8_t* dst, int width);
void ARGBMirrorRow_C(const uint8_t* src, uint8_t* dst, int width);
void ARGBMirrorRow_Any_AVX2(const uint8_t* src_ptr,
uint8_t* dst_ptr,
@ -1733,6 +1815,7 @@ void ARGBMirrorRow_Any_NEON(const uint8_t* src_ptr,
int width);
void ARGBMirrorRow_Any_MSA(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void ARGBMirrorRow_Any_MMI(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void ARGBMirrorRow_Any_LASX(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void RGB24MirrorRow_SSSE3(const uint8_t* src_rgb24,
uint8_t* dst_rgb24,
@ -2537,6 +2620,10 @@ void ARGBShuffleRow_MMI(const uint8_t* src_argb,
uint8_t* dst_argb,
const uint8_t* shuffler,
int width);
void ARGBShuffleRow_LASX(const uint8_t* src_argb,
uint8_t* dst_argb,
const uint8_t* shuffler,
int width);
void ARGBShuffleRow_Any_SSSE3(const uint8_t* src_ptr,
uint8_t* dst_ptr,
const uint8_t* param,
@ -2557,6 +2644,10 @@ void ARGBShuffleRow_Any_MMI(const uint8_t* src_ptr,
uint8_t* dst_ptr,
const uint8_t* param,
int width);
void ARGBShuffleRow_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_ptr,
const uint8_t* param,
int width);
void RGB24ToARGBRow_SSSE3(const uint8_t* src_rgb24,
uint8_t* dst_argb,
@ -2777,6 +2868,20 @@ void ARGBToRGB565DitherRow_MSA(const uint8_t* src_argb,
uint8_t* dst_rgb,
const uint32_t dither4,
int width);
void ARGBToRGB565DitherRow_LASX(const uint8_t* src_argb,
uint8_t* dst_rgb,
const uint32_t dither4,
int width);
void ARGBToRGB24Row_LASX(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
void ARGBToRAWRow_LASX(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
void ARGBToRGB565Row_LASX(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
void ARGBToARGB1555Row_LASX(const uint8_t* src_argb,
uint8_t* dst_rgb,
int width);
void ARGBToARGB4444Row_LASX(const uint8_t* src_argb,
uint8_t* dst_rgb,
int width);
void ARGBToRGB24Row_MMI(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
void ARGBToRAWRow_MMI(const uint8_t* src_argb, uint8_t* dst_rgb, int width);
@ -3896,6 +4001,14 @@ void ARGBMultiplyRow_Any_MMI(const uint8_t* y_buf,
const uint8_t* uv_buf,
uint8_t* dst_ptr,
int width);
void ARGBMultiplyRow_LASX(const uint8_t* src_argb0,
const uint8_t* src_argb1,
uint8_t* dst_argb,
int width);
void ARGBMultiplyRow_Any_LASX(const uint8_t* y_buf,
const uint8_t* uv_buf,
uint8_t* dst_ptr,
int width);
// ARGB add images.
void ARGBAddRow_C(const uint8_t* src_argb,
@ -3942,6 +4055,14 @@ void ARGBAddRow_Any_MMI(const uint8_t* y_buf,
const uint8_t* uv_buf,
uint8_t* dst_ptr,
int width);
void ARGBAddRow_LASX(const uint8_t* src_argb0,
const uint8_t* src_argb1,
uint8_t* dst_argb,
int width);
void ARGBAddRow_Any_LASX(const uint8_t* y_buf,
const uint8_t* uv_buf,
uint8_t* dst_ptr,
int width);
// ARGB subtract images. Same API as Blend, but these require
// pointer and width alignment for SSE2.
@ -3989,6 +4110,14 @@ void ARGBSubtractRow_Any_MMI(const uint8_t* y_buf,
const uint8_t* uv_buf,
uint8_t* dst_ptr,
int width);
void ARGBSubtractRow_LASX(const uint8_t* src_argb0,
const uint8_t* src_argb1,
uint8_t* dst_argb,
int width);
void ARGBSubtractRow_Any_LASX(const uint8_t* y_buf,
const uint8_t* uv_buf,
uint8_t* dst_ptr,
int width);
void ARGBToRGB24Row_Any_SSSE3(const uint8_t* src_ptr,
uint8_t* dst_ptr,
@ -4077,6 +4206,24 @@ void ARGBToRGB565DitherRow_Any_MSA(const uint8_t* src_ptr,
uint8_t* dst_ptr,
const uint32_t param,
int width);
void ARGBToRGB565DitherRow_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_ptr,
const uint32_t param,
int width);
void ARGBToRGB24Row_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
void ARGBToRAWRow_Any_LASX(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void ARGBToRGB565Row_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
void ARGBToARGB1555Row_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
void ARGBToARGB4444Row_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
void ARGBToRGB24Row_Any_MMI(const uint8_t* src_ptr,
uint8_t* dst_ptr,
@ -4291,24 +4438,48 @@ void I422ToRGB24Row_Any_MSA(const uint8_t* y_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void I422ToRGB24Row_Any_LASX(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void I422ToRGB565Row_Any_MSA(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void I422ToRGB565Row_Any_LASX(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB4444Row_Any_MSA(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB4444Row_Any_LASX(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB1555Row_Any_MSA(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void I422ToARGB1555Row_Any_LASX(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
const struct YuvConstants* yuvconstants,
int width);
void NV12ToARGBRow_Any_MSA(const uint8_t* y_buf,
const uint8_t* uv_buf,
uint8_t* dst_ptr,
@ -4365,11 +4536,17 @@ void YUY2ToUV422Row_NEON(const uint8_t* src_yuy2,
int width);
void YUY2ToYRow_MSA(const uint8_t* src_yuy2, uint8_t* dst_y, int width);
void YUY2ToYRow_MMI(const uint8_t* src_yuy2, uint8_t* dst_y, int width);
void YUY2ToYRow_LASX(const uint8_t* src_yuy2, uint8_t* dst_y, int width);
void YUY2ToUVRow_MSA(const uint8_t* src_yuy2,
int src_stride_yuy2,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUVRow_LASX(const uint8_t* src_yuy2,
int src_stride_yuy2,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUVRow_MMI(const uint8_t* src_yuy2,
int src_stride_yuy2,
uint8_t* dst_u,
@ -4379,6 +4556,10 @@ void YUY2ToUV422Row_MSA(const uint8_t* src_yuy2,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUV422Row_LASX(const uint8_t* src_yuy2,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUV422Row_MMI(const uint8_t* src_yuy2,
uint8_t* dst_u,
uint8_t* dst_v,
