|
16174
|
1 Index: liba52/a52_internal.h
|
|
|
2 ===================================================================
|
|
|
3 RCS file: /cvsroot/mplayer/main/liba52/a52_internal.h,v
|
|
|
4 retrieving revision 1.4
|
|
|
5 diff -u -r1.4 a52_internal.h
|
|
|
6 --- liba52/a52_internal.h 22 Mar 2005 23:27:18 -0000 1.4
|
|
|
7 +++ liba52/a52_internal.h 31 Jul 2005 21:20:09 -0000
|
|
|
8 @@ -41,6 +41,20 @@
|
|
|
9 #define DELTA_BIT_NONE (2)
|
|
|
10 #define DELTA_BIT_RESERVED (3)
|
|
|
11
|
|
|
12 +#ifdef ARCH_X86_64
|
|
|
13 +# define REG_a "rax"
|
|
|
14 +# define REG_d "rdx"
|
|
|
15 +# define REG_S "rsi"
|
|
|
16 +# define REG_D "rdi"
|
|
|
17 +# define REG_BP "rbp"
|
|
|
18 +#else
|
|
|
19 +# define REG_a "eax"
|
|
|
20 +# define REG_d "edx"
|
|
|
21 +# define REG_S "esi"
|
|
|
22 +# define REG_D "edi"
|
|
|
23 +# define REG_BP "ebp"
|
|
|
24 +#endif
|
|
|
25 +
|
|
|
26 void bit_allocate (a52_state_t * state, a52_ba_t * ba, int bndstart,
|
|
|
27 int start, int end, int fastleak, int slowleak,
|
|
|
28 uint8_t * exp, int8_t * bap);
|
|
|
29 Index: liba52/downmix.c
|
|
|
30 ===================================================================
|
|
|
31 RCS file: /cvsroot/mplayer/main/liba52/downmix.c,v
|
|
|
32 retrieving revision 1.17
|
|
|
33 diff -u -r1.17 downmix.c
|
|
|
34 --- liba52/downmix.c 22 Mar 2005 23:27:18 -0000 1.17
|
|
|
35 +++ liba52/downmix.c 31 Jul 2005 21:20:09 -0000
|
|
|
36 @@ -56,7 +56,7 @@
|
|
|
37 {
|
|
|
38 upmix= upmix_C;
|
|
|
39 downmix= downmix_C;
|
|
|
40 -#ifdef ARCH_X86
|
|
|
41 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
42 if(mm_accel & MM_ACCEL_X86_MMX) upmix= upmix_MMX;
|
|
|
43 if(mm_accel & MM_ACCEL_X86_SSE) downmix= downmix_SSE;
|
|
|
44 if(mm_accel & MM_ACCEL_X86_3DNOW) downmix= downmix_3dnow;
|
|
|
45 @@ -684,27 +684,27 @@
|
|
|
46 }
|
|
|
47 }
|
|
|
48
|
|
|
49 -#ifdef ARCH_X86
|
|
|
50 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
51 static void mix2to1_SSE (sample_t * dest, sample_t * src, sample_t bias)
|
|
|
52 {
|
|
|
53 asm volatile(
|
|
|
54 "movlps %2, %%xmm7 \n\t"
|
|
|
55 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
56 - "movl $-1024, %%esi \n\t"
|
|
|
57 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
58 ".balign 16\n\t"
|
|
|
59 "1: \n\t"
|
|
|
60 - "movaps (%0, %%esi), %%xmm0 \n\t"
|
|
|
61 - "movaps 16(%0, %%esi), %%xmm1 \n\t"
|
|
|
62 - "addps (%1, %%esi), %%xmm0 \n\t"
|
|
|
63 - "addps 16(%1, %%esi), %%xmm1 \n\t"
|
|
|
64 + "movaps (%0, %%"REG_S"), %%xmm0 \n\t"
|
|
|
65 + "movaps 16(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
