generic-library/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge swscale bloatup

Browse Source 
This will be cleaned up in the next merge

Authorship / merged commits:
commit f668afd4896ee65683619b6f165dda4cdd46766f
Author: Janne Grunau <janne-libav@jannau.net>
Date:   Fri Apr 15 09:12:34 2011 +0200

    swscale: fix "ISO C90 forbids mixed declarations and code" warning

    only hit with --enable-runtime-cpudetect

commit 7f2ae5c7af374dfe254195a9375974a2ff9395a7
Author: Janne Grunau <janne-libav@jannau.net>
Date:   Fri Apr 15 02:09:44 2011 +0200

    swscale: fix compilation with --enable-runtime-cpudetect

commit b6cad3df822969b31bb93eaf677e52a72416bc97
Author: Janne Grunau <janne-libav@jannau.net>
Date:   Fri Apr 15 00:31:04 2011 +0200

    swscale: correct include path to fix ppc altivec build

commit 6216fc70b74e01a5272085329aa92f5ac797f9cf
Author: Luca Barbato <lu_zero@gentoo.org>
Date:   Thu Apr 14 22:03:45 2011 +0200

    swscale: simplify rgb2rgb templating

    MMX is always built. Drop the ifdefs

commit 33a0421bbaa64f4e9c3d852b7f225ede8dad1388
Author: Josh Allmann <joshua.allmann@gmail.com>
Date:   Wed Apr 13 20:57:32 2011 +0200

    swscale: simplify initialization code

    Simplify the fallthrough case when no accelerated functions
    can be initialized.

commit 735bf1951171a1e0ee4292e84a1b1beac45dd0ab
Author: Josh Allmann <joshua.allmann@gmail.com>
Date:   Wed Apr 13 20:57:31 2011 +0200

    swscale: further cleanup swscale.c

    Move x86-specific constants out of swscale.c

commit 86330b4c9258d5e583c0db033d1e68f46443307c
Author: Luca Barbato <lu_zero@gentoo.org>
Date:   Wed Apr 13 20:57:30 2011 +0200

    swscale: partially move the arch specific code left

    PPC and x86 code is split off from swscale_template.c. Lots of code is
    still duplicated and should be removed later.

    Again uniformize the init system to be more similar to the dsputil one.

    Unset h*scale_fast in the x86 init in order to make the output
    consistent with the previous status. Thanks to Josh for spotting it.

commit c0038328830d7b341c28d7c99b0236a33617fd21
Author: Luca Barbato <lu_zero@gentoo.org>
Date:   Wed Apr 13 20:57:29 2011 +0200

    swscale: move away x86 specific code from rgb2rgb

    Keep only the plain C code in the main rgb2rgb.c and move the x86
    specific optimizations to x86/rgb2rgb.c
    Change the initialization pattern a little so some of it can be
    factorized to behave more like dsputils.

Conflicts:
	libswscale/rgb2rgb.c
	libswscale/swscale_template.c
This commit is contained in:
Michael Niedermayer 2011-05-25 05:23:39 +02:00
parent d9d56036f3
commit d1adad3cca
11 changed files with 6914 additions and 4606 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
libswscale/Makefile
View File

@ -12,7 +12,8 @@ OBJS-$(ARCH_BFIN) += bfin/internal_bfin.o \
bfin/yuv2rgb_bfin.o
OBJS-$(CONFIG_MLIB) += mlib/yuv2rgb_mlib.o
OBJS-$(HAVE_ALTIVEC) += ppc/yuv2rgb_altivec.o
OBJS-$(HAVE_MMX) += x86/yuv2rgb_mmx.o
OBJS-$(HAVE_MMX) += x86/rgb2rgb.o \
x86/yuv2rgb_mmx.o
OBJS-$(HAVE_VIS) += sparc/yuv2rgb_vis.o
TESTPROGS = colorspace swscale