@ -4425,11 +4606,17 @@ void YUY2ToUV422Row_Any_NEON(const uint8_t* src_ptr,
int width);
void YUY2ToYRow_Any_MSA(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void YUY2ToYRow_Any_MMI(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void YUY2ToYRow_Any_LASX(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void YUY2ToUVRow_Any_MSA(const uint8_t* src_ptr,
int src_stride_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUVRow_Any_LASX(const uint8_t* src_ptr,
int src_stride_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUVRow_Any_MMI(const uint8_t* src_ptr,
int src_stride_ptr,
uint8_t* dst_u,
@ -4439,6 +4626,10 @@ void YUY2ToUV422Row_Any_MSA(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUV422Row_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void YUY2ToUV422Row_Any_MMI(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
@ -4485,11 +4676,17 @@ void UYVYToUV422Row_NEON(const uint8_t* src_uyvy,
int width);
void UYVYToYRow_MSA(const uint8_t* src_uyvy, uint8_t* dst_y, int width);
void UYVYToYRow_MMI(const uint8_t* src_uyvy, uint8_t* dst_y, int width);
void UYVYToYRow_LASX(const uint8_t* src_uyvy, uint8_t* dst_y, int width);
void UYVYToUVRow_MSA(const uint8_t* src_uyvy,
int src_stride_uyvy,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUVRow_LASX(const uint8_t* src_uyvy,
int src_stride_uyvy,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUVRow_MMI(const uint8_t* src_uyvy,
int src_stride_uyvy,
uint8_t* dst_u,
@ -4499,6 +4696,10 @@ void UYVYToUV422Row_MSA(const uint8_t* src_uyvy,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUV422Row_LASX(const uint8_t* src_uyvy,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUV422Row_MMI(const uint8_t* src_uyvy,
uint8_t* dst_u,
uint8_t* dst_v,
@ -4546,11 +4747,17 @@ void UYVYToUV422Row_Any_NEON(const uint8_t* src_ptr,
int width);
void UYVYToYRow_Any_MSA(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void UYVYToYRow_Any_MMI(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void UYVYToYRow_Any_LASX(const uint8_t* src_ptr, uint8_t* dst_ptr, int width);
void UYVYToUVRow_Any_MSA(const uint8_t* src_ptr,
int src_stride_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUVRow_Any_LASX(const uint8_t* src_ptr,
int src_stride_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUVRow_Any_MMI(const uint8_t* src_ptr,
int src_stride_ptr,
uint8_t* dst_u,
@ -4560,6 +4767,10 @@ void UYVYToUV422Row_Any_MSA(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUV422Row_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
int width);
void UYVYToUV422Row_Any_MMI(const uint8_t* src_ptr,
uint8_t* dst_u,
uint8_t* dst_v,
@ -4679,6 +4890,11 @@ void I422ToYUY2Row_MMI(const uint8_t* src_y,
const uint8_t* src_v,
uint8_t* dst_yuy2,
int width);
void I422ToYUY2Row_LASX(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_yuy2,
int width);
void I422ToUYVYRow_MSA(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
@ -4689,6 +4905,11 @@ void I422ToUYVYRow_MMI(const uint8_t* src_y,
const uint8_t* src_v,
uint8_t* dst_uyvy,
int width);
void I422ToUYVYRow_LASX(const uint8_t* src_y,
const uint8_t* src_u,
const uint8_t* src_v,
uint8_t* dst_uyvy,
int width);
void I422ToYUY2Row_Any_MSA(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
@ -4699,6 +4920,11 @@ void I422ToYUY2Row_Any_MMI(const uint8_t* y_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
int width);
void I422ToYUY2Row_Any_LASX(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
int width);
void I422ToUYVYRow_Any_MSA(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
@ -4709,6 +4935,11 @@ void I422ToUYVYRow_Any_MMI(const uint8_t* y_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
int width);
void I422ToUYVYRow_Any_LASX(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* dst_ptr,
int width);
// Effects related row functions.
void ARGBAttenuateRow_C(const uint8_t* src_argb, uint8_t* dst_argb, int width);
@ -4727,6 +4958,9 @@ void ARGBAttenuateRow_MSA(const uint8_t* src_argb,
void ARGBAttenuateRow_MMI(const uint8_t* src_argb,
uint8_t* dst_argb,
int width);
void ARGBAttenuateRow_LASX(const uint8_t* src_argb,
uint8_t* dst_argb,
int width);
void ARGBAttenuateRow_Any_SSSE3(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
@ -4742,6 +4976,9 @@ void ARGBAttenuateRow_Any_MSA(const uint8_t* src_ptr,
void ARGBAttenuateRow_Any_MMI(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
void ARGBAttenuateRow_Any_LASX(const uint8_t* src_ptr,
uint8_t* dst_ptr,
int width);
// Inverse table for unattenuate, shared by C and SSE2.
extern const uint32_t fixed_invtbl8[256];
@ -4766,12 +5003,14 @@ void ARGBGrayRow_SSSE3(const uint8_t* src_argb, uint8_t* dst_argb, int width);
void ARGBGrayRow_NEON(const uint8_t* src_argb, uint8_t* dst_argb, int width);
void ARGBGrayRow_MSA(const uint8_t* src_argb, uint8_t* dst_argb, int width);
void ARGBGrayRow_MMI(const uint8_t* src_argb, uint8_t* dst_argb, int width);
void ARGBGrayRow_LASX(const uint8_t* src_argb, uint8_t* dst_argb, int width);
void ARGBSepiaRow_C(uint8_t* dst_argb, int width);
void ARGBSepiaRow_SSSE3(uint8_t* dst_argb, int width);
void ARGBSepiaRow_NEON(uint8_t* dst_argb, int width);
void ARGBSepiaRow_MSA(uint8_t* dst_argb, int width);
void ARGBSepiaRow_MMI(uint8_t* dst_argb, int width);
void ARGBSepiaRow_LASX(uint8_t* dst_argb, int width);
void ARGBColorMatrixRow_C(const uint8_t* src_argb,
uint8_t* dst_argb,
@ -4849,6 +5088,10 @@ void ARGBShadeRow_MMI(const uint8_t* src_argb,
uint8_t* dst_argb,
int width,
uint32_t value);
void ARGBShadeRow_LASX(const uint8_t* src_argb,
uint8_t* dst_argb,
int width,
uint32_t value);
// Used for blur.
void CumulativeSumToAverageRow_SSE2(const int32_t* topleft,

