1 /*
2 SDL - Simple DirectMedia Layer
3 Copyright (C) 1997-2012 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(SDL_BUGGY_MMX_MIXERS) /* buggy, so we're disabling them. --ryan. */
39 #if defined(__GNUC__) && defined(__i386__) && defined(SDL_ASSEMBLY_ROUTINES)
SDL_MixAudio_MMX_S16(char * dst,char * src,unsigned int size,int volume)40 void SDL_MixAudio_MMX_S16(char* dst,char* src,unsigned int size,int volume)
41 {
42 __asm__ __volatile__ (
43
44 " movl %3,%%eax\n" /* eax = volume */
45
46 " movl %2,%%edx\n" /* edx = size */
47
48 " shrl $4,%%edx\n" /* process 16 bytes per iteration = 8 samples */
49
50 " jz .endS16\n"
51
52 " pxor %%mm0,%%mm0\n"
53
54 " movd %%eax,%%mm0\n"
55 " movq %%mm0,%%mm1\n"
56 " psllq $16,%%mm0\n"
57 " por %%mm1,%%mm0\n"
58 " psllq $16,%%mm0\n"
59 " por %%mm1,%%mm0\n"
60 " psllq $16,%%mm0\n"
61 " por %%mm1,%%mm0\n" /* mm0 = vol|vol|vol|vol */
62
63 ".align 8\n"
64 " .mixloopS16:\n"
65
66 " movq (%1),%%mm1\n" /* mm1 = a|b|c|d */
67
68 " movq %%mm1,%%mm2\n" /* mm2 = a|b|c|d */
69
70 " movq 8(%1),%%mm4\n" /* mm4 = e|f|g|h */
71
72 /* pr� charger le buffer dst dans mm7 */
73 " movq (%0),%%mm7\n" /* mm7 = dst[0] */
74
75 /* multiplier par le volume */
76 " pmullw %%mm0,%%mm1\n" /* mm1 = l(a*v)|l(b*v)|l(c*v)|l(d*v) */
77
78 " pmulhw %%mm0,%%mm2\n" /* mm2 = h(a*v)|h(b*v)|h(c*v)|h(d*v) */
79 " movq %%mm4,%%mm5\n" /* mm5 = e|f|g|h */
80
81 " pmullw %%mm0,%%mm4\n" /* mm4 = l(e*v)|l(f*v)|l(g*v)|l(h*v) */
82
83 " pmulhw %%mm0,%%mm5\n" /* mm5 = h(e*v)|h(f*v)|h(g*v)|h(h*v) */
84 " movq %%mm1,%%mm3\n" /* mm3 = l(a*v)|l(b*v)|l(c*v)|l(d*v) */
85
86 " punpckhwd %%mm2,%%mm1\n" /* mm1 = a*v|b*v */
87
88 " movq %%mm4,%%mm6\n" /* mm6 = l(e*v)|l(f*v)|l(g*v)|l(h*v) */
89 " punpcklwd %%mm2,%%mm3\n" /* mm3 = c*v|d*v */
90
91 " punpckhwd %%mm5,%%mm4\n" /* mm4 = e*f|f*v */
92
93 " punpcklwd %%mm5,%%mm6\n" /* mm6 = g*v|h*v */
94
95 /* pr� charger le buffer dst dans mm5 */
96 " movq 8(%0),%%mm5\n" /* mm5 = dst[1] */
97
98 /* diviser par 128 */
99 " psrad $7,%%mm1\n" /* mm1 = a*v/128|b*v/128 , 128 = SDL_MIX_MAXVOLUME */
100 " add $16,%1\n"
101
102 " psrad $7,%%mm3\n" /* mm3 = c*v/128|d*v/128 */
103
104 " psrad $7,%%mm4\n" /* mm4 = e*v/128|f*v/128 */
105
106 /* mm1 = le sample avec le volume modifi� */
107 " packssdw %%mm1,%%mm3\n" /* mm3 = s(a*v|b*v|c*v|d*v) */
108
109 " psrad $7,%%mm6\n" /* mm6= g*v/128|h*v/128 */
110 " paddsw %%mm7,%%mm3\n" /* mm3 = adjust_volume(src)+dst */
