coffee/libyuv
1
0
Fork 0
mirror of https://chromium.googlesource.com/libyuv/libyuv synced 2025-12-07 17:26:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

NaCL port of CopyAlpha

Browse Source 
BUG=none
TEST=ncval
R=nfullagar@google.com

Review URL: https://webrtc-codereview.appspot.com/2999004

git-svn-id: http://libyuv.googlecode.com/svn/trunk@827 16f28f9a-4ce2-e073-06de-1de4eb20be90
This commit is contained in:
fbarchard@google.com 2013-10-25 21:26:40 +00:00
parent 08b24a4232
commit fb99c03008
4 changed files with 148 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.chromium
View File

@ -1,6 +1,6 @@
Name: libyuv
URL: http://code.google.com/p/libyuv/
Version: 826
Version: 827
License: BSD
License File: LICENSE

8
include/libyuv/row.h
View File

@ -63,6 +63,8 @@ extern "C" {
#define HAS_SOBELXROW_SSE2
#define HAS_SOBELXYROW_SSE2
#define HAS_SOBELYROW_SSE2
#define HAS_ARGBCOPYALPHAROW_SSE2
#define HAS_ARGBCOPYYTOALPHAROW_SSE2
// Conversions:
#define HAS_ABGRTOUVROW_SSSE3
@ -163,6 +165,8 @@ extern "C" {
// Effects:
#define HAS_ARGBPOLYNOMIALROW_AVX2
#define HAS_ARGBSHUFFLEROW_AVX2
#define HAS_ARGBCOPYALPHAROW_AVX2
#define HAS_ARGBCOPYYTOALPHAROW_AVX2
#endif
// The following are Windows only:
@ -171,8 +175,6 @@ extern "C" {
// Effects:
// TODO(fbarchard): Optimize and enable
// #define HAS_ARGBLUMACOLORTABLEROW_SSSE3
#define HAS_ARGBCOPYALPHAROW_SSE2
#define HAS_ARGBCOPYYTOALPHAROW_SSE2
// Caveat: Visual C 2012 required for AVX2.
#if _MSC_VER >= 1700
@ -191,8 +193,6 @@ extern "C" {
#define HAS_YUY2TOUV422ROW_AVX2
#define HAS_YUY2TOUVROW_AVX2
#define HAS_YUY2TOYROW_AVX2
#define HAS_ARGBCOPYALPHAROW_AVX2
#define HAS_ARGBCOPYYTOALPHAROW_AVX2
// Effects:
#define HAS_ARGBADDROW_AVX2

2
include/libyuv/version.h
View File

@ -11,6 +11,6 @@
#ifndef INCLUDE_LIBYUV_VERSION_H_ // NOLINT
#define INCLUDE_LIBYUV_VERSION_H_
#define LIBYUV_VERSION 826
#define LIBYUV_VERSION 827
#endif // INCLUDE_LIBYUV_VERSION_H_ NOLINT

