Mercurial > sdl-ios-xcode

comparison src/audio/SDL_mixer_MMX_VC.c @ 1662:782fd950bd46 SDL-1.3

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Revamp of the video system in progress - adding support for multiple displays, multiple windows, and a full video mode selection API. WARNING: None of the video drivers have been updated for the new API yet! The API is still under design and very fluid. The code is now run through a consistent indent format: indent -i4 -nut -nsc -br -ce The headers are being converted to automatically generate doxygen documentation.
author Sam Lantinga <slouken@libsdl.org>
date Sun, 28 May 2006 13:04:16 +0000
parents e3242177fe4a
children 4da1ee79c9af
comparison
equal deleted inserted replaced
1661:281d3f4870e5 1662:782fd950bd46
35 35
36 //////////////////////////////////////////////// 36 ////////////////////////////////////////////////
37 // Mixing for 16 bit signed buffers 37 // Mixing for 16 bit signed buffers
38 //////////////////////////////////////////////// 38 ////////////////////////////////////////////////
39 39
40 void SDL_MixAudio_MMX_S16_VC(char* dst,char* src,unsigned int nSize,int volume) 40 void
41 SDL_MixAudio_MMX_S16_VC (char *dst, char *src, unsigned int nSize, int volume)
41 { 42 {
42 __asm 43 __asm {
43 {
44 44
45 push edi 45 push edi push esi push ebx mov edi, dst // edi = dst
46 push esi 46 mov esi, src // esi = src
47 push ebx 47 mov eax, volume // eax = volume
48 48 mov ebx, nSize // ebx = size
49 mov edi, dst // edi = dst 49 shr ebx, 4 // process 16 bytes per iteration = 8 samples
50 mov esi, src // esi = src 50 jz endS16 pxor mm0, mm0 movd mm0, eax //%%eax,%%mm0
51 mov eax, volume // eax = volume 51 movq mm1, mm0 //%%mm0,%%mm1
52 mov ebx, nSize // ebx = size 52 psllq mm0, 16 //$16,%%mm0
53 shr ebx, 4 // process 16 bytes per iteration = 8 samples 53 por mm0, mm1 //%%mm1,%%mm0
54 jz endS16 54 psllq mm0, 16 //$16,%%mm0
55 55 por mm0, mm1 //%%mm1,%%mm0
56 pxor mm0, mm0 56 psllq mm0, 16 //$16,%%mm0
57 movd mm0, eax //%%eax,%%mm0 57 por mm0, mm1 //%%mm1,%%mm0 // mm0 = vol|vol|vol|vol
58 movq mm1, mm0 //%%mm0,%%mm1 58 #ifndef __WATCOMC__
59 psllq mm0, 16 //$16,%%mm0 59 align 16
60 por mm0, mm1 //%%mm1,%%mm0 60 #endif
61 psllq mm0, 16 //$16,%%mm0 61 mixloopS16:movq mm1,[esi] //(%%esi),%%mm1\n" // mm1 = a|b|c|d
62 por mm0, mm1 //%%mm1,%%mm0 62 movq mm2, mm1 //%%mm1,%%mm2\n" // mm2 = a|b|c|d
63 psllq mm0, 16 //$16,%%mm0 63 movq mm4,[esi + 8] //8(%%esi),%%mm4\n" // mm4 = e|f|g|h
64 por mm0, mm1 //%%mm1,%%mm0 // mm0 = vol|vol|vol|vol 64 // pre charger le buffer dst dans mm7
65 65 movq mm7,[edi] //(%%edi),%%mm7\n" // mm7 = dst[0]"
66 #ifndef __WATCOMC__ 66 // multiplier par le volume
67 align 16 67 pmullw mm1, mm0 //%%mm0,%%mm1\n" // mm1 = l(a*v)|l(b*v)|l(c*v)|l(d*v)
68 #endif 68 pmulhw mm2, mm0 //%%mm0,%%mm2\n" // mm2 = h(a*v)|h(b*v)|h(c*v)|h(d*v)