918
libswscale/ppc/swscale_template.c Normal file
View File

@ -0,0 +1,918 @@
/*
* Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#if COMPILE_TEMPLATE_ALTIVEC
#include "swscale_altivec_template.c"
#endif
static inline void RENAME(yuv2yuvX)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize, const int16_t **alpSrc,
uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, long dstW, long chrDstW)
{
#if COMPILE_TEMPLATE_ALTIVEC
yuv2yuvX_altivec_real(lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
dest, uDest, vDest, dstW, chrDstW);
#else //COMPILE_TEMPLATE_ALTIVEC
yuv2yuvXinC(lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
alpSrc, dest, uDest, vDest, aDest, dstW, chrDstW);
#endif //!COMPILE_TEMPLATE_ALTIVEC
}
static inline void RENAME(yuv2nv12X)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, enum PixelFormat dstFormat)
{
yuv2nv12XinC(lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
dest, uDest, dstW, chrDstW, dstFormat);
}
static inline void RENAME(yuv2yuv1)(SwsContext *c, const int16_t *lumSrc, const int16_t *chrSrc, const int16_t *alpSrc,
uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, long dstW, long chrDstW)
{
int i;
for (i=0; i<dstW; i++) {
int val= (lumSrc[i]+64)>>7;
if (val&256) {
if (val<0) val=0;
else val=255;
}
dest[i]= val;
}
if (uDest)
for (i=0; i<chrDstW; i++) {
int u=(chrSrc[i ]+64)>>7;
int v=(chrSrc[i + VOFW]+64)>>7;
if ((u|v)&256) {
if (u<0) u=0;
else if (u>255) u=255;
if (v<0) v=0;
else if (v>255) v=255;
}
uDest[i]= u;
vDest[i]= v;
}
if (CONFIG_SWSCALE_ALPHA && aDest)
for (i=0; i<dstW; i++) {
int val= (alpSrc[i]+64)>>7;
aDest[i]= av_clip_uint8(val);
}
}
/**
* vertical scale YV12 to RGB
*/
static inline void RENAME(yuv2packedX)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
const int16_t **alpSrc, uint8_t *dest, long dstW, long dstY)
{
#if COMPILE_TEMPLATE_ALTIVEC
/* The following list of supported dstFormat values should
match what's found in the body of ff_yuv2packedX_altivec() */
if (!(c->flags & SWS_BITEXACT) && !c->alpPixBuf &&
(c->dstFormat==PIX_FMT_ABGR || c->dstFormat==PIX_FMT_BGRA ||
c->dstFormat==PIX_FMT_BGR24 || c->dstFormat==PIX_FMT_RGB24 ||
c->dstFormat==PIX_FMT_RGBA || c->dstFormat==PIX_FMT_ARGB))
ff_yuv2packedX_altivec(c, lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
dest, dstW, dstY);
else
#endif
yuv2packedXinC(c, lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
alpSrc, dest, dstW, dstY);
}
/**
* vertical bilinear scale YV12 to RGB
*/
static inline void RENAME(yuv2packed2)(SwsContext *c, const uint16_t *buf0, const uint16_t *buf1, const uint16_t *uvbuf0, const uint16_t *uvbuf1,
const uint16_t *abuf0, const uint16_t *abuf1, uint8_t *dest, int dstW, int yalpha, int uvalpha, int y)
{
int yalpha1=4095- yalpha;
int uvalpha1=4095-uvalpha;
int i;
YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB2_C, YSCALE_YUV_2_PACKED2_C(void,0), YSCALE_YUV_2_GRAY16_2_C, YSCALE_YUV_2_MONO2_C)
}
/**
* YV12 to RGB without scaling or interpolating
*/
static inline void RENAME(yuv2packed1)(SwsContext *c, const uint16_t *buf0, const uint16_t *uvbuf0, const uint16_t *uvbuf1,
const uint16_t *abuf0, uint8_t *dest, int dstW, int uvalpha, enum PixelFormat dstFormat, int flags, int y)
{
const int yalpha1=0;
int i;
const uint16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
const int yalpha= 4096; //FIXME ...
if (flags&SWS_FULL_CHR_H_INT) {
c->yuv2packed2(c, buf0, buf0, uvbuf0, uvbuf1, abuf0, abuf0, dest, dstW, 0, uvalpha, y);
return;
}
if (uvalpha < 2048) {
YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1_C, YSCALE_YUV_2_PACKED1_C(void,0), YSCALE_YUV_2_GRAY16_1_C, YSCALE_YUV_2_MONO2_C)
} else {
YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1B_C, YSCALE_YUV_2_PACKED1B_C(void,0), YSCALE_YUV_2_GRAY16_1_C, YSCALE_YUV_2_MONO2_C)
}
}
//FIXME yuy2* can read up to 7 samples too much
static inline void RENAME(yuy2ToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++)
dst[i]= src[2*i];
}
static inline void RENAME(yuy2ToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
dstU[i]= src1[4*i + 1];
dstV[i]= src1[4*i + 3];
}
assert(src1 == src2);
}
static inline void RENAME(LEToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
dstU[i]= src1[2*i + 1];
dstV[i]= src2[2*i + 1];
}
}
/* This is almost identical to the previous, end exists only because
* yuy2ToY/UV)(dst, src+1, ...) would have 100% unaligned accesses. */
static inline void RENAME(uyvyToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++)
dst[i]= src[2*i+1];
}
static inline void RENAME(uyvyToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
dstU[i]= src1[4*i + 0];
dstV[i]= src1[4*i + 2];
}
assert(src1 == src2);
}
static inline void RENAME(BEToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
dstU[i]= src1[2*i];
dstV[i]= src2[2*i];
}
}
static inline void RENAME(nvXXtoUV)(uint8_t *dst1, uint8_t *dst2,
const uint8_t *src, long width)
{
int i;
for (i = 0; i < width; i++) {
dst1[i] = src[2*i+0];
dst2[i] = src[2*i+1];
}
}
static inline void RENAME(nv12ToUV)(uint8_t *dstU, uint8_t *dstV,
const uint8_t *src1, const uint8_t *src2,
long width, uint32_t *unused)
{
RENAME(nvXXtoUV)(dstU, dstV, src1, width);
}
static inline void RENAME(nv21ToUV)(uint8_t *dstU, uint8_t *dstV,
const uint8_t *src1, const uint8_t *src2,
long width, uint32_t *unused)
{
RENAME(nvXXtoUV)(dstV, dstU, src1, width);
}
static inline void RENAME(bgr24ToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
int b= src[i*3+0];
int g= src[i*3+1];
int r= src[i*3+2];
dst[i]= ((RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
}
}
static inline void RENAME(bgr24ToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
int b= src1[3*i + 0];
int g= src1[3*i + 1];
int r= src1[3*i + 2];
dstU[i]= (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
dstV[i]= (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
}
assert(src1 == src2);
}
static inline void RENAME(bgr24ToUV_half)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
int b= src1[6*i + 0] + src1[6*i + 3];
int g= src1[6*i + 1] + src1[6*i + 4];
int r= src1[6*i + 2] + src1[6*i + 5];
dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
}
assert(src1 == src2);
}
static inline void RENAME(rgb24ToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
{
int i;
for (i=0; i<width; i++) {
int r= src[i*3+0];
int g= src[i*3+1];
int b= src[i*3+2];
dst[i]= ((RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
}
}
static inline void RENAME(rgb24ToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
assert(src1==src2);
for (i=0; i<width; i++) {
int r= src1[3*i + 0];
int g= src1[3*i + 1];
int b= src1[3*i + 2];
dstU[i]= (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
dstV[i]= (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
}
}
static inline void RENAME(rgb24ToUV_half)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
{
int i;
assert(src1==src2);
for (i=0; i<width; i++) {
int r= src1[6*i + 0] + src1[6*i + 3];
int g= src1[6*i + 1] + src1[6*i + 4];
int b= src1[6*i + 2] + src1[6*i + 5];
dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
}
}
// bilinear / bicubic scaling
static inline void RENAME(hScale)(int16_t *dst, int dstW, const uint8_t *src, int srcW, int xInc,
const int16_t *filter, const int16_t *filterPos, long filterSize)
{
#if COMPILE_TEMPLATE_ALTIVEC