66 + "addps (%1, %%"REG_S"), %%xmm0 \n\t"
|
|
|
67 + "addps 16(%1, %%"REG_S"), %%xmm1\n\t"
|
|
|
68 "addps %%xmm7, %%xmm0 \n\t"
|
|
|
69 "addps %%xmm7, %%xmm1 \n\t"
|
|
|
70 - "movaps %%xmm0, (%1, %%esi) \n\t"
|
|
|
71 - "movaps %%xmm1, 16(%1, %%esi) \n\t"
|
|
|
72 - "addl $32, %%esi \n\t"
|
|
|
73 + "movaps %%xmm0, (%1, %%"REG_S") \n\t"
|
|
|
74 + "movaps %%xmm1, 16(%1, %%"REG_S")\n\t"
|
|
|
75 + "add $32, %%"REG_S" \n\t"
|
|
|
76 " jnz 1b \n\t"
|
|
|
77 :: "r" (src+256), "r" (dest+256), "m" (bias)
|
|
|
78 - : "%esi"
|
|
|
79 + : "%"REG_S
|
|
|
80 );
|
|
|
81 }
|
|
|
82
|
|
|
83 @@ -713,19 +713,19 @@
|
|
|
84 asm volatile(
|
|
|
85 "movlps %1, %%xmm7 \n\t"
|
|
|
86 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
87 - "movl $-1024, %%esi \n\t"
|
|
|
88 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
89 ".balign 16\n\t"
|
|
|
90 "1: \n\t"
|
|
|
91 - "movaps (%0, %%esi), %%xmm0 \n\t"
|
|
|
92 - "movaps 1024(%0, %%esi), %%xmm1 \n\t"
|
|
|
93 - "addps 2048(%0, %%esi), %%xmm0 \n\t"
|
|
|
94 + "movaps (%0, %%"REG_S"), %%xmm0 \n\t"
|
|
|
95 + "movaps 1024(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
96 + "addps 2048(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
97 "addps %%xmm7, %%xmm1 \n\t"
|
|
|
98 "addps %%xmm1, %%xmm0 \n\t"
|
|
|
99 - "movaps %%xmm0, (%0, %%esi) \n\t"
|
|
|
100 - "addl $16, %%esi \n\t"
|
|
|
101 + "movaps %%xmm0, (%0, %%"REG_S") \n\t"
|
|
|
102 + "add $16, %%"REG_S" \n\t"
|
|
|
103 " jnz 1b \n\t"
|
|
|
104 :: "r" (samples+256), "m" (bias)
|
|
|
105 - : "%esi"
|
|
|
106 + : "%"REG_S
|
|
|
107 );
|
|
|
108 }
|
|
|
109
|
|
|
110 @@ -734,20 +734,20 @@
|
|
|
111 asm volatile(
|
|
|
112 "movlps %1, %%xmm7 \n\t"
|
|
|
113 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
114 - "movl $-1024, %%esi \n\t"
|
|
|
115 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
116 ".balign 16\n\t"
|
|
|
117 "1: \n\t"
|
|
|
118 - "movaps (%0, %%esi), %%xmm0 \n\t"
|
|
|
119 - "movaps 1024(%0, %%esi), %%xmm1 \n\t"
|
|
|
120 - "addps 2048(%0, %%esi), %%xmm0 \n\t"
|
|
|
121 - "addps 3072(%0, %%esi), %%xmm1 \n\t"
|
|
|
122 + "movaps (%0, %%"REG_S"), %%xmm0 \n\t"
|
|
|
123 + "movaps 1024(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
124 + "addps 2048(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
125 + "addps 3072(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
126 "addps %%xmm7, %%xmm0 \n\t"
|
|
|
127 "addps %%xmm1, %%xmm0 \n\t"
|
|
|
128 - "movaps %%xmm0, (%0, %%esi) \n\t"
|
|
|
129 - "addl $16, %%esi \n\t"
|
|
|
130 + "movaps %%xmm0, (%0, %%"REG_S") \n\t"
|
|
|
131 + "add $16, %%"REG_S" \n\t"
|
|
|
132 " jnz 1b \n\t"