40
source/convert.cc
View File

@ -1095,6 +1095,16 @@ int YUY2ToI420(const uint8_t* src_yuy2,
}
}
#endif
#if defined(HAS_YUY2TOYROW_LASX) && defined(HAS_YUY2TOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
YUY2ToYRow = YUY2ToYRow_Any_LASX;
YUY2ToUVRow = YUY2ToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
YUY2ToYRow = YUY2ToYRow_LASX;
YUY2ToUVRow = YUY2ToUVRow_LASX;
}
}
#endif
for (y = 0; y < height - 1; y += 2) {
YUY2ToUVRow(src_yuy2, src_stride_yuy2, dst_u, dst_v, width);
@ -1186,6 +1196,16 @@ int UYVYToI420(const uint8_t* src_uyvy,
}
}
#endif
#if defined(HAS_UYVYTOYROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
UYVYToYRow = UYVYToYRow_Any_LASX;
UYVYToUVRow = UYVYToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
UYVYToYRow = UYVYToYRow_LASX;
UYVYToUVRow = UYVYToUVRow_LASX;
}
}
#endif
for (y = 0; y < height - 1; y += 2) {
UYVYToUVRow(src_uyvy, src_stride_uyvy, dst_u, dst_v, width);
@ -1440,6 +1460,16 @@ int ARGBToI420(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX) && defined(HAS_ARGBTOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
ARGBToUVRow = ARGBToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
ARGBToUVRow = ARGBToUVRow_LASX;
}
}
#endif
for (y = 0; y < height - 1; y += 2) {
ARGBToUVRow(src_argb, src_stride_argb, dst_u, dst_v, width);
@ -2924,6 +2954,16 @@ int ARGB4444ToI420(const uint8_t* src_argb4444,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX) && defined(HAS_ARGBTOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
ARGBToUVRow = ARGBToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
ARGBToUVRow = ARGBToUVRow_LASX;
}
}
#endif
#endif
{

48
source/convert_argb.cc
View File

@ -4668,6 +4668,14 @@ int I420ToRGB24Matrix(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TORGB24ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToRGB24Row = I422ToRGB24Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToRGB24Row = I422ToRGB24Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
I422ToRGB24Row(src_y, src_u, src_v, dst_rgb24, yuvconstants, width);
@ -4856,6 +4864,14 @@ int I420ToARGB1555(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TOARGB1555ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToARGB1555Row = I422ToARGB1555Row_Any_LASX;
if (IS_ALIGNED(width, 8)) {
I422ToARGB1555Row = I422ToARGB1555Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
I422ToARGB1555Row(src_y, src_u, src_v, dst_argb1555, &kYuvI601Constants,
@ -4937,6 +4953,14 @@ int I420ToARGB4444(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TOARGB4444ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToARGB4444Row = I422ToARGB4444Row_Any_LASX;
if (IS_ALIGNED(width, 8)) {
I422ToARGB4444Row = I422ToARGB4444Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
I422ToARGB4444Row(src_y, src_u, src_v, dst_argb4444, &kYuvI601Constants,
@ -5018,6 +5042,14 @@ int I420ToRGB565Matrix(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TORGB565ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToRGB565Row = I422ToRGB565Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToRGB565Row = I422ToRGB565Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
I422ToRGB565Row(src_y, src_u, src_v, dst_rgb565, yuvconstants, width);
@ -5140,6 +5172,14 @@ int I422ToRGB565(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TORGB565ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToRGB565Row = I422ToRGB565Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToRGB565Row = I422ToRGB565Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
I422ToRGB565Row(src_y, src_u, src_v, dst_rgb565, &kYuvI601Constants, width);
@ -5284,6 +5324,14 @@ int I420ToRGB565Dither(const uint8_t* src_y,
ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_MSA;
}
}
#endif
#if defined(HAS_ARGBTORGB565DITHERROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LASX;
}
}
#endif
{
// Allocate a row of argb.