111
112 /* mm4 = le sample avec le volume modifi� */
113 " packssdw %%mm4,%%mm6\n" /* mm6 = s(e*v|f*v|g*v|h*v) */
114 " movq %%mm3,(%0)\n"
115
116 " paddsw %%mm5,%%mm6\n" /* mm6 = adjust_volume(src)+dst */
117
118 " movq %%mm6,8(%0)\n"
119
120 " add $16,%0\n"
121
122 " dec %%edx\n"
123
124 " jnz .mixloopS16\n"
125
126 " emms\n"
127
128 ".endS16:\n"
129 :
130 : "r" (dst), "r"(src),"m"(size),
131 "m"(volume)
132 : "eax","edx","memory"
133 );
134 }
135
136
137
138 /*////////////////////////////////////////////// */
139 /* Mixing for 8 bit signed buffers */
140 /*////////////////////////////////////////////// */
141
SDL_MixAudio_MMX_S8(char * dst,char * src,unsigned int size,int volume)142 void SDL_MixAudio_MMX_S8(char* dst,char* src,unsigned int size,int volume)
143 {
144 __asm__ __volatile__ (
145
146 " movl %3,%%eax\n" /* eax = volume */
147
148 " movd %%eax,%%mm0\n"
149 " movq %%mm0,%%mm1\n"
150 " psllq $16,%%mm0\n"
151 " por %%mm1,%%mm0\n"
152 " psllq $16,%%mm0\n"
153 " por %%mm1,%%mm0\n"
154 " psllq $16,%%mm0\n"
155 " por %%mm1,%%mm0\n"
156
157 " movl %2,%%edx\n" /* edx = size */
158 " shr $3,%%edx\n" /* process 8 bytes per iteration = 8 samples */
159
160 " cmp $0,%%edx\n"
161 " je .endS8\n"
162
163 ".align 8\n"
164 " .mixloopS8:\n"
165
166 " pxor %%mm2,%%mm2\n" /* mm2 = 0 */
167 " movq (%1),%%mm1\n" /* mm1 = a|b|c|d|e|f|g|h */
168
169 " movq %%mm1,%%mm3\n" /* mm3 = a|b|c|d|e|f|g|h */
170
171 /* on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0 */
172 " pcmpgtb %%mm1,%%mm2\n" /* mm2 = 11111111|00000000|00000000.... */
173
174 " punpckhbw %%mm2,%%mm1\n" /* mm1 = 0|a|0|b|0|c|0|d */
175
176 " punpcklbw %%mm2,%%mm3\n" /* mm3 = 0|e|0|f|0|g|0|h */
177 " movq (%0),%%mm2\n" /* mm2 = destination */
178
179 " pmullw %%mm0,%%mm1\n" /* mm1 = v*a|v*b|v*c|v*d */
180 " add $8,%1\n"
181
182 " pmullw %%mm0,%%mm3\n" /* mm3 = v*e|v*f|v*g|v*h */
183 " psraw $7,%%mm1\n" /* mm1 = v*a/128|v*b/128|v*c/128|v*d/128 */
184
185 " psraw $7,%%mm3\n" /* mm3 = v*e/128|v*f/128|v*g/128|v*h/128 */
186
187 " 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 */
188
189 " paddsb %%mm2,%%mm3\n" /* add to destination buffer */
190
191 " movq %%mm3,(%0)\n" /* store back to ram */
192 " add $8,%0\n"
193
194 " dec %%edx\n"
195
196 " jnz .mixloopS8\n"
197
198 ".endS8:\n"
199 " emms\n"
200 :
201 : "r" (dst), "r"(src),"m"(size),
202 "m"(volume)
203 : "eax","edx","memory"
204 );
205 }
206 #endif
207 #endif
208