|
|
|
133 :: "r" (samples+256), "m" (bias)
|
|
|
134 - : "%esi"
|
|
|
135 + : "%"REG_S
|
|
|
136 );
|
|
|
137 }
|
|
|
138
|
|
|
139 @@ -756,21 +756,21 @@
|
|
|
140 asm volatile(
|
|
|
141 "movlps %1, %%xmm7 \n\t"
|
|
|
142 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
143 - "movl $-1024, %%esi \n\t"
|
|
|
144 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
145 ".balign 16\n\t"
|
|
|
146 "1: \n\t"
|
|
|
147 - "movaps (%0, %%esi), %%xmm0 \n\t"
|
|
|
148 - "movaps 1024(%0, %%esi), %%xmm1 \n\t"
|
|
|
149 - "addps 2048(%0, %%esi), %%xmm0 \n\t"
|
|
|
150 - "addps 3072(%0, %%esi), %%xmm1 \n\t"
|
|
|
151 + "movaps (%0, %%"REG_S"), %%xmm0 \n\t"
|
|
|
152 + "movaps 1024(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
153 + "addps 2048(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
154 + "addps 3072(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
155 "addps %%xmm7, %%xmm0 \n\t"
|
|
|
156 - "addps 4096(%0, %%esi), %%xmm1 \n\t"
|
|
|
157 + "addps 4096(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
158 "addps %%xmm1, %%xmm0 \n\t"
|
|
|
159 - "movaps %%xmm0, (%0, %%esi) \n\t"
|
|
|
160 - "addl $16, %%esi \n\t"
|
|
|
161 + "movaps %%xmm0, (%0, %%"REG_S") \n\t"
|
|
|
162 + "add $16, %%"REG_S" \n\t"
|
|
|
163 " jnz 1b \n\t"
|
|
|
164 :: "r" (samples+256), "m" (bias)
|
|
|
165 - : "%esi"
|
|
|
166 + : "%"REG_S
|
|
|
167 );
|
|
|
168 }
|
|
|
169
|
|
|
170 @@ -779,21 +779,21 @@
|
|
|
171 asm volatile(
|
|
|
172 "movlps %1, %%xmm7 \n\t"
|
|
|
173 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
174 - "movl $-1024, %%esi \n\t"
|
|
|
175 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
176 ".balign 16\n\t"
|
|
|
177 "1: \n\t"
|
|
|
178 - "movaps 1024(%0, %%esi), %%xmm0 \n\t"
|
|
|
179 + "movaps 1024(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
180 "addps %%xmm7, %%xmm0 \n\t" //common
|
|
|
181 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
182 - "movaps 2048(%0, %%esi), %%xmm2 \n\t"
|
|
|
183 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
184 + "movaps 2048(%0, %%"REG_S"), %%xmm2\n\t"
|
|
|
185 "addps %%xmm0, %%xmm1 \n\t"
|
|
|
186 "addps %%xmm0, %%xmm2 \n\t"
|
|
|
187 - "movaps %%xmm1, (%0, %%esi) \n\t"
|
|
|
188 - "movaps %%xmm2, 1024(%0, %%esi) \n\t"
|
|
|
189 - "addl $16, %%esi \n\t"
|
|
|
190 + "movaps %%xmm1, (%0, %%"REG_S") \n\t"
|
|
|
191 + "movaps %%xmm2, 1024(%0, %%"REG_S")\n\t"
|
|
|
192 + "add $16, %%"REG_S" \n\t"
|
|
|
193 " jnz 1b \n\t"
|
|
|
194 :: "r" (samples+256), "m" (bias)
|
|
|
195 - : "%esi"
|
|
|
196 + : "%"REG_S
|
|
|
197 );
|
|
|
198 }
|
|
|
199
|
|
|