hScale_altivec_real(dst, dstW, src, srcW, xInc, filter, filterPos, filterSize);
#else
int i;
for (i=0; i<dstW; i++) {
int j;
int srcPos= filterPos[i];
int val=0;
//printf("filterPos: %d\n", filterPos[i]);
for (j=0; j<filterSize; j++) {
//printf("filter: %d, src: %d\n", filter[i], src[srcPos + j]);
val += ((int)src[srcPos + j])*filter[filterSize*i + j];
}
//filter += hFilterSize;
dst[i] = FFMIN(val>>7, (1<<15)-1); // the cubic equation does overflow ...
//dst[i] = val>>7;
}
#endif /* COMPILE_TEMPLATE_ALTIVEC */
}
//FIXME all pal and rgb srcFormats could do this convertion as well
//FIXME all scalers more complex than bilinear could do half of this transform
static void RENAME(chrRangeToJpeg)(uint16_t *dst, int width)
{
int i;
for (i = 0; i < width; i++) {
dst[i ] = (FFMIN(dst[i ],30775)*4663 - 9289992)>>12; //-264
dst[i+VOFW] = (FFMIN(dst[i+VOFW],30775)*4663 - 9289992)>>12; //-264
}
}
static void RENAME(chrRangeFromJpeg)(uint16_t *dst, int width)
{
int i;
for (i = 0; i < width; i++) {
dst[i ] = (dst[i ]*1799 + 4081085)>>11; //1469
dst[i+VOFW] = (dst[i+VOFW]*1799 + 4081085)>>11; //1469
}
}
static void RENAME(lumRangeToJpeg)(uint16_t *dst, int width)
{
int i;
for (i = 0; i < width; i++)
dst[i] = (FFMIN(dst[i],30189)*19077 - 39057361)>>14;
}
static void RENAME(lumRangeFromJpeg)(uint16_t *dst, int width)
{
int i;
for (i = 0; i < width; i++)
dst[i] = (dst[i]*14071 + 33561947)>>14;
}
static inline void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
long dstWidth, const uint8_t *src, int srcW,
int xInc)
{
int i;
unsigned int xpos=0;
for (i=0;i<dstWidth;i++) {
register unsigned int xx=xpos>>16;
register unsigned int xalpha=(xpos&0xFFFF)>>9;
dst[i]= (src[xx]<<7) + (src[xx+1] - src[xx])*xalpha;
xpos+=xInc;
}
}
// *** horizontal scale Y line to temp buffer
static inline void RENAME(hyscale)(SwsContext *c, uint16_t *dst, long dstWidth, const uint8_t *src, int srcW, int xInc,
const int16_t *hLumFilter,
const int16_t *hLumFilterPos, int hLumFilterSize,
uint8_t *formatConvBuffer,
uint32_t *pal, int isAlpha)
{
void (*toYV12)(uint8_t *, const uint8_t *, long, uint32_t *) = isAlpha ? c->alpToYV12 : c->lumToYV12;
void (*convertRange)(int16_t *, int) = isAlpha ? NULL : c->lumConvertRange;
src += isAlpha ? c->alpSrcOffset : c->lumSrcOffset;
if (toYV12) {
toYV12(formatConvBuffer, src, srcW, pal);
src= formatConvBuffer;
}
if (c->hScale16) {
c->hScale16(dst, dstWidth, (uint16_t*)src, srcW, xInc, hLumFilter, hLumFilterPos, hLumFilterSize, av_pix_fmt_descriptors[c->srcFormat].comp[0].depth_minus1);
} else if (!c->hyscale_fast) {
c->hScale(dst, dstWidth, src, srcW, xInc, hLumFilter, hLumFilterPos, hLumFilterSize);
} else { // fast bilinear upscale / crap downscale
c->hyscale_fast(c, dst, dstWidth, src, srcW, xInc);
}
if (convertRange)
convertRange(dst, dstWidth);
}
static inline void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst,
long dstWidth, const uint8_t *src1,
const uint8_t *src2, int srcW, int xInc)
{
int i;
unsigned int xpos=0;
for (i=0;i<dstWidth;i++) {
register unsigned int xx=xpos>>16;
register unsigned int xalpha=(xpos&0xFFFF)>>9;
dst[i]=(src1[xx]*(xalpha^127)+src1[xx+1]*xalpha);
dst[i+VOFW]=(src2[xx]*(xalpha^127)+src2[xx+1]*xalpha);
/* slower
dst[i]= (src1[xx]<<7) + (src1[xx+1] - src1[xx])*xalpha;
dst[i+VOFW]=(src2[xx]<<7) + (src2[xx+1] - src2[xx])*xalpha;
*/
xpos+=xInc;
}
}
inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth, const uint8_t *src1, const uint8_t *src2,
int srcW, int xInc, const int16_t *hChrFilter,
const int16_t *hChrFilterPos, int hChrFilterSize,
uint8_t *formatConvBuffer,
uint32_t *pal)
{
src1 += c->chrSrcOffset;
src2 += c->chrSrcOffset;
if (c->chrToYV12) {
c->chrToYV12(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW, pal);
src1= formatConvBuffer;
src2= formatConvBuffer+VOFW;
}
if (c->hScale16) {
c->hScale16(dst , dstWidth, (uint16_t*)src1, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize, av_pix_fmt_descriptors[c->srcFormat].comp[0].depth_minus1);
c->hScale16(dst+VOFW, dstWidth, (uint16_t*)src2, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize, av_pix_fmt_descriptors[c->srcFormat].comp[0].depth_minus1);
} else if (!c->hcscale_fast) {
c->hScale(dst , dstWidth, src1, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
c->hScale(dst+VOFW, dstWidth, src2, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
} else { // fast bilinear upscale / crap downscale
c->hcscale_fast(c, dst, dstWidth, src1, src2, srcW, xInc);
}
if (c->chrConvertRange)
c->chrConvertRange(dst, dstWidth);
}
#define DEBUG_SWSCALE_BUFFERS 0
#define DEBUG_BUFFERS(...) if (DEBUG_SWSCALE_BUFFERS) av_log(c, AV_LOG_DEBUG, __VA_ARGS__)
static int RENAME(swScale)(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
int srcSliceH, uint8_t* dst[], int dstStride[])
{
/* load a few things into local vars to make the code more readable? and faster */
const int srcW= c->srcW;
const int dstW= c->dstW;
const int dstH= c->dstH;
const int chrDstW= c->chrDstW;
const int chrSrcW= c->chrSrcW;
const int lumXInc= c->lumXInc;
const int chrXInc= c->chrXInc;
const enum PixelFormat dstFormat= c->dstFormat;
const int flags= c->flags;
int16_t *vLumFilterPos= c->vLumFilterPos;
int16_t *vChrFilterPos= c->vChrFilterPos;
int16_t *hLumFilterPos= c->hLumFilterPos;
int16_t *hChrFilterPos= c->hChrFilterPos;
int16_t *vLumFilter= c->vLumFilter;
int16_t *vChrFilter= c->vChrFilter;
int16_t *hLumFilter= c->hLumFilter;
int16_t *hChrFilter= c->hChrFilter;
int32_t *lumMmxFilter= c->lumMmxFilter;
int32_t *chrMmxFilter= c->chrMmxFilter;
int32_t av_unused *alpMmxFilter= c->alpMmxFilter;
const int vLumFilterSize= c->vLumFilterSize;
const int vChrFilterSize= c->vChrFilterSize;
const int hLumFilterSize= c->hLumFilterSize;
const int hChrFilterSize= c->hChrFilterSize;
int16_t **lumPixBuf= c->lumPixBuf;
int16_t **chrPixBuf= c->chrPixBuf;
int16_t **alpPixBuf= c->alpPixBuf;
const int vLumBufSize= c->vLumBufSize;
const int vChrBufSize= c->vChrBufSize;
uint8_t *formatConvBuffer= c->formatConvBuffer;
const int chrSrcSliceY= srcSliceY >> c->chrSrcVSubSample;
const int chrSrcSliceH= -((-srcSliceH) >> c->chrSrcVSubSample);
int lastDstY;
uint32_t *pal=c->pal_yuv;
/* vars which will change and which we need to store back in the context */
int dstY= c->dstY;
int lumBufIndex= c->lumBufIndex;
int chrBufIndex= c->chrBufIndex;
int lastInLumBuf= c->lastInLumBuf;
int lastInChrBuf= c->lastInChrBuf;
if (isPacked(c->srcFormat)) {
src[0]=
src[1]=
src[2]=
src[3]= src[0];
srcStride[0]=
srcStride[1]=
srcStride[2]=
srcStride[3]= srcStride[0];
}
srcStride[1]<<= c->vChrDrop;
srcStride[2]<<= c->vChrDrop;
DEBUG_BUFFERS("swScale() %p[%d] %p[%d] %p[%d] %p[%d] -> %p[%d] %p[%d] %p[%d] %p[%d]\n",
src[0], srcStride[0], src[1], srcStride[1], src[2], srcStride[2], src[3], srcStride[3],
dst[0], dstStride[0], dst[1], dstStride[1], dst[2], dstStride[2], dst[3], dstStride[3]);
DEBUG_BUFFERS("srcSliceY: %d srcSliceH: %d dstY: %d dstH: %d\n",
srcSliceY, srcSliceH, dstY, dstH);
DEBUG_BUFFERS("vLumFilterSize: %d vLumBufSize: %d vChrFilterSize: %d vChrBufSize: %d\n",
vLumFilterSize, vLumBufSize, vChrFilterSize, vChrBufSize);
if (dstStride[0]%8 !=0 || dstStride[1]%8 !=0 || dstStride[2]%8 !=0 || dstStride[3]%8 != 0) {
static int warnedAlready=0; //FIXME move this into the context perhaps