24
source/convert_from.cc
View File

@ -452,6 +452,14 @@ int I420ToYUY2(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TOYUY2ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToYUY2Row = I422ToYUY2Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToYUY2Row = I422ToYUY2Row_LASX;
}
}
#endif
for (y = 0; y < height - 1; y += 2) {
I422ToYUY2Row(src_y, src_u, src_v, dst_yuy2, width);
@ -539,6 +547,14 @@ int I422ToUYVY(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TOUYVYROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToUYVYRow = I422ToUYVYRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToUYVYRow = I422ToUYVYRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
I422ToUYVYRow(src_y, src_u, src_v, dst_uyvy, width);
@ -614,6 +630,14 @@ int I420ToUYVY(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_I422TOUYVYROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToUYVYRow = I422ToUYVYRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToUYVYRow = I422ToUYVYRow_LASX;
}
}
#endif
for (y = 0; y < height - 1; y += 2) {
I422ToUYVYRow(src_y, src_u, src_v, dst_uyvy, width);

139
source/convert_from_argb.cc
View File

@ -84,6 +84,14 @@ int ARGBToI444(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOUV444ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToUV444Row = ARGBToUV444Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToUV444Row = ARGBToUV444Row_LASX;
}
}
#endif
#if defined(HAS_ARGBTOYROW_SSSE3)
if (TestCpuFlag(kCpuHasSSSE3)) {
ARGBToYRow = ARGBToYRow_Any_SSSE3;
@ -124,6 +132,14 @@ int ARGBToI444(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToUV444Row(src_argb, dst_u, dst_v, width);
@ -245,6 +261,17 @@ int ARGBToI422(const uint8_t* src_argb,
}
#endif
#if defined(HAS_ARGBTOYROW_LASX) && defined(HAS_ARGBTOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
ARGBToUVRow = ARGBToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
ARGBToUVRow = ARGBToUVRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToUVRow(src_argb, 0, dst_u, dst_v, width);
ARGBToYRow(src_argb, dst_y, width);
@ -355,6 +382,16 @@ int ARGBToNV12(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX) && defined(HAS_ARGBTOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
ARGBToUVRow = ARGBToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
ARGBToUVRow = ARGBToUVRow_LASX;
}
}
#endif
#if defined(HAS_MERGEUVROW_SSE2)
if (TestCpuFlag(kCpuHasSSE2)) {
MergeUVRow_ = MergeUVRow_Any_SSE2;
@ -519,6 +556,16 @@ int ARGBToNV21(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX) && defined(HAS_ARGBTOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
ARGBToUVRow = ARGBToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
ARGBToUVRow = ARGBToUVRow_LASX;
}
}
#endif
#if defined(HAS_MERGEUVROW_SSE2)
if (TestCpuFlag(kCpuHasSSE2)) {
MergeUVRow_ = MergeUVRow_Any_SSE2;
@ -1015,6 +1062,16 @@ int ARGBToYUY2(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX) && defined(HAS_ARGBTOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
ARGBToUVRow = ARGBToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
ARGBToUVRow = ARGBToUVRow_LASX;
}
}
#endif
#if defined(HAS_I422TOYUY2ROW_SSE2)
if (TestCpuFlag(kCpuHasSSE2)) {
I422ToYUY2Row = I422ToYUY2Row_Any_SSE2;
@ -1055,6 +1112,14 @@ int ARGBToYUY2(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_I422TOYUY2ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToYUY2Row = I422ToYUY2Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToYUY2Row = I422ToYUY2Row_LASX;
}
}
#endif
{
// Allocate a rows of yuv.
@ -1180,6 +1245,16 @@ int ARGBToUYVY(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX) && defined(HAS_ARGBTOUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
ARGBToUVRow = ARGBToUVRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
ARGBToUVRow = ARGBToUVRow_LASX;
}
}
#endif
#if defined(HAS_I422TOUYVYROW_SSE2)
if (TestCpuFlag(kCpuHasSSE2)) {
I422ToUYVYRow = I422ToUYVYRow_Any_SSE2;
@ -1220,6 +1295,14 @@ int ARGBToUYVY(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_I422TOUYVYROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
I422ToUYVYRow = I422ToUYVYRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
I422ToUYVYRow = I422ToUYVYRow_LASX;
}
}
#endif
{
// Allocate a rows of yuv.
@ -1305,6 +1388,14 @@ int ARGBToI400(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOYROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToYRow = ARGBToYRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToYRow = ARGBToYRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToYRow(src_argb, dst_y, width);
@ -1403,6 +1494,14 @@ int ARGBToRGB24(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTORGB24ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToRGB24Row = ARGBToRGB24Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToRGB24Row = ARGBToRGB24Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToRGB24Row(src_argb, dst_rgb24, width);
@ -1477,6 +1576,14 @@ int ARGBToRAW(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTORAWROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToRAWRow = ARGBToRAWRow_Any_LASX;
if (IS_ALIGNED(width, 32)) {
ARGBToRAWRow = ARGBToRAWRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToRAWRow(src_argb, dst_raw, width);
@ -1555,6 +1662,14 @@ int ARGBToRGB565Dither(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTORGB565DITHERROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToRGB565DitherRow(src_argb, dst_rgb565,
@ -1632,6 +1747,14 @@ int ARGBToRGB565(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTORGB565ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToRGB565Row = ARGBToRGB565Row_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBToRGB565Row = ARGBToRGB565Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToRGB565Row(src_argb, dst_rgb565, width);
@ -1706,6 +1829,14 @@ int ARGBToARGB1555(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOARGB1555ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToARGB1555Row = ARGBToARGB1555Row_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBToARGB1555Row = ARGBToARGB1555Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToARGB1555Row(src_argb, dst_argb1555, width);
@ -1780,6 +1911,14 @@ int ARGBToARGB4444(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBTOARGB4444ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBToARGB4444Row = ARGBToARGB4444Row_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBToARGB4444Row = ARGBToARGB4444Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBToARGB4444Row(src_argb, dst_argb4444, width);