142
source/row_posix.cc
View File

@ -3276,6 +3276,148 @@ void CopyRow_ERMS(const uint8* src, uint8* dst, int width) {
}
#endif // HAS_COPYROW_ERMS
#ifdef HAS_ARGBCOPYALPHAROW_SSE2
// width in pixels
void ARGBCopyAlphaRow_SSE2(const uint8* src, uint8* dst, int width) {
asm volatile (
"pcmpeqb %%xmm0,%%xmm0 \n"
"pslld $0x18,%%xmm0 \n"
"pcmpeqb %%xmm1,%%xmm1 \n"
"psrld $0x8,%%xmm1 \n"
".p2align 4 \n"
"1: \n"
"movdqa " MEMACCESS(0) ",%%xmm2 \n"
"movdqa " MEMACCESS2(0x10,0) ",%%xmm3 \n"
"lea " MEMLEA(0x20,0) ",%0 \n"
"movdqa " MEMACCESS(1) ",%%xmm4 \n"
"movdqa " MEMACCESS2(0x10,1) ",%%xmm5 \n"
"pand %%xmm0,%%xmm2 \n"
"pand %%xmm0,%%xmm3 \n"
"pand %%xmm1,%%xmm4 \n"
"pand %%xmm1,%%xmm5 \n"
"por %%xmm4,%%xmm2 \n"
"por %%xmm5,%%xmm3 \n"
"movdqa %%xmm2," MEMACCESS(1) " \n"
"movdqa %%xmm3," MEMACCESS2(0x10,1) " \n"
"lea " MEMLEA(0x20,1) ",%1 \n"
"sub $0x8,%2 \n"
"jg 1b \n"
: "+r"(src), // %0
"+r"(dst), // %1
"+r"(width) // %2
:
: "memory", "cc"
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
#endif
);
}
#endif // HAS_ARGBCOPYALPHAROW_SSE2
#ifdef HAS_ARGBCOPYALPHAROW_AVX2
// width in pixels
void ARGBCopyAlphaRow_AVX2(const uint8* src, uint8* dst, int width) {
asm volatile (
"vpcmpeqb %%ymm0,%%ymm0,%%ymm0 \n"
"vpsrld $0x8,%%ymm0,%%ymm0 \n"
".p2align 4 \n"
"1: \n"
"vmovdqu " MEMACCESS(0) ",%%ymm1 \n"
"vmovdqu " MEMACCESS2(0x20,0) ",%%ymm2 \n"
"lea " MEMLEA(0x40,0) ",%0 \n"
"vpblendvb %%ymm0," MEMACCESS(1) ",%%ymm1,%%ymm1 \n"
"vpblendvb %%ymm0," MEMACCESS2(0x20,1) ",%%ymm2,%%ymm2 \n"
"vmovdqu %%ymm1," MEMACCESS(1) " \n"
"vmovdqu %%ymm2," MEMACCESS2(0x20,1) " \n"
"lea " MEMLEA(0x40,1) ",%1 \n"
"sub $0x10,%2 \n"
"jg 1b \n"
"vzeroupper \n"
: "+r"(src), // %0
"+r"(dst), // %1
"+r"(width) // %2
:
: "memory", "cc"
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2"
#endif
);
}
#endif // HAS_ARGBCOPYALPHAROW_AVX2
#ifdef HAS_ARGBCOPYYTOALPHAROW_SSE2
// width in pixels
void ARGBCopyYToAlphaRow_SSE2(const uint8* src, uint8* dst, int width) {
asm volatile (
"pcmpeqb %%xmm0,%%xmm0 \n"
"pslld $0x18,%%xmm0 \n"
"pcmpeqb %%xmm1,%%xmm1 \n"
"psrld $0x8,%%xmm1 \n"
".p2align 4 \n"
"1: \n"
"movq " MEMACCESS(0) ",%%xmm2 \n"
"lea " MEMLEA(0x8,0) ",%0 \n"
"punpcklbw %%xmm2,%%xmm2 \n"
"punpckhwd %%xmm2,%%xmm3 \n"
"punpcklwd %%xmm2,%%xmm2 \n"
"movdqa " MEMACCESS(1) ",%%xmm4 \n"
"movdqa " MEMACCESS2(0x10,1) ",%%xmm5 \n"
"pand %%xmm0,%%xmm2 \n"
"pand %%xmm0,%%xmm3 \n"
"pand %%xmm1,%%xmm4 \n"
"pand %%xmm1,%%xmm5 \n"
"por %%xmm4,%%xmm2 \n"
"por %%xmm5,%%xmm3 \n"
"movdqa %%xmm2," MEMACCESS(1) " \n"
"movdqa %%xmm3," MEMACCESS2(0x10,1) " \n"
"lea " MEMLEA(0x20,1) ",%1 \n"
"sub $0x8,%2 \n"
"jg 1b \n"
: "+r"(src), // %0
"+r"(dst), // %1
"+r"(width) // %2
:
: "memory", "cc"
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
#endif
);
}
#endif // HAS_ARGBCOPYYTOALPHAROW_SSE2
#ifdef HAS_ARGBCOPYYTOALPHAROW_AVX2
// width in pixels
void ARGBCopyYToAlphaRow_AVX2(const uint8* src, uint8* dst, int width) {
asm volatile (
"vpcmpeqb %%ymm0,%%ymm0,%%ymm0 \n"
"vpsrld $0x8,%%ymm0,%%ymm0 \n"
".p2align 4 \n"
"1: \n"
"vpmovzxbd " MEMACCESS(0) ",%%ymm1 \n"
"vpmovzxbd " MEMACCESS2(0x8,0) ",%%ymm2 \n"
"lea " MEMLEA(0x10,0) ",%0 \n"
"vpslld $0x18,%%ymm1,%%ymm1 \n"
"vpslld $0x18,%%ymm2,%%ymm2 \n"
"vpblendvb %%ymm0," MEMACCESS(1) ",%%ymm1,%%ymm1 \n"
"vpblendvb %%ymm0," MEMACCESS2(0x20,1) ",%%ymm2,%%ymm2 \n"
"vmovdqu %%ymm1," MEMACCESS(1) " \n"
"vmovdqu %%ymm2," MEMACCESS2(0x20,1) " \n"
"lea " MEMLEA(0x40,1) ",%1 \n"
"sub $0x10,%2 \n"
"jg 1b \n"
"vzeroupper \n"
: "+r"(src), // %0
"+r"(dst), // %1
"+r"(width) // %2
:
: "memory", "cc"
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2"
#endif
);
}
#endif // HAS_ARGBCOPYYTOALPHAROW_AVX2
#ifdef HAS_SETROW_X86
void SetRow_X86(uint8* dst, uint32 v32, int width) {
size_t width_tmp = static_cast<size_t>(width);