if (flags & SWS_PRINT_INFO && !warnedAlready) {
av_log(c, AV_LOG_WARNING, "Warning: dstStride is not aligned!\n"
" ->cannot do aligned memory accesses anymore\n");
warnedAlready=1;
}
}
/* Note the user might start scaling the picture in the middle so this
will not get executed. This is not really intended but works
currently, so people might do it. */
if (srcSliceY ==0) {
lumBufIndex=-1;
chrBufIndex=-1;
dstY=0;
lastInLumBuf= -1;
lastInChrBuf= -1;
}
lastDstY= dstY;
for (;dstY < dstH; dstY++) {
unsigned char *dest =dst[0]+dstStride[0]*dstY;
const int chrDstY= dstY>>c->chrDstVSubSample;
unsigned char *uDest=dst[1]+dstStride[1]*chrDstY;
unsigned char *vDest=dst[2]+dstStride[2]*chrDstY;
unsigned char *aDest=(CONFIG_SWSCALE_ALPHA && alpPixBuf) ? dst[3]+dstStride[3]*dstY : NULL;
const int firstLumSrcY= vLumFilterPos[dstY]; //First line needed as input
const int firstLumSrcY2= vLumFilterPos[FFMIN(dstY | ((1<<c->chrDstVSubSample) - 1), dstH-1)];
const int firstChrSrcY= vChrFilterPos[chrDstY]; //First line needed as input
int lastLumSrcY= firstLumSrcY + vLumFilterSize -1; // Last line needed as input
int lastLumSrcY2=firstLumSrcY2+ vLumFilterSize -1; // Last line needed as input
int lastChrSrcY= firstChrSrcY + vChrFilterSize -1; // Last line needed as input
int enough_lines;
//handle holes (FAST_BILINEAR & weird filters)
if (firstLumSrcY > lastInLumBuf) lastInLumBuf= firstLumSrcY-1;
if (firstChrSrcY > lastInChrBuf) lastInChrBuf= firstChrSrcY-1;
assert(firstLumSrcY >= lastInLumBuf - vLumBufSize + 1);
assert(firstChrSrcY >= lastInChrBuf - vChrBufSize + 1);
DEBUG_BUFFERS("dstY: %d\n", dstY);
DEBUG_BUFFERS("\tfirstLumSrcY: %d lastLumSrcY: %d lastInLumBuf: %d\n",
firstLumSrcY, lastLumSrcY, lastInLumBuf);
DEBUG_BUFFERS("\tfirstChrSrcY: %d lastChrSrcY: %d lastInChrBuf: %d\n",
firstChrSrcY, lastChrSrcY, lastInChrBuf);
// Do we have enough lines in this slice to output the dstY line
enough_lines = lastLumSrcY2 < srcSliceY + srcSliceH && lastChrSrcY < -((-srcSliceY - srcSliceH)>>c->chrSrcVSubSample);
if (!enough_lines) {
lastLumSrcY = srcSliceY + srcSliceH - 1;
lastChrSrcY = chrSrcSliceY + chrSrcSliceH - 1;
DEBUG_BUFFERS("buffering slice: lastLumSrcY %d lastChrSrcY %d\n",
lastLumSrcY, lastChrSrcY);
}
//Do horizontal scaling
while(lastInLumBuf < lastLumSrcY) {
const uint8_t *src1= src[0]+(lastInLumBuf + 1 - srcSliceY)*srcStride[0];
const uint8_t *src2= src[3]+(lastInLumBuf + 1 - srcSliceY)*srcStride[3];
lumBufIndex++;
assert(lumBufIndex < 2*vLumBufSize);
assert(lastInLumBuf + 1 - srcSliceY < srcSliceH);
assert(lastInLumBuf + 1 - srcSliceY >= 0);
RENAME(hyscale)(c, lumPixBuf[ lumBufIndex ], dstW, src1, srcW, lumXInc,
hLumFilter, hLumFilterPos, hLumFilterSize,
formatConvBuffer,
pal, 0);
if (CONFIG_SWSCALE_ALPHA && alpPixBuf)
RENAME(hyscale)(c, alpPixBuf[ lumBufIndex ], dstW, src2, srcW, lumXInc,
hLumFilter, hLumFilterPos, hLumFilterSize,
formatConvBuffer,
pal, 1);
lastInLumBuf++;
DEBUG_BUFFERS("\t\tlumBufIndex %d: lastInLumBuf: %d\n",
lumBufIndex, lastInLumBuf);
}
while(lastInChrBuf < lastChrSrcY) {
const uint8_t *src1= src[1]+(lastInChrBuf + 1 - chrSrcSliceY)*srcStride[1];
const uint8_t *src2= src[2]+(lastInChrBuf + 1 - chrSrcSliceY)*srcStride[2];
chrBufIndex++;
assert(chrBufIndex < 2*vChrBufSize);
assert(lastInChrBuf + 1 - chrSrcSliceY < (chrSrcSliceH));
assert(lastInChrBuf + 1 - chrSrcSliceY >= 0);
//FIXME replace parameters through context struct (some at least)
if (c->needs_hcscale)
RENAME(hcscale)(c, chrPixBuf[ chrBufIndex ], chrDstW, src1, src2, chrSrcW, chrXInc,
hChrFilter, hChrFilterPos, hChrFilterSize,
formatConvBuffer,
pal);
lastInChrBuf++;
DEBUG_BUFFERS("\t\tchrBufIndex %d: lastInChrBuf: %d\n",
chrBufIndex, lastInChrBuf);
}
//wrap buf index around to stay inside the ring buffer
if (lumBufIndex >= vLumBufSize) lumBufIndex-= vLumBufSize;
if (chrBufIndex >= vChrBufSize) chrBufIndex-= vChrBufSize;
if (!enough_lines)
break; //we can't output a dstY line so let's try with the next slice
if (dstY < dstH-2) {
const int16_t **lumSrcPtr= (const int16_t **) lumPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize;
const int16_t **chrSrcPtr= (const int16_t **) chrPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
const int16_t **alpSrcPtr= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? (const int16_t **) alpPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize : NULL;
if (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21) {
const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
if (dstY&chrSkipMask) uDest= NULL; //FIXME split functions in lumi / chromi
c->yuv2nv12X(c,
vLumFilter+dstY*vLumFilterSize , lumSrcPtr, vLumFilterSize,
vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
dest, uDest, dstW, chrDstW, dstFormat);
} else if (isPlanarYUV(dstFormat) || dstFormat==PIX_FMT_GRAY8) { //YV12 like
const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
if ((dstY&chrSkipMask) || isGray(dstFormat)) uDest=vDest= NULL; //FIXME split functions in lumi / chromi
if (is16BPS(dstFormat) || isNBPS(dstFormat)) {
yuv2yuvX16inC(
vLumFilter+dstY*vLumFilterSize , lumSrcPtr, vLumFilterSize,
vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, (uint16_t *) dest, (uint16_t *) uDest, (uint16_t *) vDest, (uint16_t *) aDest, dstW, chrDstW,
dstFormat);
} else if (vLumFilterSize == 1 && vChrFilterSize == 1) { // unscaled YV12
const int16_t *lumBuf = lumSrcPtr[0];
const int16_t *chrBuf= chrSrcPtr[0];
const int16_t *alpBuf= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? alpSrcPtr[0] : NULL;
c->yuv2yuv1(c, lumBuf, chrBuf, alpBuf, dest, uDest, vDest, aDest, dstW, chrDstW);
} else { //General YV12
c->yuv2yuvX(c,
vLumFilter+dstY*vLumFilterSize , lumSrcPtr, vLumFilterSize,
vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, uDest, vDest, aDest, dstW, chrDstW);
}
} else {
assert(lumSrcPtr + vLumFilterSize - 1 < lumPixBuf + vLumBufSize*2);
assert(chrSrcPtr + vChrFilterSize - 1 < chrPixBuf + vChrBufSize*2);
if (vLumFilterSize == 1 && vChrFilterSize == 2) { //unscaled RGB
int chrAlpha= vChrFilter[2*dstY+1];
if(flags & SWS_FULL_CHR_H_INT) {
yuv2rgbXinC_full(c, //FIXME write a packed1_full function
vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, dstW, dstY);
} else {
c->yuv2packed1(c, *lumSrcPtr, *chrSrcPtr, *(chrSrcPtr+1),
alpPixBuf ? *alpSrcPtr : NULL,
dest, dstW, chrAlpha, dstFormat, flags, dstY);
}
} else if (vLumFilterSize == 2 && vChrFilterSize == 2) { //bilinear upscale RGB
int lumAlpha= vLumFilter[2*dstY+1];
int chrAlpha= vChrFilter[2*dstY+1];
lumMmxFilter[2]=
lumMmxFilter[3]= vLumFilter[2*dstY ]*0x10001;
chrMmxFilter[2]=
chrMmxFilter[3]= vChrFilter[2*chrDstY]*0x10001;
if(flags & SWS_FULL_CHR_H_INT) {
yuv2rgbXinC_full(c, //FIXME write a packed2_full function
vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, dstW, dstY);
} else {
c->yuv2packed2(c, *lumSrcPtr, *(lumSrcPtr+1), *chrSrcPtr, *(chrSrcPtr+1),
alpPixBuf ? *alpSrcPtr : NULL, alpPixBuf ? *(alpSrcPtr+1) : NULL,
dest, dstW, lumAlpha, chrAlpha, dstY);
}
} else { //general RGB
if(flags & SWS_FULL_CHR_H_INT) {
yuv2rgbXinC_full(c,
vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, dstW, dstY);
} else {
c->yuv2packedX(c,
vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, dstW, dstY);
}
}
}
} else { // hmm looks like we can't use MMX here without overwriting this array's tail