104
source/planar_functions.cc
View File

@ -1728,6 +1728,16 @@ int YUY2ToI422(const uint8_t* src_yuy2,
}
}
#endif
#if defined(HAS_YUY2TOYROW_LASX) && defined(HAS_YUY2TOUV422ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
YUY2ToYRow = YUY2ToYRow_Any_LASX;
YUY2ToUV422Row = YUY2ToUV422Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
YUY2ToYRow = YUY2ToYRow_LASX;
YUY2ToUV422Row = YUY2ToUV422Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
YUY2ToUV422Row(src_yuy2, dst_u, dst_v, width);
@ -1824,6 +1834,16 @@ int UYVYToI422(const uint8_t* src_uyvy,
}
}
#endif
#if defined(HAS_UYVYTOYROW_LASX) && defined(HAS_UYVYTOUV422ROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
UYVYToYRow = UYVYToYRow_Any_LASX;
UYVYToUV422Row = UYVYToUV422Row_Any_LASX;
if (IS_ALIGNED(width, 32)) {
UYVYToYRow = UYVYToYRow_LASX;
UYVYToUV422Row = UYVYToUV422Row_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
UYVYToUV422Row(src_uyvy, dst_u, dst_v, width);
@ -1968,6 +1988,14 @@ void MirrorPlane(const uint8_t* src_y,
}
}
#endif
#if defined(HAS_MIRRORROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
MirrorRow = MirrorRow_Any_LASX;
if (IS_ALIGNED(width, 64)) {
MirrorRow = MirrorRow_LASX;
}
}
#endif
// Mirror plane
for (y = 0; y < height; ++y) {
@ -2026,6 +2054,14 @@ void MirrorUVPlane(const uint8_t* src_uv,
}
}
#endif
#if defined(HAS_MIRRORUVROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
MirrorUVRow = MirrorUVRow_Any_LASX;
if (IS_ALIGNED(width, 16)) {
MirrorUVRow = MirrorUVRow_LASX;
}
}
#endif
// MirrorUV plane
for (y = 0; y < height; ++y) {
@ -2194,6 +2230,14 @@ int ARGBMirror(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBMIRRORROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBMirrorRow = ARGBMirrorRow_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBMirrorRow = ARGBMirrorRow_LASX;
}
}
#endif
// Mirror plane
for (y = 0; y < height; ++y) {
@ -2602,6 +2646,14 @@ int ARGBMultiply(const uint8_t* src_argb0,
}
}
#endif
#if defined(HAS_ARGBMULTIPLYROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBMultiplyRow = ARGBMultiplyRow_Any_LASX;
if (IS_ALIGNED(width, 8)) {
ARGBMultiplyRow = ARGBMultiplyRow_LASX;
}
}
#endif
// Multiply plane
for (y = 0; y < height; ++y) {
@ -2687,6 +2739,14 @@ int ARGBAdd(const uint8_t* src_argb0,
}
}
#endif
#if defined(HAS_ARGBADDROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBAddRow = ARGBAddRow_Any_LASX;
if (IS_ALIGNED(width, 8)) {
ARGBAddRow = ARGBAddRow_LASX;
}
}
#endif
// Add plane
for (y = 0; y < height; ++y) {
@ -2767,6 +2827,14 @@ int ARGBSubtract(const uint8_t* src_argb0,
}
}
#endif
#if defined(HAS_ARGBSUBTRACTROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBSubtractRow = ARGBSubtractRow_Any_LASX;
if (IS_ALIGNED(width, 8)) {
ARGBSubtractRow = ARGBSubtractRow_LASX;
}
}
#endif
// Subtract plane
for (y = 0; y < height; ++y) {
@ -3073,6 +3141,14 @@ int ARGBAttenuate(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBATTENUATEROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBAttenuateRow = ARGBAttenuateRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBAttenuateRow(src_argb, dst_argb, width);
@ -3178,6 +3254,11 @@ int ARGBGrayTo(const uint8_t* src_argb,
ARGBGrayRow = ARGBGrayRow_MSA;
}
#endif
#if defined(HAS_ARGBGRAYROW_LASX)
if (TestCpuFlag(kCpuHasLASX) && IS_ALIGNED(width, 16)) {
ARGBGrayRow = ARGBGrayRow_LASX;
}
#endif
for (y = 0; y < height; ++y) {
ARGBGrayRow(src_argb, dst_argb, width);
@ -3228,6 +3309,11 @@ int ARGBGray(uint8_t* dst_argb,
ARGBGrayRow = ARGBGrayRow_MSA;
}
#endif
#if defined(HAS_ARGBGRAYROW_LASX)
if (TestCpuFlag(kCpuHasLASX) && IS_ALIGNED(width, 16)) {
ARGBGrayRow = ARGBGrayRow_LASX;
}
#endif
for (y = 0; y < height; ++y) {
ARGBGrayRow(dst, dst, width);
@ -3276,6 +3362,11 @@ int ARGBSepia(uint8_t* dst_argb,
ARGBSepiaRow = ARGBSepiaRow_MSA;
}
#endif
#if defined(HAS_ARGBSEPIAROW_LASX)
if (TestCpuFlag(kCpuHasLASX) && IS_ALIGNED(width, 16)) {
ARGBSepiaRow = ARGBSepiaRow_LASX;
}
#endif
for (y = 0; y < height; ++y) {
ARGBSepiaRow(dst, width);
@ -3706,6 +3797,11 @@ int ARGBShade(const uint8_t* src_argb,
ARGBShadeRow = ARGBShadeRow_MSA;
}
#endif
#if defined(HAS_ARGBSHADEROW_LASX)
if (TestCpuFlag(kCpuHasLASX) && IS_ALIGNED(width, 8)) {
ARGBShadeRow = ARGBShadeRow_LASX;
}
#endif
for (y = 0; y < height; ++y) {
ARGBShadeRow(src_argb, dst_argb, width, value);
@ -3916,6 +4012,14 @@ int ARGBShuffle(const uint8_t* src_bgra,
}
}
#endif
#if defined(HAS_ARGBSHUFFLEROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBShuffleRow = ARGBShuffleRow_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBShuffleRow = ARGBShuffleRow_LASX;
}
}
#endif
for (y = 0; y < height; ++y) {
ARGBShuffleRow(src_bgra, dst_argb, shuffler, width);