69 mixloopS16: 69 movq mm5, mm4 //%%mm4,%%mm5\n" // mm5 = e|f|g|h
70 movq mm1, [esi] //(%%esi),%%mm1\n" // mm1 = a|b|c|d 70 pmullw mm4, mm0 //%%mm0,%%mm4\n" // mm4 = l(e*v)|l(f*v)|l(g*v)|l(h*v)
71 movq mm2, mm1 //%%mm1,%%mm2\n" // mm2 = a|b|c|d 71 pmulhw mm5, mm0 //%%mm0,%%mm5\n" // mm5 = h(e*v)|h(f*v)|h(g*v)|h(h*v)
72 movq mm4, [esi + 8] //8(%%esi),%%mm4\n" // mm4 = e|f|g|h 72 movq mm3, mm1 //%%mm1,%%mm3\n" // mm3 = l(a*v)|l(b*v)|l(c*v)|l(d*v)
73 // pre charger le buffer dst dans mm7 73 punpckhwd mm1, mm2 //%%mm2,%%mm1\n" // mm1 = a*v|b*v
74 movq mm7, [edi] //(%%edi),%%mm7\n" // mm7 = dst[0]" 74 movq mm6, mm4 //%%mm4,%%mm6\n" // mm6 = l(e*v)|l(f*v)|l(g*v)|l(h*v)
75 // multiplier par le volume 75 punpcklwd mm3, mm2 //%%mm2,%%mm3\n" // mm3 = c*v|d*v
76 pmullw mm1, mm0 //%%mm0,%%mm1\n" // mm1 = l(a*v)|l(b*v)|l(c*v)|l(d*v) 76 punpckhwd mm4, mm5 //%%mm5,%%mm4\n" // mm4 = e*f|f*v
77 pmulhw mm2, mm0 //%%mm0,%%mm2\n" // mm2 = h(a*v)|h(b*v)|h(c*v)|h(d*v) 77 punpcklwd mm6, mm5 //%%mm5,%%mm6\n" // mm6 = g*v|h*v
78 movq mm5, mm4 //%%mm4,%%mm5\n" // mm5 = e|f|g|h 78 // pre charger le buffer dst dans mm5
79 pmullw mm4, mm0 //%%mm0,%%mm4\n" // mm4 = l(e*v)|l(f*v)|l(g*v)|l(h*v) 79 movq mm5,[edi + 8] //8(%%edi),%%mm5\n" // mm5 = dst[1]
80 pmulhw mm5, mm0 //%%mm0,%%mm5\n" // mm5 = h(e*v)|h(f*v)|h(g*v)|h(h*v) 80 // diviser par 128
81 movq mm3, mm1 //%%mm1,%%mm3\n" // mm3 = l(a*v)|l(b*v)|l(c*v)|l(d*v) 81 psrad mm1, 7 //$7,%%mm1\n" // mm1 = a*v/128|b*v/128 , 128 = SDL_MIX_MAXVOLUME
82 punpckhwd mm1, mm2 //%%mm2,%%mm1\n" // mm1 = a*v|b*v 82 add esi, 16 //$16,%%esi\n"
83 movq mm6, mm4 //%%mm4,%%mm6\n" // mm6 = l(e*v)|l(f*v)|l(g*v)|l(h*v) 83 psrad mm3, 7 //$7,%%mm3\n" // mm3 = c*v/128|d*v/128
84 punpcklwd mm3, mm2 //%%mm2,%%mm3\n" // mm3 = c*v|d*v 84 psrad mm4, 7 //$7,%%mm4\n" // mm4 = e*v/128|f*v/128
85 punpckhwd mm4, mm5 //%%mm5,%%mm4\n" // mm4 = e*f|f*v 85 // mm1 = le sample avec le volume modifie
86 punpcklwd mm6, mm5 //%%mm5,%%mm6\n" // mm6 = g*v|h*v 86 packssdw mm3, mm1 //%%mm1,%%mm3\n" // mm3 = s(a*v|b*v|c*v|d*v)
87 // pre charger le buffer dst dans mm5 87 psrad mm6, 7 //$7,%%mm6\n" // mm6= g*v/128|h*v/128
88 movq mm5, [edi + 8] //8(%%edi),%%mm5\n" // mm5 = dst[1] 88 paddsw mm3, mm7 //%%mm7,%%mm3\n" // mm3 = adjust_volume(src)+dst
89 // diviser par 128 89 // mm4 = le sample avec le volume modifie
90 psrad mm1, 7 //$7,%%mm1\n" // mm1 = a*v/128|b*v/128 , 128 = SDL_MIX_MAXVOLUME 90 packssdw mm6, mm4 //%%mm4,%%mm6\n" // mm6 = s(e*v|f*v|g*v|h*v)
91 add esi, 16 //$16,%%esi\n" 91 movq[edi], mm3 //%%mm3,(%%edi)\n"
92 psrad mm3, 7 //$7,%%mm3\n" // mm3 = c*v/128|d*v/128 92 paddsw mm6, mm5 //%%mm5,%%mm6\n" // mm6 = adjust_volume(src)+dst
93 psrad mm4, 7 //$7,%%mm4\n" // mm4 = e*v/128|f*v/128 93 movq[edi + 8], mm6 //%%mm6,8(%%edi)\n"
