1 /*
2 SDL - Simple DirectMedia Layer
3 Copyright (C) 1997-2006 Sam Lantinga
4
5 This library is free software; you can redistribute it and/or
6 modify it under the terms of the GNU Lesser General Public
7 License as published by the Free Software Foundation; either
8 version 2.1 of the License, or (at your option) any later version.
9
10 This library is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 Lesser General Public License for more details.
14
15 You should have received a copy of the GNU Lesser General Public
16 License along with this library; if not, write to the Free Software
17 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
18
19 Sam Lantinga
20 slouken@libsdl.org
21 */
22 #include "SDL_config.h"
23
24 #include "SDL_mixer_MMX_VC.h"
25
26 #if ((defined(_MSC_VER) && defined(_M_IX86)) || defined(__WATCOMC__)) && defined(SDL_ASSEMBLY_ROUTINES)
27 // MMX assembler version of SDL_MixAudio for signed little endian 16 bit samples and signed 8 bit samples
28 // Copyright 2002 Stephane Marchesin (stephane.marchesin@wanadoo.fr)
29 // Converted to Intel ASM notation by Cth
30 // This code is licensed under the LGPL (see COPYING for details)
31 //
32 // Assumes buffer size in bytes is a multiple of 16
33 // Assumes SDL_MIX_MAXVOLUME = 128
34
35
36 ////////////////////////////////////////////////
37 // Mixing for 16 bit signed buffers
38 ////////////////////////////////////////////////
39
SDL_MixAudio_MMX_S16_VC(char * dst,char * src,unsigned int nSize,int volume)40 void SDL_MixAudio_MMX_S16_VC(char* dst,char* src,unsigned int nSize,int volume)
41 {
42 __asm
43 {
44
45 push edi
46 push esi
47 push ebx
48
49 mov edi, dst // edi = dst
50 mov esi, src // esi = src
51 mov eax, volume // eax = volume
52 mov ebx, nSize // ebx = size
53 shr ebx, 4 // process 16 bytes per iteration = 8 samples
54 jz endS16
55
56 pxor mm0, mm0
57 movd mm0, eax //%%eax,%%mm0
58 movq mm1, mm0 //%%mm0,%%mm1
59 psllq mm0, 16 //$16,%%mm0
60 por mm0, mm1 //%%mm1,%%mm0
61 psllq mm0, 16 //$16,%%mm0
62 por mm0, mm1 //%%mm1,%%mm0
63 psllq mm0, 16 //$16,%%mm0
64 por mm0, mm1 //%%mm1,%%mm0 // mm0 = vol|vol|vol|vol
65
66 #ifndef __WATCOMC__
67 align 16
68 #endif
69 mixloopS16:
70 movq mm1, [esi] //(%%esi),%%mm1\n" // mm1 = a|b|c|d
71 movq mm2, mm1 //%%mm1,%%mm2\n" // mm2 = a|b|c|d
72 movq mm4, [esi + 8] //8(%%esi),%%mm4\n" // mm4 = e|f|g|h
73 // pre charger le buffer dst dans mm7
74 movq mm7, [edi] //(%%edi),%%mm7\n" // mm7 = dst[0]"
75 // multiplier par le volume
76 pmullw mm1, mm0 //%%mm0,%%mm1\n" // mm1 = l(a*v)|l(b*v)|l(c*v)|l(d*v)
77 pmulhw mm2, mm0 //%%mm0,%%mm2\n" // mm2 = h(a*v)|h(b*v)|h(c*v)|h(d*v)
78 movq mm5, mm4 //%%mm4,%%mm5\n" // mm5 = e|f|g|h
79 pmullw mm4, mm0 //%%mm0,%%mm4\n" // mm4 = l(e*v)|l(f*v)|l(g*v)|l(h*v)