const int16_t **lumSrcPtr= (const int16_t **)lumPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize;
const int16_t **chrSrcPtr= (const int16_t **)chrPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
const int16_t **alpSrcPtr= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? (const int16_t **)alpPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize : NULL;
if (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21) {
const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
if (dstY&chrSkipMask) uDest= NULL; //FIXME split functions in lumi / chromi
yuv2nv12XinC(
vLumFilter+dstY*vLumFilterSize , lumSrcPtr, vLumFilterSize,
vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
dest, uDest, dstW, chrDstW, dstFormat);
} else if (isPlanarYUV(dstFormat) || dstFormat==PIX_FMT_GRAY8) { //YV12
const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
if ((dstY&chrSkipMask) || isGray(dstFormat)) uDest=vDest= NULL; //FIXME split functions in lumi / chromi
if (is16BPS(dstFormat) || isNBPS(dstFormat)) {
yuv2yuvX16inC(
vLumFilter+dstY*vLumFilterSize , lumSrcPtr, vLumFilterSize,
vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, (uint16_t *) dest, (uint16_t *) uDest, (uint16_t *) vDest, (uint16_t *) aDest, dstW, chrDstW,
dstFormat);
} else {
yuv2yuvXinC(
vLumFilter+dstY*vLumFilterSize , lumSrcPtr, vLumFilterSize,
vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, uDest, vDest, aDest, dstW, chrDstW);
}
} else {
assert(lumSrcPtr + vLumFilterSize - 1 < lumPixBuf + vLumBufSize*2);
assert(chrSrcPtr + vChrFilterSize - 1 < chrPixBuf + vChrBufSize*2);
if(flags & SWS_FULL_CHR_H_INT) {
yuv2rgbXinC_full(c,
vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, dstW, dstY);
} else {
yuv2packedXinC(c,
vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
alpSrcPtr, dest, dstW, dstY);
}
}
}
}
if ((dstFormat == PIX_FMT_YUVA420P) && !alpPixBuf)
fillPlane(dst[3], dstStride[3], dstW, dstY-lastDstY, lastDstY, 255);
/* store changed local vars back in the context */
c->dstY= dstY;
c->lumBufIndex= lumBufIndex;
c->chrBufIndex= chrBufIndex;
c->lastInLumBuf= lastInLumBuf;
c->lastInChrBuf= lastInChrBuf;
return dstY - lastDstY;
}
static void RENAME(sws_init_swScale)(SwsContext *c)
{
enum PixelFormat srcFormat = c->srcFormat;
c->yuv2nv12X = RENAME(yuv2nv12X );
c->yuv2yuv1 = RENAME(yuv2yuv1 );
c->yuv2yuvX = RENAME(yuv2yuvX );
c->yuv2packed1 = RENAME(yuv2packed1 );
c->yuv2packed2 = RENAME(yuv2packed2 );
c->yuv2packedX = RENAME(yuv2packedX );
c->hScale = RENAME(hScale );
if (c->flags & SWS_FAST_BILINEAR)
{
c->hyscale_fast = RENAME(hyscale_fast);
c->hcscale_fast = RENAME(hcscale_fast);
}
c->chrToYV12 = NULL;
switch(srcFormat) {
case PIX_FMT_YUYV422 : c->chrToYV12 = RENAME(yuy2ToUV); break;
case PIX_FMT_UYVY422 : c->chrToYV12 = RENAME(uyvyToUV); break;
case PIX_FMT_NV12 : c->chrToYV12 = RENAME(nv12ToUV); break;
case PIX_FMT_NV21 : c->chrToYV12 = RENAME(nv21ToUV); break;
case PIX_FMT_RGB8 :
case PIX_FMT_BGR8 :
case PIX_FMT_PAL8 :
case PIX_FMT_BGR4_BYTE:
case PIX_FMT_RGB4_BYTE: c->chrToYV12 = palToUV; break;
case PIX_FMT_GRAY16BE :
case PIX_FMT_YUV420P9BE:
case PIX_FMT_YUV422P10BE:
case PIX_FMT_YUV420P10BE:
case PIX_FMT_YUV420P16BE:
case PIX_FMT_YUV422P16BE:
case PIX_FMT_YUV444P16BE: c->hScale16= HAVE_BIGENDIAN ? RENAME(hScale16) : RENAME(hScale16X); break;
case PIX_FMT_GRAY16LE :
case PIX_FMT_YUV420P9LE:
case PIX_FMT_YUV422P10LE:
case PIX_FMT_YUV420P10LE:
case PIX_FMT_YUV420P16LE:
case PIX_FMT_YUV422P16LE:
case PIX_FMT_YUV444P16LE: c->hScale16= HAVE_BIGENDIAN ? RENAME(hScale16X) : RENAME(hScale16); break;
}
if (c->chrSrcHSubSample) {
switch(srcFormat) {
case PIX_FMT_RGB48BE:
case PIX_FMT_RGB48LE: c->chrToYV12 = rgb48ToUV_half; break;
case PIX_FMT_BGR48BE:
case PIX_FMT_BGR48LE: c->chrToYV12 = bgr48ToUV_half; break;
case PIX_FMT_RGB32 : c->chrToYV12 = bgr32ToUV_half; break;
case PIX_FMT_RGB32_1: c->chrToYV12 = bgr321ToUV_half; break;
case PIX_FMT_BGR24 : c->chrToYV12 = RENAME(bgr24ToUV_half); break;
case PIX_FMT_BGR565 : c->chrToYV12 = bgr16ToUV_half; break;
case PIX_FMT_BGR555 : c->chrToYV12 = bgr15ToUV_half; break;
case PIX_FMT_BGR32 : c->chrToYV12 = rgb32ToUV_half; break;
case PIX_FMT_BGR32_1: c->chrToYV12 = rgb321ToUV_half; break;
case PIX_FMT_RGB24 : c->chrToYV12 = RENAME(rgb24ToUV_half); break;
case PIX_FMT_RGB565 : c->chrToYV12 = rgb16ToUV_half; break;
case PIX_FMT_RGB555 : c->chrToYV12 = rgb15ToUV_half; break;
}
} else {
switch(srcFormat) {
case PIX_FMT_RGB48BE:
case PIX_FMT_RGB48LE: c->chrToYV12 = rgb48ToUV; break;
case PIX_FMT_BGR48BE:
case PIX_FMT_BGR48LE: c->chrToYV12 = bgr48ToUV; break;
case PIX_FMT_RGB32 : c->chrToYV12 = bgr32ToUV; break;
case PIX_FMT_RGB32_1: c->chrToYV12 = bgr321ToUV; break;
case PIX_FMT_BGR24 : c->chrToYV12 = RENAME(bgr24ToUV); break;
case PIX_FMT_BGR565 : c->chrToYV12 = bgr16ToUV; break;
case PIX_FMT_BGR555 : c->chrToYV12 = bgr15ToUV; break;
case PIX_FMT_BGR32 : c->chrToYV12 = rgb32ToUV; break;
case PIX_FMT_BGR32_1: c->chrToYV12 = rgb321ToUV; break;
case PIX_FMT_RGB24 : c->chrToYV12 = RENAME(rgb24ToUV); break;
case PIX_FMT_RGB565 : c->chrToYV12 = rgb16ToUV; break;
case PIX_FMT_RGB555 : c->chrToYV12 = rgb15ToUV; break;
}
}
c->lumToYV12 = NULL;
c->alpToYV12 = NULL;
switch (srcFormat) {
case PIX_FMT_YUYV422 :
case PIX_FMT_GRAY8A :
c->lumToYV12 = RENAME(yuy2ToY); break;
case PIX_FMT_UYVY422 :
c->lumToYV12 = RENAME(uyvyToY); break;
case PIX_FMT_BGR24 : c->lumToYV12 = RENAME(bgr24ToY); break;
case PIX_FMT_BGR565 : c->lumToYV12 = bgr16ToY; break;
case PIX_FMT_BGR555 : c->lumToYV12 = bgr15ToY; break;
case PIX_FMT_RGB24 : c->lumToYV12 = RENAME(rgb24ToY); break;
case PIX_FMT_RGB565 : c->lumToYV12 = rgb16ToY; break;
case PIX_FMT_RGB555 : c->lumToYV12 = rgb15ToY; break;
case PIX_FMT_RGB8 :
case PIX_FMT_BGR8 :
case PIX_FMT_PAL8 :
case PIX_FMT_BGR4_BYTE:
case PIX_FMT_RGB4_BYTE: c->lumToYV12 = palToY; break;
case PIX_FMT_MONOBLACK: c->lumToYV12 = monoblack2Y; break;
case PIX_FMT_MONOWHITE: c->lumToYV12 = monowhite2Y; break;
case PIX_FMT_RGB32 : c->lumToYV12 = bgr32ToY; break;
case PIX_FMT_RGB32_1: c->lumToYV12 = bgr321ToY; break;
case PIX_FMT_BGR32 : c->lumToYV12 = rgb32ToY; break;
case PIX_FMT_BGR32_1: c->lumToYV12 = rgb321ToY; break;
case PIX_FMT_RGB48BE:
case PIX_FMT_RGB48LE: c->lumToYV12 = rgb48ToY; break;
case PIX_FMT_BGR48BE:
case PIX_FMT_BGR48LE: c->lumToYV12 = bgr48ToY; break;
}
if (c->alpPixBuf) {
switch (srcFormat) {
case PIX_FMT_RGB32 :
case PIX_FMT_RGB32_1:
case PIX_FMT_BGR32 :
case PIX_FMT_BGR32_1: c->alpToYV12 = abgrToA; break;
case PIX_FMT_GRAY8A : c->alpToYV12 = RENAME(yuy2ToY); break;
case PIX_FMT_PAL8 : c->alpToYV12 = palToA; break;
}
}
switch (srcFormat) {
case PIX_FMT_GRAY8A :
c->alpSrcOffset = 1;
break;
case PIX_FMT_RGB32 :
case PIX_FMT_BGR32 :
c->alpSrcOffset = 3;
break;
case PIX_FMT_RGB48LE:
case PIX_FMT_BGR48LE:
c->lumSrcOffset = 1;
c->chrSrcOffset = 1;
c->alpSrcOffset = 1;
break;
}
if (c->srcRange != c->dstRange && !isAnyRGB(c->dstFormat)) {
if (c->srcRange) {
c->lumConvertRange = RENAME(lumRangeFromJpeg);
c->chrConvertRange = RENAME(chrRangeFromJpeg);
} else {
c->lumConvertRange = RENAME(lumRangeToJpeg);
c->chrConvertRange = RENAME(chrRangeToJpeg);
}
}
if (!(isGray(srcFormat) || isGray(c->dstFormat) ||
srcFormat == PIX_FMT_MONOBLACK || srcFormat == PIX_FMT_MONOWHITE))
c->needs_hcscale = 1;
}

