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 /*
25 MMX assembler version of SDL_MixAudio for signed little endian 16 bit samples and signed 8 bit samples
26 Copyright 2002 Stephane Marchesin (stephane.marchesin@wanadoo.fr)
27 This code is licensed under the LGPL (see COPYING for details)
28
29 Assumes buffer size in bytes is a multiple of 16
30 Assumes SDL_MIX_MAXVOLUME = 128
31 */
32
33
34 /***********************************************
35 * Mixing for 16 bit signed buffers
36 ***********************************************/
37
38 #if defined(__GNUC__) && defined(__i386__) && defined(SDL_ASSEMBLY_ROUTINES)
SDL_MixAudio_MMX_S16(char * dst,char * src,unsigned int size,int volume)39 void SDL_MixAudio_MMX_S16(char* dst,char* src,unsigned int size,int volume)
40 {
41 __asm__ __volatile__ (
42
43 " movl %3,%%eax\n" /* eax = volume */
44
45 " movl %2,%%edx\n" /* edx = size */
46
47 " shrl $4,%%edx\n" /* process 16 bytes per iteration = 8 samples */
48
49 " jz .endS16\n"
50
51 " pxor %%mm0,%%mm0\n"
52
53 " movd %%eax,%%mm0\n"
54 " movq %%mm0,%%mm1\n"
55 " psllq $16,%%mm0\n"
56 " por %%mm1,%%mm0\n"
57 " psllq $16,%%mm0\n"
58 " por %%mm1,%%mm0\n"
59 " psllq $16,%%mm0\n"
60 " por %%mm1,%%mm0\n" /* mm0 = vol|vol|vol|vol */
61
62 ".align 8\n"
63 " .mixloopS16:\n"
64
65 " movq (%1),%%mm1\n" /* mm1 = a|b|c|d */
66
67 " movq %%mm1,%%mm2\n" /* mm2 = a|b|c|d */
68
69 " movq 8(%1),%%mm4\n" /* mm4 = e|f|g|h */
70
71 /* pr� charger le buffer dst dans mm7 */
72 " movq (%0),%%mm7\n" /* mm7 = dst[0] */
73
74 /* multiplier par le volume */
75 " pmullw %%mm0,%%mm1\n" /* mm1 = l(a*v)|l(b*v)|l(c*v)|l(d*v) */
76
77 " pmulhw %%mm0,%%mm2\n" /* mm2 = h(a*v)|h(b*v)|h(c*v)|h(d*v) */
78 " movq %%mm4,%%mm5\n" /* mm5 = e|f|g|h */
79
80 " pmullw %%mm0,%%mm4\n" /* mm4 = l(e*v)|l(f*v)|l(g*v)|l(h*v) */
81
82 " pmulhw %%mm0,%%mm5\n" /* mm5 = h(e*v)|h(f*v)|h(g*v)|h(h*v) */
83 " movq %%mm1,%%mm3\n" /* mm3 = l(a*v)|l(b*v)|l(c*v)|l(d*v) */
84
85 " punpckhwd %%mm2,%%mm1\n" /* mm1 = a*v|b*v */
86
87 " movq %%mm4,%%mm6\n" /* mm6 = l(e*v)|l(f*v)|l(g*v)|l(h*v) */
88 " punpcklwd %%mm2,%%mm3\n" /* mm3 = c*v|d*v */
89
90 " punpckhwd %%mm5,%%mm4\n" /* mm4 = e*f|f*v */
91
92 " punpcklwd %%mm5,%%mm6\n" /* mm6 = g*v|h*v */
93
94 /* pr� charger le buffer dst dans mm5 */
95 " movq 8(%0),%%mm5\n" /* mm5 = dst[1] */
96
97 /* diviser par 128 */
98 " psrad $7,%%mm1\n" /* mm1 = a*v/128|b*v/128 , 128 = SDL_MIX_MAXVOLUME */
99 " add $16,%1\n"
100
101 " psrad $7,%%mm3\n" /* mm3 = c*v/128|d*v/128 */
102
103 " psrad $7,%%mm4\n" /* mm4 = e*v/128|f*v/128 */
104
105 /* mm1 = le sample avec le volume modifi� */