8
source/rotate_argb.cc
View File

@ -163,6 +163,14 @@ static int ARGBRotate180(const uint8_t* src_argb,
}
}
#endif
#if defined(HAS_ARGBMIRRORROW_LASX)
if (TestCpuFlag(kCpuHasLASX)) {
ARGBMirrorRow = ARGBMirrorRow_Any_LASX;
if (IS_ALIGNED(width, 16)) {
ARGBMirrorRow = ARGBMirrorRow_LASX;
}
}
#endif
#if defined(HAS_COPYROW_SSE2)
if (TestCpuFlag(kCpuHasSSE2)) {
CopyRow = IS_ALIGNED(width * 4, 32) ? CopyRow_SSE2 : CopyRow_Any_SSE2;

72
source/row_any.cc
View File

@ -297,6 +297,9 @@ ANY31(I422ToYUY2Row_Any_MSA, I422ToYUY2Row_MSA, 1, 1, 4, 31)
#ifdef HAS_I422TOYUY2ROW_MMI
ANY31(I422ToYUY2Row_Any_MMI, I422ToYUY2Row_MMI, 1, 1, 4, 7)
#endif
#ifdef HAS_I422TOYUY2ROW_LASX
ANY31(I422ToYUY2Row_Any_LASX, I422ToYUY2Row_LASX, 1, 1, 4, 31)
#endif
#ifdef HAS_I422TOUYVYROW_NEON
ANY31(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, 1, 1, 4, 15)
#endif
@ -306,6 +309,9 @@ ANY31(I422ToUYVYRow_Any_MSA, I422ToUYVYRow_MSA, 1, 1, 4, 31)
#ifdef HAS_I422TOUYVYROW_MMI
ANY31(I422ToUYVYRow_Any_MMI, I422ToUYVYRow_MMI, 1, 1, 4, 7)
#endif
#ifdef HAS_I422TOUYVYROW_LASX
ANY31(I422ToUYVYRow_Any_LASX, I422ToUYVYRow_LASX, 1, 1, 4, 31)
#endif
#ifdef HAS_BLENDPLANEROW_AVX2
ANY31(BlendPlaneRow_Any_AVX2, BlendPlaneRow_AVX2, 0, 0, 1, 31)
#endif
@ -425,6 +431,10 @@ ANY31C(I422ToRGBARow_Any_MMI, I422ToRGBARow_MMI, 1, 0, 4, 7)
#ifdef HAS_I422TOARGBROW_LASX
ANY31C(I422ToARGBRow_Any_LASX, I422ToARGBRow_LASX, 1, 0, 4, 31)
ANY31C(I422ToRGBARow_Any_LASX, I422ToRGBARow_LASX, 1, 0, 4, 31)
ANY31C(I422ToRGB24Row_Any_LASX, I422ToRGB24Row_LASX, 1, 0, 3, 31)
ANY31C(I422ToRGB565Row_Any_LASX, I422ToRGB565Row_LASX, 1, 0, 2, 31)
ANY31C(I422ToARGB4444Row_Any_LASX, I422ToARGB4444Row_LASX, 1, 0, 2, 31)
ANY31C(I422ToARGB1555Row_Any_LASX, I422ToARGB1555Row_LASX, 1, 0, 2, 31)
#endif
#undef ANY31C
@ -631,18 +641,27 @@ ANY21(ARGBMultiplyRow_Any_MSA, ARGBMultiplyRow_MSA, 0, 4, 4, 4, 3)
#ifdef HAS_ARGBMULTIPLYROW_MMI
ANY21(ARGBMultiplyRow_Any_MMI, ARGBMultiplyRow_MMI, 0, 4, 4, 4, 1)
#endif
#ifdef HAS_ARGBMULTIPLYROW_LASX
ANY21(ARGBMultiplyRow_Any_LASX, ARGBMultiplyRow_LASX, 0, 4, 4, 4, 7)
#endif
#ifdef HAS_ARGBADDROW_MSA
ANY21(ARGBAddRow_Any_MSA, ARGBAddRow_MSA, 0, 4, 4, 4, 7)
#endif
#ifdef HAS_ARGBADDROW_MMI
ANY21(ARGBAddRow_Any_MMI, ARGBAddRow_MMI, 0, 4, 4, 4, 1)
#endif
#ifdef HAS_ARGBADDROW_LASX
ANY21(ARGBAddRow_Any_LASX, ARGBAddRow_LASX, 0, 4, 4, 4, 7)
#endif
#ifdef HAS_ARGBSUBTRACTROW_MSA
ANY21(ARGBSubtractRow_Any_MSA, ARGBSubtractRow_MSA, 0, 4, 4, 4, 7)
#endif
#ifdef HAS_ARGBSUBTRACTROW_MMI
ANY21(ARGBSubtractRow_Any_MMI, ARGBSubtractRow_MMI, 0, 4, 4, 4, 1)
#endif
#ifdef HAS_ARGBSUBTRACTROW_LASX
ANY21(ARGBSubtractRow_Any_LASX, ARGBSubtractRow_LASX, 0, 4, 4, 4, 7)
#endif
#ifdef HAS_SOBELROW_SSE2