95
libswscale/rgb2rgb.c
View File

@ -24,7 +24,6 @@
*/
#include <inttypes.h>
#include "config.h"
#include "libavutil/x86_cpu.h"
#include "libavutil/bswap.h"
#include "rgb2rgb.h"
#include "swscale.h"
@ -95,45 +94,6 @@ void (*yuyvtoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t
long width, long height,
long lumStride, long chromStride, long srcStride);
#if ARCH_X86
DECLARE_ASM_CONST(8, uint64_t, mmx_ff) = 0x00000000000000FFULL;
DECLARE_ASM_CONST(8, uint64_t, mmx_null) = 0x0000000000000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mmx_one) = 0xFFFFFFFFFFFFFFFFULL;
DECLARE_ASM_CONST(8, uint64_t, mask32b) = 0x000000FF000000FFULL;
DECLARE_ASM_CONST(8, uint64_t, mask32g) = 0x0000FF000000FF00ULL;
DECLARE_ASM_CONST(8, uint64_t, mask32r) = 0x00FF000000FF0000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask32a) = 0xFF000000FF000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask32) = 0x00FFFFFF00FFFFFFULL;
DECLARE_ASM_CONST(8, uint64_t, mask3216br) = 0x00F800F800F800F8ULL;
DECLARE_ASM_CONST(8, uint64_t, mask3216g) = 0x0000FC000000FC00ULL;
DECLARE_ASM_CONST(8, uint64_t, mask3215g) = 0x0000F8000000F800ULL;
DECLARE_ASM_CONST(8, uint64_t, mul3216) = 0x2000000420000004ULL;
DECLARE_ASM_CONST(8, uint64_t, mul3215) = 0x2000000820000008ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24b) = 0x00FF0000FF0000FFULL;
DECLARE_ASM_CONST(8, uint64_t, mask24g) = 0xFF0000FF0000FF00ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24r) = 0x0000FF0000FF0000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24l) = 0x0000000000FFFFFFULL;
DECLARE_ASM_CONST(8, uint64_t, mask24h) = 0x0000FFFFFF000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24hh) = 0xffff000000000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24hhh) = 0xffffffff00000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24hhhh) = 0xffffffffffff0000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask15b) = 0x001F001F001F001FULL; /* 00000000 00011111 xxB */
DECLARE_ASM_CONST(8, uint64_t, mask15rg) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000 RGx */
DECLARE_ASM_CONST(8, uint64_t, mask15s) = 0xFFE0FFE0FFE0FFE0ULL;
DECLARE_ASM_CONST(8, uint64_t, mask15g) = 0x03E003E003E003E0ULL;
DECLARE_ASM_CONST(8, uint64_t, mask15r) = 0x7C007C007C007C00ULL;
#define mask16b mask15b
DECLARE_ASM_CONST(8, uint64_t, mask16g) = 0x07E007E007E007E0ULL;
DECLARE_ASM_CONST(8, uint64_t, mask16r) = 0xF800F800F800F800ULL;
DECLARE_ASM_CONST(8, uint64_t, red_16mask) = 0x0000f8000000f800ULL;
DECLARE_ASM_CONST(8, uint64_t, green_16mask) = 0x000007e0000007e0ULL;
DECLARE_ASM_CONST(8, uint64_t, blue_16mask) = 0x0000001f0000001fULL;
DECLARE_ASM_CONST(8, uint64_t, red_15mask) = 0x00007c0000007c00ULL;
DECLARE_ASM_CONST(8, uint64_t, green_15mask) = 0x000003e0000003e0ULL;
DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
#endif /* ARCH_X86 */
#define RGB2YUV_SHIFT 8
#define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
#define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
@ -145,50 +105,9 @@ DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
#define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
#define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
//Note: We have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW + MMX2 one.
//plain C versions
#define COMPILE_TEMPLATE_MMX 0
#define COMPILE_TEMPLATE_MMX2 0
#define COMPILE_TEMPLATE_AMD3DNOW 0
#define COMPILE_TEMPLATE_SSE2 0
#define RENAME(a) a ## _C
#include "rgb2rgb_template.c"
#if ARCH_X86
//MMX versions
#undef RENAME
#undef COMPILE_TEMPLATE_MMX
#define COMPILE_TEMPLATE_MMX 1
#define RENAME(a) a ## _MMX
#include "rgb2rgb_template.c"
//MMX2 versions
#undef RENAME
#undef COMPILE_TEMPLATE_MMX2
#define COMPILE_TEMPLATE_MMX2 1
#define RENAME(a) a ## _MMX2
#include "rgb2rgb_template.c"
//SSE2 versions
#undef RENAME
#undef COMPILE_TEMPLATE_SSE2
#define COMPILE_TEMPLATE_SSE2 1
#define RENAME(a) a ## _SSE2
#include "rgb2rgb_template.c"
//3DNOW versions
#undef RENAME
#undef COMPILE_TEMPLATE_MMX2
#undef COMPILE_TEMPLATE_SSE2
#undef COMPILE_TEMPLATE_AMD3DNOW
#define COMPILE_TEMPLATE_MMX2 0
#define COMPILE_TEMPLATE_SSE2 1
#define COMPILE_TEMPLATE_AMD3DNOW 1
#define RENAME(a) a ## _3DNOW
#include "rgb2rgb_template.c"
#endif //ARCH_X86 || ARCH_X86_64
/*
RGB15->RGB16 original by Strepto/Astral
@ -199,18 +118,10 @@ DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
void sws_rgb2rgb_init(int flags)
{
#if HAVE_SSE2 || HAVE_MMX2 || HAVE_AMD3DNOW || HAVE_MMX
if (flags & SWS_CPU_CAPS_SSE2)
rgb2rgb_init_SSE2();
else if (flags & SWS_CPU_CAPS_MMX2)
rgb2rgb_init_MMX2();
else if (flags & SWS_CPU_CAPS_3DNOW)
rgb2rgb_init_3DNOW();
else if (flags & SWS_CPU_CAPS_MMX)
rgb2rgb_init_MMX();
else
rgb2rgb_init_c();
#if HAVE_MMX2 || HAVE_AMD3DNOW || HAVE_MMX
rgb2rgb_init_x86(flags);
#endif /* HAVE_MMX2 || HAVE_AMD3DNOW || HAVE_MMX */
rgb2rgb_init_C();
}
#if LIBSWSCALE_VERSION_MAJOR < 1

2
libswscale/rgb2rgb.h
View File

@ -168,4 +168,6 @@ extern void (*yuyvtoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const u
void sws_rgb2rgb_init(int flags);
void rgb2rgb_init_x86(int flags);
#endif /* SWSCALE_RGB2RGB_H */