200 @@ -802,21 +802,21 @@
|
|
|
201 asm volatile(
|
|
|
202 "movlps %2, %%xmm7 \n\t"
|
|
|
203 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
204 - "movl $-1024, %%esi \n\t"
|
|
|
205 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
206 ".balign 16\n\t"
|
|
|
207 "1: \n\t"
|
|
|
208 - "movaps 1024(%1, %%esi), %%xmm0 \n\t"
|
|
|
209 + "movaps 1024(%1, %%"REG_S"), %%xmm0\n\t"
|
|
|
210 "addps %%xmm7, %%xmm0 \n\t" //common
|
|
|
211 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
212 - "movaps (%1, %%esi), %%xmm2 \n\t"
|
|
|
213 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
214 + "movaps (%1, %%"REG_S"), %%xmm2 \n\t"
|
|
|
215 "addps %%xmm0, %%xmm1 \n\t"
|
|
|
216 "addps %%xmm0, %%xmm2 \n\t"
|
|
|
217 - "movaps %%xmm1, (%0, %%esi) \n\t"
|
|
|
218 - "movaps %%xmm2, (%1, %%esi) \n\t"
|
|
|
219 - "addl $16, %%esi \n\t"
|
|
|
220 + "movaps %%xmm1, (%0, %%"REG_S") \n\t"
|
|
|
221 + "movaps %%xmm2, (%1, %%"REG_S") \n\t"
|
|
|
222 + "add $16, %%"REG_S" \n\t"
|
|
|
223 " jnz 1b \n\t"
|
|
|
224 :: "r" (left+256), "r" (right+256), "m" (bias)
|
|
|
225 - : "%esi"
|
|
|
226 + : "%"REG_S
|
|
|
227 );
|
|
|
228 }
|
|
|
229
|
|
|
230 @@ -825,22 +825,22 @@
|
|
|
231 asm volatile(
|
|
|
232 "movlps %1, %%xmm7 \n\t"
|
|
|
233 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
234 - "movl $-1024, %%esi \n\t"
|
|
|
235 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
236 ".balign 16\n\t"
|
|
|
237 "1: \n\t"
|
|
|
238 - "movaps 2048(%0, %%esi), %%xmm0 \n\t" // surround
|
|
|
239 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
240 - "movaps 1024(%0, %%esi), %%xmm2 \n\t"
|
|
|
241 + "movaps 2048(%0, %%"REG_S"), %%xmm0\n\t" // surround
|
|
|
242 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
243 + "movaps 1024(%0, %%"REG_S"), %%xmm2\n\t"
|
|
|
244 "addps %%xmm7, %%xmm1 \n\t"
|
|
|
245 "addps %%xmm7, %%xmm2 \n\t"
|
|
|
246 "subps %%xmm0, %%xmm1 \n\t"
|
|
|
247 "addps %%xmm0, %%xmm2 \n\t"
|
|
|
248 - "movaps %%xmm1, (%0, %%esi) \n\t"
|
|
|
249 - "movaps %%xmm2, 1024(%0, %%esi) \n\t"
|
|
|
250 - "addl $16, %%esi \n\t"
|
|
|
251 + "movaps %%xmm1, (%0, %%"REG_S") \n\t"
|
|
|
252 + "movaps %%xmm2, 1024(%0, %%"REG_S")\n\t"
|
|
|
253 + "add $16, %%"REG_S" \n\t"
|
|
|
254 " jnz 1b \n\t"
|
|
|
255 :: "r" (samples+256), "m" (bias)
|
|
|
256 - : "%esi"
|
|
|
257 + : "%"REG_S
|
|
|
258 );
|
|
|
259 }
|
|
|
260
|
|
|
261 @@ -849,22 +849,22 @@
|
|
|
262 asm volatile(
|
|
|
263 "movlps %1, %%xmm7 \n\t"
|
|
|
264 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
265 - "movl $-1024, %%esi \n\t"
|
|
|
266 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