106 " packssdw %%mm1,%%mm3\n" /* mm3 = s(a*v|b*v|c*v|d*v) */
107
108 " psrad $7,%%mm6\n" /* mm6= g*v/128|h*v/128 */
109 " paddsw %%mm7,%%mm3\n" /* mm3 = adjust_volume(src)+dst */
110
111 /* mm4 = le sample avec le volume modifi� */
112 " packssdw %%mm4,%%mm6\n" /* mm6 = s(e*v|f*v|g*v|h*v) */
113 " movq %%mm3,(%0)\n"
114
115 " paddsw %%mm5,%%mm6\n" /* mm6 = adjust_volume(src)+dst */
116
117 " movq %%mm6,8(%0)\n"
118
119 " add $16,%0\n"
120
121 " dec %%edx\n"
122
123 " jnz .mixloopS16\n"
124
125 " emms\n"
126
127 ".endS16:\n"
128 :
129 : "r" (dst), "r"(src),"m"(size),
130 "m"(volume)
131 : "eax","edx","memory"
132 );
133 }
134
135
136
137 /*////////////////////////////////////////////// */
138 /* Mixing for 8 bit signed buffers */
139 /*////////////////////////////////////////////// */
140
SDL_MixAudio_MMX_S8(char * dst,char * src,unsigned int size,int volume)141 void SDL_MixAudio_MMX_S8(char* dst,char* src,unsigned int size,int volume)
142 {
143 __asm__ __volatile__ (
144
145 " movl %3,%%eax\n" /* eax = volume */
146
147 " movd %%eax,%%mm0\n"
148 " movq %%mm0,%%mm1\n"
149 " psllq $16,%%mm0\n"
150 " por %%mm1,%%mm0\n"
151 " psllq $16,%%mm0\n"
152 " por %%mm1,%%mm0\n"
153 " psllq $16,%%mm0\n"
154 " por %%mm1,%%mm0\n"
155
156 " movl %2,%%edx\n" /* edx = size */
157 " shr $3,%%edx\n" /* process 8 bytes per iteration = 8 samples */
158
159 " cmp $0,%%edx\n"
160 " je .endS8\n"
161
162 ".align 8\n"
163 " .mixloopS8:\n"
164
165 " pxor %%mm2,%%mm2\n" /* mm2 = 0 */
166 " movq (%1),%%mm1\n" /* mm1 = a|b|c|d|e|f|g|h */
167
168 " movq %%mm1,%%mm3\n" /* mm3 = a|b|c|d|e|f|g|h */
169
170 /* on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0 */
171 " pcmpgtb %%mm1,%%mm2\n" /* mm2 = 11111111|00000000|00000000.... */
172
173 " punpckhbw %%mm2,%%mm1\n" /* mm1 = 0|a|0|b|0|c|0|d */
174
175 " punpcklbw %%mm2,%%mm3\n" /* mm3 = 0|e|0|f|0|g|0|h */
176 " movq (%0),%%mm2\n" /* mm2 = destination */
177
178 " pmullw %%mm0,%%mm1\n" /* mm1 = v*a|v*b|v*c|v*d */
179 " add $8,%1\n"
180
181 " pmullw %%mm0,%%mm3\n" /* mm3 = v*e|v*f|v*g|v*h */
182 " psraw $7,%%mm1\n" /* mm1 = v*a/128|v*b/128|v*c/128|v*d/128 */
183
184 " psraw $7,%%mm3\n" /* mm3 = v*e/128|v*f/128|v*g/128|v*h/128 */
185
186 " packsswb %%mm1,%%mm3\n" /* 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 */
187
188 " paddsb %%mm2,%%mm3\n" /* add to destination buffer */
189
190 " movq %%mm3,(%0)\n" /* store back to ram */
191 " add $8,%0\n"
192
193 " dec %%edx\n"
194
195 " jnz .mixloopS8\n"
196
197 ".endS8:\n"
198 " emms\n"
199 :
200 : "r" (dst), "r"(src),"m"(size),
201 "m"(volume)
202 : "eax","edx","memory"
203 );
204 }
205 #endif
206
207