2305
libswscale/rgb2rgb_template.c
View File

File diff suppressed because it is too large Load Diff

93
libswscale/swscale.c
View File

@ -122,63 +122,6 @@ add BGR4 output support
write special BGR->BGR scaler
*/
#if ARCH_X86
DECLARE_ASM_CONST(8, uint64_t, bF8)= 0xF8F8F8F8F8F8F8F8LL;
DECLARE_ASM_CONST(8, uint64_t, bFC)= 0xFCFCFCFCFCFCFCFCLL;
DECLARE_ASM_CONST(8, uint64_t, w10)= 0x0010001000100010LL;
DECLARE_ASM_CONST(8, uint64_t, w02)= 0x0002000200020002LL;
DECLARE_ASM_CONST(8, uint64_t, bm00001111)=0x00000000FFFFFFFFLL;
DECLARE_ASM_CONST(8, uint64_t, bm00000111)=0x0000000000FFFFFFLL;
DECLARE_ASM_CONST(8, uint64_t, bm11111000)=0xFFFFFFFFFF000000LL;
DECLARE_ASM_CONST(8, uint64_t, bm01010101)=0x00FF00FF00FF00FFLL;
const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
0x0103010301030103LL,
0x0200020002000200LL,};
const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
0x0602060206020602LL,
0x0004000400040004LL,};
DECLARE_ASM_CONST(8, uint64_t, b16Mask)= 0x001F001F001F001FLL;
DECLARE_ASM_CONST(8, uint64_t, g16Mask)= 0x07E007E007E007E0LL;
DECLARE_ASM_CONST(8, uint64_t, r16Mask)= 0xF800F800F800F800LL;
DECLARE_ASM_CONST(8, uint64_t, b15Mask)= 0x001F001F001F001FLL;
DECLARE_ASM_CONST(8, uint64_t, g15Mask)= 0x03E003E003E003E0LL;
DECLARE_ASM_CONST(8, uint64_t, r15Mask)= 0x7C007C007C007C00LL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24A) = 0x00FF0000FF0000FFLL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24B) = 0xFF0000FF0000FF00LL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24C) = 0x0000FF0000FF0000LL;
#ifdef FAST_BGR2YV12
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000000210041000DULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000FFEEFFDC0038ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00000038FFD2FFF8ULL;
#else
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000020E540830C8BULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000ED0FDAC23831ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00003831D0E6F6EAULL;
#endif /* FAST_BGR2YV12 */
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset) = 0x1010101010101010ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_w1111) = 0x0001000100010001ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY1Coeff) = 0x0C88000040870C88ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY2Coeff) = 0x20DE4087000020DEULL;
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY1Coeff) = 0x20DE0000408720DEULL;
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY2Coeff) = 0x0C88408700000C88ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toYOffset) = 0x0008400000084000ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUV)[2][4] = {
{0x38380000DAC83838ULL, 0xECFFDAC80000ECFFULL, 0xF6E40000D0E3F6E4ULL, 0x3838D0E300003838ULL},
{0xECFF0000DAC8ECFFULL, 0x3838DAC800003838ULL, 0x38380000D0E33838ULL, 0xF6E4D0E30000F6E4ULL},
};
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUVOffset)= 0x0040400000404000ULL;
#endif /* ARCH_X86 */
DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_4)[2][8]={
{ 1, 3, 1, 3, 1, 3, 1, 3, },
{ 2, 0, 2, 0, 2, 0, 2, 0, },
@ -1367,17 +1310,14 @@ static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uin
#define COMPILE_TEMPLATE_AMD3DNOW 0
#define COMPILE_TEMPLATE_ALTIVEC 0
#if COMPILE_C
#define RENAME(a) a ## _C
#include "swscale_template.c"
#endif
#if COMPILE_ALTIVEC
#undef RENAME
#undef COMPILE_TEMPLATE_ALTIVEC
#define COMPILE_TEMPLATE_ALTIVEC 1
#define RENAME(a) a ## _altivec
#include "swscale_template.c"
#include "ppc/swscale_template.c"
#endif
#if ARCH_X86
@ -1392,7 +1332,7 @@ static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uin
#define COMPILE_TEMPLATE_MMX2 0
#define COMPILE_TEMPLATE_AMD3DNOW 0
#define RENAME(a) a ## _MMX
#include "swscale_template.c"
#include "x86/swscale_template.c"
#endif
//MMX2 versions
@ -1405,7 +1345,7 @@ static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uin
#define COMPILE_TEMPLATE_MMX2 1
#define COMPILE_TEMPLATE_AMD3DNOW 0
#define RENAME(a) a ## _MMX2
#include "swscale_template.c"
#include "x86/swscale_template.c"
#endif
//3DNOW versions
@ -1418,44 +1358,36 @@ static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uin
#define COMPILE_TEMPLATE_MMX2 0
#define COMPILE_TEMPLATE_AMD3DNOW 1
#define RENAME(a) a ## _3DNow
#include "swscale_template.c"
#include "x86/swscale_template.c"
#endif
#endif //ARCH_X86
SwsFunc ff_getSwsFunc(SwsContext *c)
{
#if CONFIG_RUNTIME_CPUDETECT
int flags = c->flags;
sws_init_swScale_c(c);
#if CONFIG_RUNTIME_CPUDETECT
#if ARCH_X86
// ordered per speed fastest first
if (flags & SWS_CPU_CAPS_MMX2) {
if (c->flags & SWS_CPU_CAPS_MMX2) {
sws_init_swScale_MMX2(c);
return swScale_MMX2;
} else if (flags & SWS_CPU_CAPS_3DNOW) {
} else if (c->flags & SWS_CPU_CAPS_3DNOW) {
sws_init_swScale_3DNow(c);
return swScale_3DNow;
} else if (flags & SWS_CPU_CAPS_MMX) {
} else if (c->flags & SWS_CPU_CAPS_MMX) {
sws_init_swScale_MMX(c);
return swScale_MMX;
} else {
sws_init_swScale_C(c);
return swScale_C;
}
#else
#if COMPILE_ALTIVEC
if (flags & SWS_CPU_CAPS_ALTIVEC) {
if (c->flags & SWS_CPU_CAPS_ALTIVEC) {
sws_init_swScale_altivec(c);
return swScale_altivec;
} else {
sws_init_swScale_C(c);
return swScale_C;
}
#endif
sws_init_swScale_C(c);
return swScale_C;
#endif /* ARCH_X86 */
#else //CONFIG_RUNTIME_CPUDETECT
#if COMPILE_TEMPLATE_MMX2
@ -1470,11 +1402,10 @@ SwsFunc ff_getSwsFunc(SwsContext *c)
#elif COMPILE_TEMPLATE_ALTIVEC
sws_init_swScale_altivec(c);
return swScale_altivec;
#else
sws_init_swScale_C(c);
return swScale_C;
#endif
#endif //!CONFIG_RUNTIME_CPUDETECT
return swScale_c;
}
static void copyPlane(const uint8_t *src, int srcStride,