267 ".balign 16\n\t"
|
|
|
268 "1: \n\t"
|
|
|
269 - "movaps 1024(%0, %%esi), %%xmm0 \n\t"
|
|
|
270 - "addps 3072(%0, %%esi), %%xmm0 \n\t"
|
|
|
271 + "movaps 1024(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
272 + "addps 3072(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
273 "addps %%xmm7, %%xmm0 \n\t" // common
|
|
|
274 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
275 - "movaps 2048(%0, %%esi), %%xmm2 \n\t"
|
|
|
276 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
277 + "movaps 2048(%0, %%"REG_S"), %%xmm2\n\t"
|
|
|
278 "addps %%xmm0, %%xmm1 \n\t"
|
|
|
279 "addps %%xmm0, %%xmm2 \n\t"
|
|
|
280 - "movaps %%xmm1, (%0, %%esi) \n\t"
|
|
|
281 - "movaps %%xmm2, 1024(%0, %%esi) \n\t"
|
|
|
282 - "addl $16, %%esi \n\t"
|
|
|
283 + "movaps %%xmm1, (%0, %%"REG_S") \n\t"
|
|
|
284 + "movaps %%xmm2, 1024(%0, %%"REG_S")\n\t"
|
|
|
285 + "add $16, %%"REG_S" \n\t"
|
|
|
286 " jnz 1b \n\t"
|
|
|
287 :: "r" (samples+256), "m" (bias)
|
|
|
288 - : "%esi"
|
|
|
289 + : "%"REG_S
|
|
|
290 );
|
|
|
291 }
|
|
|
292
|
|
|
293 @@ -873,24 +873,24 @@
|
|
|
294 asm volatile(
|
|
|
295 "movlps %1, %%xmm7 \n\t"
|
|
|
296 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
297 - "movl $-1024, %%esi \n\t"
|
|
|
298 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
299 ".balign 16\n\t"
|
|
|
300 "1: \n\t"
|
|
|
301 - "movaps 1024(%0, %%esi), %%xmm0 \n\t"
|
|
|
302 - "movaps 3072(%0, %%esi), %%xmm3 \n\t" // surround
|
|
|
303 + "movaps 1024(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
304 + "movaps 3072(%0, %%"REG_S"), %%xmm3\n\t" // surround
|
|
|
305 "addps %%xmm7, %%xmm0 \n\t" // common
|
|
|
306 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
307 - "movaps 2048(%0, %%esi), %%xmm2 \n\t"
|
|
|
308 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
309 + "movaps 2048(%0, %%"REG_S"), %%xmm2\n\t"
|
|
|
310 "addps %%xmm0, %%xmm1 \n\t"
|
|
|
311 "addps %%xmm0, %%xmm2 \n\t"
|
|
|
312 "subps %%xmm3, %%xmm1 \n\t"
|
|
|
313 "addps %%xmm3, %%xmm2 \n\t"
|
|
|
314 - "movaps %%xmm1, (%0, %%esi) \n\t"
|
|
|
315 - "movaps %%xmm2, 1024(%0, %%esi) \n\t"
|
|
|
316 - "addl $16, %%esi \n\t"
|
|
|
317 + "movaps %%xmm1, (%0, %%"REG_S") \n\t"
|
|
|
318 + "movaps %%xmm2, 1024(%0, %%"REG_S")\n\t"
|
|
|
319 + "add $16, %%"REG_S" \n\t"
|
|
|
320 " jnz 1b \n\t"
|
|
|
321 :: "r" (samples+256), "m" (bias)
|
|
|
322 - : "%esi"
|
|
|
323 + : "%"REG_S
|
|
|
324 );
|
|
|
325 }
|
|
|
326
|
|
|
327 @@ -899,23 +899,23 @@
|
|
|
328 asm volatile(
|
|
|
329 "movlps %1, %%xmm7 \n\t"
|
|
|
330 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