ANY21(SobelRow_Any_SSE2, SobelRow_SSE2, 0, 1, 1, 4, 15)
#endif
@ -953,6 +972,13 @@ ANY11(ARGBToARGB1555Row_Any_MMI, ARGBToARGB1555Row_MMI, 0, 4, 2, 3)
ANY11(ARGBToARGB4444Row_Any_MMI, ARGBToARGB4444Row_MMI, 0, 4, 2, 3)
ANY11(J400ToARGBRow_Any_MMI, J400ToARGBRow_MMI, 0, 1, 4, 3)
#endif
#if defined(HAS_ARGBTORGB24ROW_LASX)
ANY11(ARGBToRGB24Row_Any_LASX, ARGBToRGB24Row_LASX, 0, 4, 3, 31)
ANY11(ARGBToRAWRow_Any_LASX, ARGBToRAWRow_LASX, 0, 4, 3, 31)
ANY11(ARGBToRGB565Row_Any_LASX, ARGBToRGB565Row_LASX, 0, 4, 2, 15)
ANY11(ARGBToARGB1555Row_Any_LASX, ARGBToARGB1555Row_LASX, 0, 4, 2, 15)
ANY11(ARGBToARGB4444Row_Any_LASX, ARGBToARGB4444Row_LASX, 0, 4, 2, 15)
#endif
#if defined(HAS_RAWTORGB24ROW_NEON)
ANY11(RAWToRGB24Row_Any_NEON, RAWToRGB24Row_NEON, 0, 3, 3, 7)
#endif
@ -1007,6 +1033,9 @@ ANY11(ARGBToYRow_Any_MSA, ARGBToYRow_MSA, 0, 4, 1, 15)
#ifdef HAS_ARGBTOYROW_MMI
ANY11(ARGBToYRow_Any_MMI, ARGBToYRow_MMI, 0, 4, 1, 7)
#endif
#ifdef HAS_ARGBTOYROW_LASX
ANY11(ARGBToYRow_Any_LASX, ARGBToYRow_LASX, 0, 4, 1, 31)
#endif
#ifdef HAS_ARGBTOYJROW_NEON
ANY11(ARGBToYJRow_Any_NEON, ARGBToYJRow_NEON, 0, 4, 1, 7)
#endif
@ -1115,12 +1144,18 @@ ANY11(UYVYToYRow_Any_NEON, UYVYToYRow_NEON, 1, 4, 1, 15)
#ifdef HAS_YUY2TOYROW_MSA
ANY11(YUY2ToYRow_Any_MSA, YUY2ToYRow_MSA, 1, 4, 1, 31)
#endif
#ifdef HAS_YUY2TOYROW_LASX
ANY11(YUY2ToYRow_Any_LASX, YUY2ToYRow_LASX, 1, 4, 1, 31)
#endif
#ifdef HAS_YUY2TOYROW_MMI
ANY11(YUY2ToYRow_Any_MMI, YUY2ToYRow_MMI, 1, 4, 1, 7)
#endif
#ifdef HAS_UYVYTOYROW_MSA
ANY11(UYVYToYRow_Any_MSA, UYVYToYRow_MSA, 1, 4, 1, 31)
#endif
#ifdef HAS_UYVYTOYROW_LASX
ANY11(UYVYToYRow_Any_LASX, UYVYToYRow_LASX, 1, 4, 1, 31)
#endif
#ifdef HAS_UYVYTOYROW_MMI
ANY11(UYVYToYRow_Any_MMI, UYVYToYRow_MMI, 1, 4, 1, 15)
#endif
@ -1205,6 +1240,9 @@ ANY11(ARGBAttenuateRow_Any_MSA, ARGBAttenuateRow_MSA, 0, 4, 4, 7)
#ifdef HAS_ARGBATTENUATEROW_MMI
ANY11(ARGBAttenuateRow_Any_MMI, ARGBAttenuateRow_MMI, 0, 4, 4, 1)
#endif
#ifdef HAS_ARGBATTENUATEROW_LASX
ANY11(ARGBAttenuateRow_Any_LASX, ARGBAttenuateRow_LASX, 0, 4, 4, 15)
#endif
#ifdef HAS_ARGBEXTRACTALPHAROW_SSE2
ANY11(ARGBExtractAlphaRow_Any_SSE2, ARGBExtractAlphaRow_SSE2, 0, 4, 1, 7)
#endif
@ -1354,6 +1392,14 @@ ANY11P(ARGBToRGB565DitherRow_Any_MMI,
2,
3)
#endif
#if defined(HAS_ARGBTORGB565DITHERROW_LASX)
ANY11P(ARGBToRGB565DitherRow_Any_LASX,
ARGBToRGB565DitherRow_LASX,
const uint32_t,
4,
2,
15)
#endif
#ifdef HAS_ARGBSHUFFLEROW_SSSE3
ANY11P(ARGBShuffleRow_Any_SSSE3, ARGBShuffleRow_SSSE3, const uint8_t*, 4, 4, 7)
#endif
@ -1369,6 +1415,9 @@ ANY11P(ARGBShuffleRow_Any_MSA, ARGBShuffleRow_MSA, const uint8_t*, 4, 4, 7)
#ifdef HAS_ARGBSHUFFLEROW_MMI
ANY11P(ARGBShuffleRow_Any_MMI, ARGBShuffleRow_MMI, const uint8_t*, 4, 4, 1)
