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

Port Sobel to Native Client

Browse Source 
BUG=253
TEST=libyuvtest,trybots
R=fbarchard@google.com

Review URL: https://webrtc-codereview.appspot.com/2037004

git-svn-id: http://libyuv.googlecode.com/svn/trunk@764 16f28f9a-4ce2-e073-06de-1de4eb20be90
This commit is contained in:
nfullagar@google.com 2013-08-15 21:26:53 +00:00
parent bb5ea8e4df
commit 1c710a0606
4 changed files with 60 additions and 38 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: 763
Version: 764
License: BSD
License File: LICENSE

8
include/libyuv/row.h
View File

@ -56,6 +56,10 @@ extern "C" {
#define HAS_CUMULATIVESUMTOAVERAGEROW_SSE2
#define HAS_INTERPOLATEROW_SSE2
#define HAS_INTERPOLATEROW_SSSE3
#define HAS_SOBELROW_SSE2
#define HAS_SOBELXROW_SSSE3
#define HAS_SOBELXYROW_SSE2
#define HAS_SOBELYROW_SSSE3
// Conversions:
#define HAS_ARGBSHUFFLEROW_SSSE3
@ -131,10 +135,6 @@ extern "C" {
#define HAS_YUY2TOYROW_SSE2
// Effects:
#define HAS_SOBELROW_SSE2
#define HAS_SOBELXROW_SSSE3
#define HAS_SOBELXYROW_SSE2
#define HAS_SOBELYROW_SSSE3
#endif
// The following are Windows only:

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 763
#define LIBYUV_VERSION 764
#endif // INCLUDE_LIBYUV_VERSION_H_ NOLINT