80 pmulhw mm5, mm0 //%%mm0,%%mm5\n" // mm5 = h(e*v)|h(f*v)|h(g*v)|h(h*v)
81 movq mm3, mm1 //%%mm1,%%mm3\n" // mm3 = l(a*v)|l(b*v)|l(c*v)|l(d*v)
82 punpckhwd mm1, mm2 //%%mm2,%%mm1\n" // mm1 = a*v|b*v
83 movq mm6, mm4 //%%mm4,%%mm6\n" // mm6 = l(e*v)|l(f*v)|l(g*v)|l(h*v)
84 punpcklwd mm3, mm2 //%%mm2,%%mm3\n" // mm3 = c*v|d*v
85 punpckhwd mm4, mm5 //%%mm5,%%mm4\n" // mm4 = e*f|f*v
86 punpcklwd mm6, mm5 //%%mm5,%%mm6\n" // mm6 = g*v|h*v
87 // pre charger le buffer dst dans mm5
88 movq mm5, [edi + 8] //8(%%edi),%%mm5\n" // mm5 = dst[1]
89 // diviser par 128
90 psrad mm1, 7 //$7,%%mm1\n" // mm1 = a*v/128|b*v/128 , 128 = SDL_MIX_MAXVOLUME
91 add esi, 16 //$16,%%esi\n"
92 psrad mm3, 7 //$7,%%mm3\n" // mm3 = c*v/128|d*v/128
93 psrad mm4, 7 //$7,%%mm4\n" // mm4 = e*v/128|f*v/128
94 // mm1 = le sample avec le volume modifie
95 packssdw mm3, mm1 //%%mm1,%%mm3\n" // mm3 = s(a*v|b*v|c*v|d*v)
96 psrad mm6, 7 //$7,%%mm6\n" // mm6= g*v/128|h*v/128
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
115 }
116
117 ////////////////////////////////////////////////
118 // Mixing for 8 bit signed buffers
119 ////////////////////////////////////////////////
120
SDL_MixAudio_MMX_S8_VC(char * dst,char * src,unsigned int nSize,int volume)121 void SDL_MixAudio_MMX_S8_VC(char* dst,char* src,unsigned int nSize,int volume)
122 {
123 _asm
124 {
125
126 push edi
127 push esi
128 push ebx
129
130 mov edi, dst //movl %0,%%edi // edi = dst
131 mov esi, src //%1,%%esi // esi = src
132 mov eax, volume //%3,%%eax // eax = volume
133
134 movd mm0, eax //%%eax,%%mm0
135 movq mm1, mm0 //%%mm0,%%mm1
136 psllq mm0, 16 //$16,%%mm0
137 por mm0, mm1 //%%mm1,%%mm0
138 psllq mm0, 16 //$16,%%mm0
139 por mm0, mm1 //%%mm1,%%mm0
140 psllq mm0, 16 //$16,%%mm0
141 por mm0, mm1 //%%mm1,%%mm0
142
143 mov ebx, nSize //%2,%%ebx // ebx = size
144 shr ebx, 3 //$3,%%ebx // process 8 bytes per iteration = 8 samples
145 cmp ebx, 0 //$0,%%ebx
146 je endS8
147
148 #ifndef __WATCOMC__
149 align 16
150 #endif
151 mixloopS8:
152 pxor mm2, mm2 //%%mm2,%%mm2 // mm2 = 0
153 movq mm1, [esi] //(%%esi),%%mm1 // mm1 = a|b|c|d|e|f|g|h
154 movq mm3, mm1 //%%mm1,%%mm3 // mm3 = a|b|c|d|e|f|g|h
155 // on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0
156 pcmpgtb mm2, mm1 //%%mm1,%%mm2 // mm2 = 11111111|00000000|00000000....
157 punpckhbw mm1, mm2 //%%mm2,%%mm1 // mm1 = 0|a|0|b|0|c|0|d
158 punpcklbw mm3, mm2 //%%mm2,%%mm3 // mm3 = 0|e|0|f|0|g|0|h
159 movq mm2, [edi] //(%%edi),%%mm2 // mm2 = destination
160 pmullw mm1, mm0 //%%mm0,%%mm1 // mm1 = v*a|v*b|v*c|v*d
161 add esi, 8 //$8,%%esi
162 pmullw mm3, mm0 //%%mm0,%%mm3 // mm3 = v*e|v*f|v*g|v*h
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 }
180
181 #endif /* SDL_ASSEMBLY_ROUTINES */
182