#endif
#ifdef HAS_ARGBSHUFFLEROW_LASX
ANY11P(ARGBShuffleRow_Any_LASX, ARGBShuffleRow_LASX, const uint8_t*, 4, 4, 15)
#endif
#undef ANY11P
#undef ANY11P
@ -1667,6 +1716,9 @@ ANY11M(MirrorRow_Any_MSA, MirrorRow_MSA, 1, 63)
#ifdef HAS_MIRRORROW_MMI
ANY11M(MirrorRow_Any_MMI, MirrorRow_MMI, 1, 7)
#endif
#ifdef HAS_MIRRORROW_LASX
ANY11M(MirrorRow_Any_LASX, MirrorRow_LASX, 1, 63)
#endif
#ifdef HAS_MIRRORUVROW_AVX2
ANY11M(MirrorUVRow_Any_AVX2, MirrorUVRow_AVX2, 2, 15)
#endif
@ -1679,6 +1731,9 @@ ANY11M(MirrorUVRow_Any_NEON, MirrorUVRow_NEON, 2, 31)
#ifdef HAS_MIRRORUVROW_MSA
ANY11M(MirrorUVRow_Any_MSA, MirrorUVRow_MSA, 2, 7)
#endif
#ifdef HAS_MIRRORUVROW_LASX
ANY11M(MirrorUVRow_Any_LASX, MirrorUVRow_LASX, 2, 15)
#endif
#ifdef HAS_ARGBMIRRORROW_AVX2
ANY11M(ARGBMirrorRow_Any_AVX2, ARGBMirrorRow_AVX2, 4, 7)
#endif
@ -1691,6 +1746,9 @@ ANY11M(ARGBMirrorRow_Any_NEON, ARGBMirrorRow_NEON, 4, 7)
#ifdef HAS_ARGBMIRRORROW_MSA
ANY11M(ARGBMirrorRow_Any_MSA, ARGBMirrorRow_MSA, 4, 15)
#endif
#ifdef HAS_ARGBMIRRORROW_LASX
ANY11M(ARGBMirrorRow_Any_LASX, ARGBMirrorRow_LASX, 4, 15)
#endif
#ifdef HAS_ARGBMIRRORROW_MMI
ANY11M(ARGBMirrorRow_Any_MMI, ARGBMirrorRow_MMI, 4, 1)
#endif
@ -1791,6 +1849,11 @@ ANY12(ARGBToUV444Row_Any_MMI, ARGBToUV444Row_MMI, 0, 4, 0, 7)
ANY12(UYVYToUV422Row_Any_MMI, UYVYToUV422Row_MMI, 1, 4, 1, 15)
ANY12(YUY2ToUV422Row_Any_MMI, YUY2ToUV422Row_MMI, 1, 4, 1, 15)
#endif
#ifdef HAS_YUY2TOUV422ROW_LASX
ANY12(ARGBToUV444Row_Any_LASX, ARGBToUV444Row_LASX, 0, 4, 0, 31)
ANY12(YUY2ToUV422Row_Any_LASX, YUY2ToUV422Row_LASX, 1, 4, 1, 31)
ANY12(UYVYToUV422Row_Any_LASX, UYVYToUV422Row_LASX, 1, 4, 1, 31)
#endif
#undef ANY12
// Any 2 16 bit planes with parameter to 1
@ -1951,6 +2014,9 @@ ANY12S(ARGBToUVRow_Any_MSA, ARGBToUVRow_MSA, 0, 4, 31)
#ifdef HAS_ARGBTOUVROW_MMI
ANY12S(ARGBToUVRow_Any_MMI, ARGBToUVRow_MMI, 0, 4, 15)
#endif
#ifdef HAS_ARGBTOUVROW_LASX
ANY12S(ARGBToUVRow_Any_LASX, ARGBToUVRow_LASX, 0, 4, 31)
#endif
#ifdef HAS_ARGBTOUVJROW_NEON
ANY12S(ARGBToUVJRow_Any_NEON, ARGBToUVJRow_NEON, 0, 4, 15)
#endif
@ -2047,9 +2113,15 @@ ANY12S(YUY2ToUVRow_Any_MSA, YUY2ToUVRow_MSA, 1, 4, 31)
#ifdef HAS_YUY2TOUVROW_MMI
ANY12S(YUY2ToUVRow_Any_MMI, YUY2ToUVRow_MMI, 1, 4, 15)
#endif
#ifdef HAS_YUY2TOUVROW_LASX
ANY12S(YUY2ToUVRow_Any_LASX, YUY2ToUVRow_LASX, 1, 4, 31)
#endif
#ifdef HAS_UYVYTOUVROW_MSA
ANY12S(UYVYToUVRow_Any_MSA, UYVYToUVRow_MSA, 1, 4, 31)
#endif
#ifdef HAS_UYVYTOUVROW_LASX
ANY12S(UYVYToUVRow_Any_LASX, UYVYToUVRow_LASX, 1, 4, 31)
#endif
#ifdef HAS_UYVYTOUVROW_MMI
ANY12S(UYVYToUVRow_Any_MMI, UYVYToUVRow_MMI, 1, 4, 15)
#endif

1042
source/row_lasx.cc
View File

File diff suppressed because it is too large Load Diff