94 // mm1 = le sample avec le volume modifie 94 add edi, 16 //$16,%%edi\n"
95 packssdw mm3, mm1 //%%mm1,%%mm3\n" // mm3 = s(a*v|b*v|c*v|d*v) 95 dec ebx //%%ebx\n"
96 psrad mm6, 7 //$7,%%mm6\n" // mm6= g*v/128|h*v/128 96 jnz mixloopS16 ends16:emms pop ebx pop esi pop edi}
97 paddsw mm3, mm7 //%%mm7,%%mm3\n" // mm3 = adjust_volume(src)+dst
98 // mm4 = le sample avec le volume modifie
99 packssdw mm6, mm4 //%%mm4,%%mm6\n" // mm6 = s(e*v|f*v|g*v|h*v)
100 movq [edi], mm3 //%%mm3,(%%edi)\n"
101 paddsw mm6, mm5 //%%mm5,%%mm6\n" // mm6 = adjust_volume(src)+dst
102 movq [edi + 8], mm6 //%%mm6,8(%%edi)\n"
103 add edi, 16 //$16,%%edi\n"
104 dec ebx //%%ebx\n"
105 jnz mixloopS16
106
107 ends16:
108 emms
109
110 pop ebx
111 pop esi
112 pop edi
113 }
114 97
115 } 98 }
116 99
117 //////////////////////////////////////////////// 100 ////////////////////////////////////////////////
118 // Mixing for 8 bit signed buffers 101 // Mixing for 8 bit signed buffers
119 //////////////////////////////////////////////// 102 ////////////////////////////////////////////////
120 103
121 void SDL_MixAudio_MMX_S8_VC(char* dst,char* src,unsigned int nSize,int volume) 104 void
105 SDL_MixAudio_MMX_S8_VC (char *dst, char *src, unsigned int nSize, int volume)
122 { 106 {
123 _asm 107 _asm {
124 {
125 108
126 push edi 109 push edi push esi push ebx mov edi, dst //movl %0,%%edi // edi = dst
127 push esi 110 mov esi, src //%1,%%esi // esi = src
128 push ebx 111 mov eax, volume //%3,%%eax // eax = volume
129 112 movd mm0, eax //%%eax,%%mm0
130 mov edi, dst //movl %0,%%edi // edi = dst 113 movq mm1, mm0 //%%mm0,%%mm1
131 mov esi, src //%1,%%esi // esi = src 114 psllq mm0, 16 //$16,%%mm0
132 mov eax, volume //%3,%%eax // eax = volume 115 por mm0, mm1 //%%mm1,%%mm0
133 116 psllq mm0, 16 //$16,%%mm0
134 movd mm0, eax //%%eax,%%mm0 117 por mm0, mm1 //%%mm1,%%mm0
135 movq mm1, mm0 //%%mm0,%%mm1 118 psllq mm0, 16 //$16,%%mm0
136 psllq mm0, 16 //$16,%%mm0 119 por mm0, mm1 //%%mm1,%%mm0
137 por mm0, mm1 //%%mm1,%%mm0 120 mov ebx, nSize //%2,%%ebx // ebx = size
138 psllq mm0, 16 //$16,%%mm0 121 shr ebx, 3 //$3,%%ebx // process 8 bytes per iteration = 8 samples
139 por mm0, mm1 //%%mm1,%%mm0 122 cmp ebx, 0 //$0,%%ebx
140 psllq mm0, 16 //$16,%%mm0 123 je endS8
141 por mm0, mm1 //%%mm1,%%mm0 124 #ifndef __WATCOMC__
142 125 align 16
143 mov ebx, nSize //%2,%%ebx // ebx = size 126 #endif
144 shr ebx, 3 //$3,%%ebx // process 8 bytes per iteration = 8 samples 127 mixloopS8:pxor mm2, mm2 //%%mm2,%%mm2 // mm2 = 0
145 cmp ebx, 0 //$0,%%ebx 128 movq mm1,[esi] //(%%esi),%%mm1 // mm1 = a|b|c|d|e|f|g|h
146 je endS8 129 movq mm3, mm1 //%%mm1,%%mm3 // mm3 = a|b|c|d|e|f|g|h
147 130 // on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0
148 #ifndef __WATCOMC__ 131 pcmpgtb mm2, mm1 //%%mm1,%%mm2 // mm2 = 11111111|00000000|00000000....