86
source/row_posix.cc
View File

@ -4362,19 +4362,22 @@ void SobelXRow_SSSE3(const uint8* src_y0, const uint8* src_y1,
// 8 pixel loop.
".p2align 4 \n"
BUNDLEALIGN
"1: \n"
"movq (%0),%%xmm0 \n"
"movq 0x2(%0),%%xmm1 \n"
"movq "MEMACCESS(0)",%%xmm0 \n"
"movq "MEMACCESS2(0x2,0)",%%xmm1 \n"
"punpcklbw %%xmm5,%%xmm0 \n"
"punpcklbw %%xmm5,%%xmm1 \n"
"psubw %%xmm1,%%xmm0 \n"
"movq (%0,%1,1),%%xmm1 \n"
"movq 0x2(%0,%1,1),%%xmm2 \n"
BUNDLEALIGN
MEMOPREG(movq,0x00,0,1,1,xmm1) // movq (%0,%1,1),%%xmm1
MEMOPREG(movq,0x02,0,1,1,xmm2) // movq 0x2(%0,%1,1),%%xmm2
"punpcklbw %%xmm5,%%xmm1 \n"
"punpcklbw %%xmm5,%%xmm2 \n"
"psubw %%xmm2,%%xmm1 \n"
"movq (%0,%2,1),%%xmm2 \n"
"movq 0x2(%0,%2,1),%%xmm3 \n"
BUNDLEALIGN
MEMOPREG(movq,0x00,0,2,1,xmm2) // movq (%0,%2,1),%%xmm2
MEMOPREG(movq,0x02,0,2,1,xmm3) // movq 0x2(%0,%2,1),%%xmm3
"punpcklbw %%xmm5,%%xmm2 \n"
"punpcklbw %%xmm5,%%xmm3 \n"
"psubw %%xmm3,%%xmm2 \n"
@ -4384,8 +4387,9 @@ void SobelXRow_SSSE3(const uint8* src_y0, const uint8* src_y1,
"pabsw %%xmm0,%%xmm0 \n"
"packuswb %%xmm0,%%xmm0 \n"
"sub $0x8,%4 \n"
"movq %%xmm0,(%0,%3,1) \n"
"lea 0x8(%0),%0 \n"
BUNDLEALIGN
MEMOPMEM(movq,xmm0,0x00,0,3,1) // movq %%xmm0,(%0,%3,1)
"lea "MEMLEA(0x8,0)",%0 \n"
"jg 1b \n"
: "+r"(src_y0), // %0
"+r"(src_y1), // %1
@ -4394,6 +4398,9 @@ void SobelXRow_SSSE3(const uint8* src_y0, const uint8* src_y1,
"+r"(width) // %4
:
: "memory", "cc"
#if defined(__native_client__) && defined(__x86_64__)
, "r14"
#endif
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
#endif
@ -4415,19 +4422,22 @@ void SobelYRow_SSSE3(const uint8* src_y0, const uint8* src_y1,
// 8 pixel loop.
".p2align 4 \n"
BUNDLEALIGN
"1: \n"
"movq (%0),%%xmm0 \n"
"movq (%0,%1,1),%%xmm1 \n"
"movq "MEMACCESS(0)",%%xmm0 \n"
MEMOPREG(movq,0x00,0,1,1,xmm1) // movq (%0,%1,1),%%xmm1
"punpcklbw %%xmm5,%%xmm0 \n"
"punpcklbw %%xmm5,%%xmm1 \n"
"psubw %%xmm1,%%xmm0 \n"
"movq 0x1(%0),%%xmm1 \n"
"movq 0x1(%0,%1,1),%%xmm2 \n"
BUNDLEALIGN
"movq "MEMACCESS2(0x1,0)",%%xmm1 \n"
MEMOPREG(movq,0x01,0,1,1,xmm2) // movq 0x1(%0,%1,1),%%xmm2
"punpcklbw %%xmm5,%%xmm1 \n"
"punpcklbw %%xmm5,%%xmm2 \n"
"psubw %%xmm2,%%xmm1 \n"
"movq 0x2(%0),%%xmm2 \n"
"movq 0x2(%0,%1,1),%%xmm3 \n"
BUNDLEALIGN
"movq "MEMACCESS2(0x2,0)",%%xmm2 \n"
MEMOPREG(movq,0x02,0,1,1,xmm3) // movq 0x2(%0,%1,1),%%xmm3
"punpcklbw %%xmm5,%%xmm2 \n"
"punpcklbw %%xmm5,%%xmm3 \n"
"psubw %%xmm3,%%xmm2 \n"
@ -4437,8 +4447,9 @@ void SobelYRow_SSSE3(const uint8* src_y0, const uint8* src_y1,
"pabsw %%xmm0,%%xmm0 \n"
"packuswb %%xmm0,%%xmm0 \n"
"sub $0x8,%3 \n"
"movq %%xmm0,(%0,%2,1) \n"
"lea 0x8(%0),%0 \n"
BUNDLEALIGN
MEMOPMEM(movq,xmm0,0x00,0,2,1) // movq %%xmm0,(%0,%2,1)
"lea "MEMLEA(0x8,0)",%0 \n"
"jg 1b \n"
: "+r"(src_y0), // %0
"+r"(src_y1), // %1
@ -4446,6 +4457,9 @@ void SobelYRow_SSSE3(const uint8* src_y0, const uint8* src_y1,
"+r"(width) // %3
:
: "memory", "cc"
#if defined(__native_client__) && defined(__x86_64__)
, "r14"
#endif
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
#endif
@ -4468,10 +4482,11 @@ void SobelRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
// 8 pixel loop.
".p2align 4 \n"
BUNDLEALIGN
"1: \n"
"movdqa (%0),%%xmm0 \n"
"movdqa (%0,%1,1),%%xmm1 \n"
"lea 0x10(%0),%0 \n"
"movdqa "MEMACCESS(0)",%%xmm0 \n"
MEMOPREG(movdqa,0x00,0,1,1,xmm1) // movdqa (%0,%1,1),%%xmm1
"lea "MEMLEA(0x10,0)",%0 \n"
"paddusb %%xmm1,%%xmm0 \n"
"movdqa %%xmm0,%%xmm2 \n"
"punpcklbw %%xmm0,%%xmm2 \n"
@ -4487,11 +4502,11 @@ void SobelRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
"por %%xmm5,%%xmm3 \n"
"por %%xmm5,%%xmm0 \n"
"sub $0x10,%3 \n"
"movdqa %%xmm1,(%2) \n"
"movdqa %%xmm2,0x10(%2) \n"
"movdqa %%xmm3,0x20(%2) \n"
"movdqa %%xmm0,0x30(%2) \n"
"lea 0x40(%2),%2 \n"
"movdqa %%xmm1,"MEMACCESS(2)" \n"
"movdqa %%xmm2,"MEMACCESS2(0x10,2)" \n"
"movdqa %%xmm3,"MEMACCESS2(0x20,2)" \n"
"movdqa %%xmm0,"MEMACCESS2(0x30,2)" \n"
"lea "MEMLEA(0x40,2)",%2 \n"
"jg 1b \n"
: "+r"(src_sobelx), // %0
"+r"(src_sobely), // %1
@ -4499,6 +4514,9 @@ void SobelRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
"+r"(width) // %3
:
: "memory", "cc"
#if defined(__native_client__) && defined(__x86_64__)
, "r14"
#endif
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
#endif
@ -4520,10 +4538,11 @@ void SobelXYRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
// 8 pixel loop.
".p2align 4 \n"
BUNDLEALIGN
"1: \n"
"movdqa (%0),%%xmm0 \n"
"movdqa (%0,%1,1),%%xmm1 \n"
"lea 0x10(%0),%0 \n"
"movdqa "MEMACCESS(0)",%%xmm0 \n"
MEMOPREG(movdqa,0x00,0,1,1,xmm1) // movdqa (%0,%1,1),%%xmm1
"lea "MEMLEA(0x10,0)",%0 \n"
"movdqa %%xmm0,%%xmm2 \n"
"paddusb %%xmm1,%%xmm2 \n"
"movdqa %%xmm0,%%xmm3 \n"
@ -4539,11 +4558,11 @@ void SobelXYRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
"punpcklwd %%xmm0,%%xmm7 \n"
"punpckhwd %%xmm0,%%xmm1 \n"
"sub $0x10,%3 \n"
"movdqa %%xmm6,(%2) \n"
"movdqa %%xmm4,0x10(%2) \n"
"movdqa %%xmm7,0x20(%2) \n"
"movdqa %%xmm1,0x30(%2) \n"
"lea 0x40(%2),%2 \n"
"movdqa %%xmm6,"MEMACCESS(2)" \n"
"movdqa %%xmm4,"MEMACCESS2(0x10,2)" \n"
"movdqa %%xmm7,"MEMACCESS2(0x20,2)" \n"
"movdqa %%xmm1,"MEMACCESS2(0x30,2)" \n"
"lea "MEMLEA(0x40,2)",%2 \n"
"jg 1b \n"
: "+r"(src_sobelx), // %0
"+r"(src_sobely), // %1
@ -4551,6 +4570,9 @@ void SobelXYRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
"+r"(width) // %3
:
: "memory", "cc"
#if defined(__native_client__) && defined(__x86_64__)
, "r14"
#endif
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
#endif