Mercurial > sdl-ios-xcode

comparison src/audio/SDL_mixer_MMX_VC.c @ 1049:b4117292e587

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Mixer: Visual C MMX asm code now aligns loops to 16 bytes. This patch came from Stephane Marchesin's codefork, but I don't know who wrote it originally.
author Ryan C. Gordon <icculus@icculus.org>
date Sun, 17 Apr 2005 10:27:00 +0000
parents 72ef7ce609ef
children ef03e8789d59
comparison
equal deleted inserted replaced
1048:a78acdd4967e 1049:b4117292e587
17 17
18 void SDL_MixAudio_MMX_S16_VC(char* dst,char* src,unsigned int nSize,int volume) 18 void SDL_MixAudio_MMX_S16_VC(char* dst,char* src,unsigned int nSize,int volume)
19 { 19 {
20 __asm 20 __asm
21 { 21 {
22 align 16
23 22
24 push edi 23 push edi
25 push esi 24 push esi
26 push ebx 25 push ebx
27 26
40 psllq mm0, 16 //$16,%%mm0 39 psllq mm0, 16 //$16,%%mm0
41 por mm0, mm1 //%%mm1,%%mm0 40 por mm0, mm1 //%%mm1,%%mm0
42 psllq mm0, 16 //$16,%%mm0 41 psllq mm0, 16 //$16,%%mm0
43 por mm0, mm1 //%%mm1,%%mm0 // mm0 = vol|vol|vol|vol 42 por mm0, mm1 //%%mm1,%%mm0 // mm0 = vol|vol|vol|vol
44 43
44 align 16
45 mixloopS16: 45 mixloopS16:
46 movq mm1, [esi] //(%%esi),%%mm1\n" // mm1 = a|b|c|d 46 movq mm1, [esi] //(%%esi),%%mm1\n" // mm1 = a|b|c|d
47 movq mm2, mm1 //%%mm1,%%mm2\n" // mm2 = a|b|c|d 47 movq mm2, mm1 //%%mm1,%%mm2\n" // mm2 = a|b|c|d
48 movq mm4, [esi + 8] //8(%%esi),%%mm4\n" // mm4 = e|f|g|h 48 movq mm4, [esi + 8] //8(%%esi),%%mm4\n" // mm4 = e|f|g|h
49 // pre charger le buffer dst dans mm7 49 // pre charger le buffer dst dans mm7
96 96
97 void SDL_MixAudio_MMX_S8_VC(char* dst,char* src,unsigned int nSize,int volume) 97 void SDL_MixAudio_MMX_S8_VC(char* dst,char* src,unsigned int nSize,int volume)
98 { 98 {
99 _asm 99 _asm
100 { 100 {
101 align 16
102 101
103 push edi 102 push edi
104 push esi 103 push esi
105 push ebx 104 push ebx
106 105
120 mov ebx, nSize //%2,%%ebx // ebx = size 119 mov ebx, nSize //%2,%%ebx // ebx = size
121 shr ebx, 3 //$3,%%ebx // process 8 bytes per iteration = 8 samples 120 shr ebx, 3 //$3,%%ebx // process 8 bytes per iteration = 8 samples
122 cmp ebx, 0 //$0,%%ebx 121 cmp ebx, 0 //$0,%%ebx
123 je endS8 122 je endS8
124 123
124 align 16
125 mixloopS8: 125 mixloopS8:
126 pxor mm2, mm2 //%%mm2,%%mm2 // mm2 = 0 126 pxor mm2, mm2 //%%mm2,%%mm2 // mm2 = 0
127 movq mm1, [esi] //(%%esi),%%mm1 // mm1 = a|b|c|d|e|f|g|h 127 movq mm1, [esi] //(%%esi),%%mm1 // mm1 = a|b|c|d|e|f|g|h
128 movq mm3, mm1 //%%mm1,%%mm3 // mm3 = a|b|c|d|e|f|g|h 128 movq mm3, mm1 //%%mm1,%%mm3 // mm3 = a|b|c|d|e|f|g|h
129 // on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0 129 // on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0