149 align 16 132 punpckhbw mm1, mm2 //%%mm2,%%mm1 // mm1 = 0|a|0|b|0|c|0|d
150 #endif 133 punpcklbw mm3, mm2 //%%mm2,%%mm3 // mm3 = 0|e|0|f|0|g|0|h
151 mixloopS8: 134 movq mm2,[edi] //(%%edi),%%mm2 // mm2 = destination
152 pxor mm2, mm2 //%%mm2,%%mm2 // mm2 = 0 135 pmullw mm1, mm0 //%%mm0,%%mm1 // mm1 = v*a|v*b|v*c|v*d
153 movq mm1, [esi] //(%%esi),%%mm1 // mm1 = a|b|c|d|e|f|g|h 136 add esi, 8 //$8,%%esi
154 movq mm3, mm1 //%%mm1,%%mm3 // mm3 = a|b|c|d|e|f|g|h 137 pmullw mm3, mm0 //%%mm0,%%mm3 // mm3 = v*e|v*f|v*g|v*h
155 // on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0 138 psraw mm1, 7 //$7,%%mm1 // mm1 = v*a/128|v*b/128|v*c/128|v*d/128
156 pcmpgtb mm2, mm1 //%%mm1,%%mm2 // mm2 = 11111111|00000000|00000000.... 139 psraw mm3, 7 //$7,%%mm3 // mm3 = v*e/128|v*f/128|v*g/128|v*h/128
157 punpckhbw mm1, mm2 //%%mm2,%%mm1 // mm1 = 0|a|0|b|0|c|0|d 140 packsswb mm3, mm1 //%%mm1,%%mm3 // mm1 = v*a/128|v*b/128|v*c/128|v*d/128|v*e/128|v*f/128|v*g/128|v*h/128
158 punpcklbw mm3, mm2 //%%mm2,%%mm3 // mm3 = 0|e|0|f|0|g|0|h 141 paddsb mm3, mm2 //%%mm2,%%mm3 // add to destination buffer
159 movq mm2, [edi] //(%%edi),%%mm2 // mm2 = destination 142 movq[edi], mm3 //%%mm3,(%%edi) // store back to ram
160 pmullw mm1, mm0 //%%mm0,%%mm1 // mm1 = v*a|v*b|v*c|v*d 143 add edi, 8 //$8,%%edi
161 add esi, 8 //$8,%%esi 144 dec ebx //%%ebx
162 pmullw mm3, mm0 //%%mm0,%%mm3 // mm3 = v*e|v*f|v*g|v*h 145 jnz mixloopS8 endS8:emms pop ebx pop esi pop edi}
163 psraw mm1, 7 //$7,%%mm1 // mm1 = v*a/128|v*b/128|v*c/128|v*d/128
164 psraw mm3, 7 //$7,%%mm3 // mm3 = v*e/128|v*f/128|v*g/128|v*h/128
165 packsswb mm3, mm1 //%%mm1,%%mm3 // mm1 = v*a/128|v*b/128|v*c/128|v*d/128|v*e/128|v*f/128|v*g/128|v*h/128
166 paddsb mm3, mm2 //%%mm2,%%mm3 // add to destination buffer
167 movq [edi], mm3 //%%mm3,(%%edi) // store back to ram
168 add edi, 8 //$8,%%edi
169 dec ebx //%%ebx
170 jnz mixloopS8
171
172 endS8:
173 emms
174
175 pop ebx
176 pop esi
177 pop edi
178 }
179 } 146 }
180 147
181 #endif /* SDL_ASSEMBLY_ROUTINES */ 148 #endif /* SDL_ASSEMBLY_ROUTINES */
149 /* vi: set ts=4 sw=4 expandtab: */