331 - "movl $-1024, %%esi \n\t"
|
|
|
332 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
333 ".balign 16\n\t"
|
|
|
334 "1: \n\t"
|
|
|
335 - "movaps 2048(%0, %%esi), %%xmm0 \n\t"
|
|
|
336 - "addps 3072(%0, %%esi), %%xmm0 \n\t" // surround
|
|
|
337 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
338 - "movaps 1024(%0, %%esi), %%xmm2 \n\t"
|
|
|
339 + "movaps 2048(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
340 + "addps 3072(%0, %%"REG_S"), %%xmm0\n\t" // surround
|
|
|
341 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
342 + "movaps 1024(%0, %%"REG_S"), %%xmm2\n\t"
|
|
|
343 "addps %%xmm7, %%xmm1 \n\t"
|
|
|
344 "addps %%xmm7, %%xmm2 \n\t"
|
|
|
345 "subps %%xmm0, %%xmm1 \n\t"
|
|
|
346 "addps %%xmm0, %%xmm2 \n\t"
|
|
|
347 - "movaps %%xmm1, (%0, %%esi) \n\t"
|
|
|
348 - "movaps %%xmm2, 1024(%0, %%esi) \n\t"
|
|
|
349 - "addl $16, %%esi \n\t"
|
|
|
350 + "movaps %%xmm1, (%0, %%"REG_S") \n\t"
|
|
|
351 + "movaps %%xmm2, 1024(%0, %%"REG_S")\n\t"
|
|
|
352 + "add $16, %%"REG_S" \n\t"
|
|
|
353 " jnz 1b \n\t"
|
|
|
354 :: "r" (samples+256), "m" (bias)
|
|
|
355 - : "%esi"
|
|
|
356 + : "%"REG_S
|
|
|
357 );
|
|
|
358 }
|
|
|
359
|
|
|
360 @@ -924,22 +924,22 @@
|
|
|
361 asm volatile(
|
|
|
362 "movlps %1, %%xmm7 \n\t"
|
|
|
363 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
364 - "movl $-1024, %%esi \n\t"
|
|
|
365 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
366 ".balign 16\n\t"
|
|
|
367 "1: \n\t"
|
|
|
368 - "movaps 1024(%0, %%esi), %%xmm0 \n\t"
|
|
|
369 + "movaps 1024(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
370 "addps %%xmm7, %%xmm0 \n\t" // common
|
|
|
371 "movaps %%xmm0, %%xmm1 \n\t" // common
|
|
|
372 - "addps (%0, %%esi), %%xmm0 \n\t"
|
|
|
373 - "addps 2048(%0, %%esi), %%xmm1 \n\t"
|
|
|
374 - "addps 3072(%0, %%esi), %%xmm0 \n\t"
|
|
|
375 - "addps 4096(%0, %%esi), %%xmm1 \n\t"
|
|
|
376 - "movaps %%xmm0, (%0, %%esi) \n\t"
|
|
|
377 - "movaps %%xmm1, 1024(%0, %%esi) \n\t"
|
|
|
378 - "addl $16, %%esi \n\t"
|
|
|
379 + "addps (%0, %%"REG_S"), %%xmm0 \n\t"
|
|
|
380 + "addps 2048(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
381 + "addps 3072(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
382 + "addps 4096(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
383 + "movaps %%xmm0, (%0, %%"REG_S") \n\t"
|
|
|
384 + "movaps %%xmm1, 1024(%0, %%"REG_S")\n\t"
|
|
|
385 + "add $16, %%"REG_S" \n\t"
|
|
|
386 " jnz 1b \n\t"
|
|
|
387 :: "r" (samples+256), "m" (bias)
|
|
|
388 - : "%esi"
|
|
|
389 + : "%"REG_S
|
|
|
390 );
|
|
|
391 }
|
|
|
392
|
|
|
