ffmpeg/FFmpeg
1
0
Fork 0
mirror of https://git.ffmpeg.org/ffmpeg.git synced 2024-09-20 13:26:39 +00:00
Code Issues Packages Projects Releases Wiki Activity

Rewrite put_signed_pixels_clamped_mmx() to eliminate mmx.h from dsputil_mmx.c.

Browse Source 
Originally committed as revision 18319 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Alex Converse 2009-04-02 21:02:42 +00:00
parent dd3bbb4903
commit 3dd6531208
1 changed files with 30 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

42
libavcodec/x86/dsputil_mmx.c
View File

@ -28,7 +28,6 @@
#include "libavcodec/mpegvideo.h" #include "libavcodec/mpegvideo.h"
#include "libavcodec/simple_idct.h" #include "libavcodec/simple_idct.h"
#include "dsputil_mmx.h" #include "dsputil_mmx.h"
#include "mmx.h"
#include "vp3dsp_mmx.h" #include "vp3dsp_mmx.h"
#include "vp3dsp_sse2.h" #include "vp3dsp_sse2.h"
#include "vp6dsp_mmx.h" #include "vp6dsp_mmx.h"
@ -273,22 +272,41 @@ void put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size
:"memory"); :"memory");
} }
static DECLARE_ALIGNED_8(const unsigned char, vector128[8]) = DECLARE_ALIGNED_8(const unsigned char, ff_vector128[8]) =
{ 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80 }; { 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80 };
#define put_signed_pixels_clamped_mmx_half(off) \
"movq "#off"(%2), %%mm1 \n\t"\
"movq 16+"#off"(%2), %%mm2 \n\t"\
"movq 32+"#off"(%2), %%mm3 \n\t"\
"movq 48+"#off"(%2), %%mm4 \n\t"\
"packsswb 8+"#off"(%2), %%mm1 \n\t"\
"packsswb 24+"#off"(%2), %%mm2 \n\t"\
"packsswb 40+"#off"(%2), %%mm3 \n\t"\
"packsswb 56+"#off"(%2), %%mm4 \n\t"\
"paddb %%mm0, %%mm1 \n\t"\
"paddb %%mm0, %%mm2 \n\t"\
"paddb %%mm0, %%mm3 \n\t"\
"paddb %%mm0, %%mm4 \n\t"\
"movq %%mm1, (%0) \n\t"\
"movq %%mm2, (%0, %3) \n\t"\
"movq %%mm3, (%0, %3, 2) \n\t"\
"movq %%mm4, (%0, %1) \n\t"
void put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size) void put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size)
{ {
int i; x86_reg line_skip = line_size;
x86_reg av_uninit(line_skip3);
movq_m2r(*vector128, mm1); __asm__ volatile (
for (i = 0; i < 8; i++) { "movq "MANGLE(ff_vector128)", %%mm0 \n\t"
movq_m2r(*(block), mm0); "lea (%3, %3, 2), %1 \n\t"
packsswb_m2r(*(block + 4), mm0); put_signed_pixels_clamped_mmx_half(0)
block += 8; "lea (%0, %3, 4), %0 \n\t"
paddb_r2r(mm1, mm0); put_signed_pixels_clamped_mmx_half(64)
movq_r2m(mm0, *pixels); :"+r" (pixels), "+r" (line_skip3)
pixels += line_size; :"r" (block), "r"(line_skip)
} :"memory");
} }
void add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size) void add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size)