virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-26 19:01:44 +02:00
Code Issues Releases Activity

pseudo-simd add_bytes and diff_bytes

Browse Source 
2x faster than scalar in 32bit, 4x faster in 64bit (as opposed to 8x in mmx)

Originally committed as revision 12165 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Loren Merritt 2008-02-21 07:54:46 +00:00
parent 4a9ca0a279
commit 469bd7b150
1 changed files with 20 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
libavcodec/dsputil.c
View File

@ -50,6 +50,9 @@ void ff_add_png_paeth_prediction(uint8_t *dst, uint8_t *src, uint8_t *top, int w
uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP] = {0, }; uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP] = {0, };
uint32_t ff_squareTbl[512] = {0, }; uint32_t ff_squareTbl[512] = {0, };
static const unsigned long pb_7f = 0x7f7f7f7f7f7f7f7fUL;
static const unsigned long pb_80 = 0x8080808080808080UL;
const uint8_t ff_zigzag_direct[64] = { const uint8_t ff_zigzag_direct[64] = {
0, 1, 8, 16, 9, 2, 3, 10, 0, 1, 8, 16, 9, 2, 3, 10,
17, 24, 32, 25, 18, 11, 4, 5, 17, 24, 32, 25, 18, 11, 4, 5,
@ -3276,34 +3279,31 @@ static void clear_blocks_c(DCTELEM *blocks)
} }
static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){ static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){
int i; long i;
for(i=0; i+7<w; i+=8){ for(i=0; i<=w-sizeof(long); i+=sizeof(long)){
dst[i+0] += src[i+0]; long a = *(long*)(src+i);
dst[i+1] += src[i+1]; long b = *(long*)(dst+i);
dst[i+2] += src[i+2]; *(long*)(dst+i) = ((a&pb_7f) + (b&pb_7f)) ^ ((a^b)&pb_80);
dst[i+3] += src[i+3];
dst[i+4] += src[i+4];
dst[i+5] += src[i+5];
dst[i+6] += src[i+6];
dst[i+7] += src[i+7];
} }
for(; i<w; i++) for(; i<w; i++)
dst[i+0] += src[i+0]; dst[i+0] += src[i+0];
} }
static void add_bytes_l2_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){ static void add_bytes_l2_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
int i; long i;
for(i=0; i<=w-sizeof(long); i+=sizeof(long)){ for(i=0; i<=w-sizeof(long); i+=sizeof(long)){
long a = *(long*)(src1+i); long a = *(long*)(src1+i);
long b = *(long*)(src2+i); long b = *(long*)(src2+i);
*(long*)(dst+i) = ((a&0x7f7f7f7f7f7f7f7fL) + (b&0x7f7f7f7f7f7f7f7fL)) ^ ((a^b)&0x8080808080808080L); *(long*)(dst+i) = ((a&pb_7f) + (b&pb_7f)) ^ ((a^b)&pb_80);
} }
for(; i<w; i++) for(; i<w; i++)
dst[i] = src1[i]+src2[i]; dst[i] = src1[i]+src2[i];
} }
static void diff_bytes_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){ static void diff_bytes_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
int i; long i;
#ifndef HAVE_FAST_UNALIGNED
if((long)src2 & (sizeof(long)-1)){
for(i=0; i+7<w; i+=8){ for(i=0; i+7<w; i+=8){
dst[i+0] = src1[i+0]-src2[i+0]; dst[i+0] = src1[i+0]-src2[i+0];
dst[i+1] = src1[i+1]-src2[i+1]; dst[i+1] = src1[i+1]-src2[i+1];
@ -3314,6 +3314,13 @@ static void diff_bytes_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
dst[i+6] = src1[i+6]-src2[i+6]; dst[i+6] = src1[i+6]-src2[i+6];
dst[i+7] = src1[i+7]-src2[i+7]; dst[i+7] = src1[i+7]-src2[i+7];
} }
}else
#endif
for(i=0; i<=w-sizeof(long); i+=sizeof(long)){
long a = *(long*)(src1+i);
long b = *(long*)(src2+i);
*(long*)(dst+i) = ((a|pb_80) - (b&pb_7f)) ^ ((a^b^pb_80)&pb_80);
}
for(; i<w; i++) for(; i<w; i++)
dst[i+0] = src1[i+0]-src2[i+0]; dst[i+0] = src1[i+0]-src2[i+0];
} }