393 @@ -948,25 +948,25 @@
|
|
|
394 asm volatile(
|
|
|
395 "movlps %1, %%xmm7 \n\t"
|
|
|
396 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
397 - "movl $-1024, %%esi \n\t"
|
|
|
398 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
399 ".balign 16\n\t"
|
|
|
400 "1: \n\t"
|
|
|
401 - "movaps 1024(%0, %%esi), %%xmm0 \n\t"
|
|
|
402 - "movaps 3072(%0, %%esi), %%xmm2 \n\t"
|
|
|
403 + "movaps 1024(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
404 + "movaps 3072(%0, %%"REG_S"), %%xmm2\n\t"
|
|
|
405 "addps %%xmm7, %%xmm0 \n\t" // common
|
|
|
406 - "addps 4096(%0, %%esi), %%xmm2 \n\t" // surround
|
|
|
407 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
408 - "movaps 2048(%0, %%esi), %%xmm3 \n\t"
|
|
|
409 + "addps 4096(%0, %%"REG_S"), %%xmm2\n\t" // surround
|
|
|
410 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
411 + "movaps 2048(%0, %%"REG_S"), %%xmm3\n\t"
|
|
|
412 "subps %%xmm2, %%xmm1 \n\t"
|
|
|
413 "addps %%xmm2, %%xmm3 \n\t"
|
|
|
414 "addps %%xmm0, %%xmm1 \n\t"
|
|
|
415 "addps %%xmm0, %%xmm3 \n\t"
|
|
|
416 - "movaps %%xmm1, (%0, %%esi) \n\t"
|
|
|
417 - "movaps %%xmm3, 1024(%0, %%esi) \n\t"
|
|
|
418 - "addl $16, %%esi \n\t"
|
|
|
419 + "movaps %%xmm1, (%0, %%"REG_S") \n\t"
|
|
|
420 + "movaps %%xmm3, 1024(%0, %%"REG_S")\n\t"
|
|
|
421 + "add $16, %%"REG_S" \n\t"
|
|
|
422 " jnz 1b \n\t"
|
|
|
423 :: "r" (samples+256), "m" (bias)
|
|
|
424 - : "%esi"
|
|
|
425 + : "%"REG_S
|
|
|
426 );
|
|
|
427 }
|
|
|
428
|
|
|
429 @@ -975,40 +975,40 @@
|
|
|
430 asm volatile(
|
|
|
431 "movlps %2, %%xmm7 \n\t"
|
|
|
432 "shufps $0x00, %%xmm7, %%xmm7 \n\t"
|
|
|
433 - "movl $-1024, %%esi \n\t"
|
|
|
434 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
435 ".balign 16\n\t"
|
|
|
436 "1: \n\t"
|
|
|
437 - "movaps (%0, %%esi), %%xmm0 \n\t"
|
|
|
438 - "movaps 16(%0, %%esi), %%xmm1 \n\t"
|
|
|
439 - "addps 1024(%0, %%esi), %%xmm0 \n\t"
|
|
|
440 - "addps 1040(%0, %%esi), %%xmm1 \n\t"
|
|
|
441 + "movaps (%0, %%"REG_S"), %%xmm0 \n\t"
|
|
|
442 + "movaps 16(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
443 + "addps 1024(%0, %%"REG_S"), %%xmm0\n\t"
|
|
|
444 + "addps 1040(%0, %%"REG_S"), %%xmm1\n\t"
|
|
|
445 "addps %%xmm7, %%xmm0 \n\t"
|
|
|
446 "addps %%xmm7, %%xmm1 \n\t"
|
|
|
447 - "movaps %%xmm0, (%1, %%esi) \n\t"
|
|
|
448 - "movaps %%xmm1, 16(%1, %%esi) \n\t"
|
|
|
449 - "addl $32, %%esi \n\t"
|
|
|
450 + "movaps %%xmm0, (%1, %%"REG_S") \n\t"
|
|
|
451 + "movaps %%xmm1, 16(%1, %%"REG_S")\n\t"
|
|
|
452 + "add $32, %%"REG_S" \n\t"
|
|
|
453 " jnz 1b \n\t"
|
|
|
454 :: "r" (src+256), "r" (dest+256), "m" (bias)