2417
libswscale/swscale_template.c
View File

File diff suppressed because it is too large Load Diff

137
libswscale/x86/rgb2rgb.c Normal file
View File

@ -0,0 +1,137 @@
/*
* software RGB to RGB converter
* pluralize by software PAL8 to RGB converter
* software YUV to YUV converter
* software YUV to RGB converter
* Written by Nick Kurshev.
* palette & YUV & runtime CPU stuff by Michael (michaelni@gmx.at)
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include <stdint.h>
#include "config.h"
#include "libavutil/x86_cpu.h"
#include "libavutil/bswap.h"
#include "libswscale/rgb2rgb.h"
#include "libswscale/swscale.h"
#include "libswscale/swscale_internal.h"
DECLARE_ASM_CONST(8, uint64_t, mmx_ff) = 0x00000000000000FFULL;
DECLARE_ASM_CONST(8, uint64_t, mmx_null) = 0x0000000000000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mmx_one) = 0xFFFFFFFFFFFFFFFFULL;
DECLARE_ASM_CONST(8, uint64_t, mask32b) = 0x000000FF000000FFULL;
DECLARE_ASM_CONST(8, uint64_t, mask32g) = 0x0000FF000000FF00ULL;
DECLARE_ASM_CONST(8, uint64_t, mask32r) = 0x00FF000000FF0000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask32a) = 0xFF000000FF000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask32) = 0x00FFFFFF00FFFFFFULL;
DECLARE_ASM_CONST(8, uint64_t, mask3216br) = 0x00F800F800F800F8ULL;
DECLARE_ASM_CONST(8, uint64_t, mask3216g) = 0x0000FC000000FC00ULL;
DECLARE_ASM_CONST(8, uint64_t, mask3215g) = 0x0000F8000000F800ULL;
DECLARE_ASM_CONST(8, uint64_t, mul3216) = 0x2000000420000004ULL;
DECLARE_ASM_CONST(8, uint64_t, mul3215) = 0x2000000820000008ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24b) = 0x00FF0000FF0000FFULL;
DECLARE_ASM_CONST(8, uint64_t, mask24g) = 0xFF0000FF0000FF00ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24r) = 0x0000FF0000FF0000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24l) = 0x0000000000FFFFFFULL;
DECLARE_ASM_CONST(8, uint64_t, mask24h) = 0x0000FFFFFF000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24hh) = 0xffff000000000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24hhh) = 0xffffffff00000000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask24hhhh) = 0xffffffffffff0000ULL;
DECLARE_ASM_CONST(8, uint64_t, mask15b) = 0x001F001F001F001FULL; /* 00000000 00011111 xxB */
DECLARE_ASM_CONST(8, uint64_t, mask15rg) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000 RGx */
DECLARE_ASM_CONST(8, uint64_t, mask15s) = 0xFFE0FFE0FFE0FFE0ULL;
DECLARE_ASM_CONST(8, uint64_t, mask15g) = 0x03E003E003E003E0ULL;
DECLARE_ASM_CONST(8, uint64_t, mask15r) = 0x7C007C007C007C00ULL;
#define mask16b mask15b
DECLARE_ASM_CONST(8, uint64_t, mask16g) = 0x07E007E007E007E0ULL;
DECLARE_ASM_CONST(8, uint64_t, mask16r) = 0xF800F800F800F800ULL;
DECLARE_ASM_CONST(8, uint64_t, red_16mask) = 0x0000f8000000f800ULL;
DECLARE_ASM_CONST(8, uint64_t, green_16mask) = 0x000007e0000007e0ULL;
DECLARE_ASM_CONST(8, uint64_t, blue_16mask) = 0x0000001f0000001fULL;
DECLARE_ASM_CONST(8, uint64_t, red_15mask) = 0x00007c0000007c00ULL;
DECLARE_ASM_CONST(8, uint64_t, green_15mask) = 0x000003e0000003e0ULL;
DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
#define RGB2YUV_SHIFT 8
#define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
#define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
#define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
#define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
#define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
#define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
#define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
#define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
#define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
//Note: We have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW + MMX2 one.
#define COMPILE_TEMPLATE_MMX2 0
#define COMPILE_TEMPLATE_AMD3DNOW 0
#define COMPILE_TEMPLATE_SSE2 0
//MMX versions
#undef RENAME
#define RENAME(a) a ## _MMX
#include "rgb2rgb_template.c"
//MMX2 versions
#undef RENAME
#undef COMPILE_TEMPLATE_MMX2
#define COMPILE_TEMPLATE_MMX2 1
#define RENAME(a) a ## _MMX2
#include "rgb2rgb_template.c"
//SSE2 versions
#undef RENAME
#undef COMPILE_TEMPLATE_SSE2
#define COMPILE_TEMPLATE_SSE2 1
#define RENAME(a) a ## _SSE2
#include "rgb2rgb_template.c"
//3DNOW versions
#undef RENAME
#undef COMPILE_TEMPLATE_MMX2
#undef COMPILE_TEMPLATE_SSE2
#undef COMPILE_TEMPLATE_AMD3DNOW
#define COMPILE_TEMPLATE_MMX2 0
#define COMPILE_TEMPLATE_SSE2 1
#define COMPILE_TEMPLATE_AMD3DNOW 1
#define RENAME(a) a ## _3DNOW
#include "rgb2rgb_template.c"
/*
RGB15->RGB16 original by Strepto/Astral
ported to gcc & bugfixed : A'rpi
MMX2, 3DNOW optimization by Nick Kurshev
32-bit C version, and and&add trick by Michael Niedermayer
*/
void rgb2rgb_init_x86(int flags)
{
#if HAVE_MMX2 || HAVE_AMD3DNOW || HAVE_MMX
if (flags & SWS_CPU_CAPS_SSE2)
rgb2rgb_init_SSE2();
else if (flags & SWS_CPU_CAPS_MMX2)
rgb2rgb_init_MMX2();
else if (flags & SWS_CPU_CAPS_3DNOW)
rgb2rgb_init_3DNOW();
else if (flags & SWS_CPU_CAPS_MMX)
rgb2rgb_init_MMX();
#endif /* HAVE_MMX2 || HAVE_AMD3DNOW || HAVE_MMX */
}

2646
libswscale/x86/rgb2rgb_template.c Normal file
View File

File diff suppressed because it is too large Load Diff

2825
libswscale/x86/swscale_template.c Normal file
View File

File diff suppressed because it is too large Load Diff

79
libswscale/x86/swscale_template.h Normal file
View File

@ -0,0 +1,79 @@
/*
* Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#ifndef SWSCALE_X86_SWSCALE_TEMPLATE_H
#define SWSCALE_X86_SWSCALE_TEMPLATE_H
DECLARE_ASM_CONST(8, uint64_t, bF8)= 0xF8F8F8F8F8F8F8F8LL;
DECLARE_ASM_CONST(8, uint64_t, bFC)= 0xFCFCFCFCFCFCFCFCLL;
DECLARE_ASM_CONST(8, uint64_t, w10)= 0x0010001000100010LL;
DECLARE_ASM_CONST(8, uint64_t, w02)= 0x0002000200020002LL;
DECLARE_ASM_CONST(8, uint64_t, bm00001111)=0x00000000FFFFFFFFLL;
DECLARE_ASM_CONST(8, uint64_t, bm00000111)=0x0000000000FFFFFFLL;
DECLARE_ASM_CONST(8, uint64_t, bm11111000)=0xFFFFFFFFFF000000LL;
DECLARE_ASM_CONST(8, uint64_t, bm01010101)=0x00FF00FF00FF00FFLL;
const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
0x0103010301030103LL,
0x0200020002000200LL,};
const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
0x0602060206020602LL,
0x0004000400040004LL,};
DECLARE_ASM_CONST(8, uint64_t, b16Mask)= 0x001F001F001F001FLL;
DECLARE_ASM_CONST(8, uint64_t, g16Mask)= 0x07E007E007E007E0LL;
DECLARE_ASM_CONST(8, uint64_t, r16Mask)= 0xF800F800F800F800LL;
DECLARE_ASM_CONST(8, uint64_t, b15Mask)= 0x001F001F001F001FLL;
DECLARE_ASM_CONST(8, uint64_t, g15Mask)= 0x03E003E003E003E0LL;
DECLARE_ASM_CONST(8, uint64_t, r15Mask)= 0x7C007C007C007C00LL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24A) = 0x00FF0000FF0000FFLL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24B) = 0xFF0000FF0000FF00LL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24C) = 0x0000FF0000FF0000LL;
#ifdef FAST_BGR2YV12
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000000210041000DULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000FFEEFFDC0038ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00000038FFD2FFF8ULL;
#else
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000020E540830C8BULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000ED0FDAC23831ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00003831D0E6F6EAULL;
#endif /* FAST_BGR2YV12 */
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset) = 0x1010101010101010ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_w1111) = 0x0001000100010001ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY1Coeff) = 0x0C88000040870C88ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY2Coeff) = 0x20DE4087000020DEULL;
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY1Coeff) = 0x20DE0000408720DEULL;
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY2Coeff) = 0x0C88408700000C88ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toYOffset) = 0x0008400000084000ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUV)[2][4] = {
{0x38380000DAC83838ULL, 0xECFFDAC80000ECFFULL, 0xF6E40000D0E3F6E4ULL, 0x3838D0E300003838ULL},
{0xECFF0000DAC8ECFFULL, 0x3838DAC800003838ULL, 0x38380000D0E33838ULL, 0xF6E4D0E30000F6E4ULL},
};
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUVOffset)= 0x0040400000404000ULL;
#endif /* SWSCALE_X86_SWSCALE_TEMPLATE_H */