|
|
|
455 - : "%esi"
|
|
|
456 + : "%"REG_S
|
|
|
457 );
|
|
|
458 }
|
|
|
459
|
|
|
460 static void zero_MMX(sample_t * samples)
|
|
|
461 {
|
|
|
462 asm volatile(
|
|
|
463 - "movl $-1024, %%esi \n\t"
|
|
|
464 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
465 "pxor %%mm0, %%mm0 \n\t"
|
|
|
466 ".balign 16\n\t"
|
|
|
467 "1: \n\t"
|
|
|
468 - "movq %%mm0, (%0, %%esi) \n\t"
|
|
|
469 - "movq %%mm0, 8(%0, %%esi) \n\t"
|
|
|
470 - "movq %%mm0, 16(%0, %%esi) \n\t"
|
|
|
471 - "movq %%mm0, 24(%0, %%esi) \n\t"
|
|
|
472 - "addl $32, %%esi \n\t"
|
|
|
473 + "movq %%mm0, (%0, %%"REG_S") \n\t"
|
|
|
474 + "movq %%mm0, 8(%0, %%"REG_S") \n\t"
|
|
|
475 + "movq %%mm0, 16(%0, %%"REG_S") \n\t"
|
|
|
476 + "movq %%mm0, 24(%0, %%"REG_S") \n\t"
|
|
|
477 + "add $32, %%"REG_S" \n\t"
|
|
|
478 " jnz 1b \n\t"
|
|
|
479 "emms"
|
|
|
480 :: "r" (samples+256)
|
|
|
481 - : "%esi"
|
|
|
482 + : "%"REG_S
|
|
|
483 );
|
|
|
484 }
|
|
|
485
|
|
|
486 @@ -1257,29 +1257,29 @@
|
|
|
487 asm volatile(
|
|
|
488 "movd %2, %%mm7 \n\t"
|
|
|
489 "punpckldq %2, %%mm7 \n\t"
|
|
|
490 - "movl $-1024, %%esi \n\t"
|
|
|
491 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
492 ".balign 16\n\t"
|
|
|
493 "1: \n\t"
|
|
|
494 - "movq (%0, %%esi), %%mm0 \n\t"
|
|
|
495 - "movq 8(%0, %%esi), %%mm1 \n\t"
|
|
|
496 - "movq 16(%0, %%esi), %%mm2 \n\t"
|
|
|
497 - "movq 24(%0, %%esi), %%mm3 \n\t"
|
|
|
498 - "pfadd (%1, %%esi), %%mm0 \n\t"
|
|
|
499 - "pfadd 8(%1, %%esi), %%mm1 \n\t"
|
|
|
500 - "pfadd 16(%1, %%esi), %%mm2 \n\t"
|
|
|
501 - "pfadd 24(%1, %%esi), %%mm3 \n\t"
|
|
|
502 + "movq (%0, %%"REG_S"), %%mm0 \n\t"
|
|
|
503 + "movq 8(%0, %%"REG_S"), %%mm1 \n\t"
|
|
|
504 + "movq 16(%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
505 + "movq 24(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
506 + "pfadd (%1, %%"REG_S"), %%mm0 \n\t"
|
|
|
507 + "pfadd 8(%1, %%"REG_S"), %%mm1 \n\t"
|
|
|
508 + "pfadd 16(%1, %%"REG_S"), %%mm2 \n\t"
|
|
|
509 + "pfadd 24(%1, %%"REG_S"), %%mm3 \n\t"
|
|
|
510 "pfadd %%mm7, %%mm0 \n\t"
|
|
|
511 "pfadd %%mm7, %%mm1 \n\t"
|
|
|
512 "pfadd %%mm7, %%mm2 \n\t"
|
|
|
513 "pfadd %%mm7, %%mm3 \n\t"
|
|
|
514 - "movq %%mm0, (%1, %%esi) \n\t"
|
|
|
515 - "movq %%mm1, 8(%1, %%esi) \n\t"
|
|
|
516 - "movq %%mm2, 16(%1, %%esi) \n\t"
|
|
|
517 - "movq %%mm3, 24(%1, %%esi) \n\t"
|
|
|
518 - "addl $32, %%esi \n\t"
|
|
|
519 + "movq %%mm0, (%1, %%"REG_S") \n\t"
|
|
|
520 + "movq %%mm1, 8(%1, %%"REG_S") \n\t"
|
|
|
521 + "movq %%mm2, 16(%1, %%"REG_S") \n\t"
|
|
|
522 + "movq %%mm3, 24(%1, %%"REG_S") \n\t"
|
|
|
523 + "add $32, %%"REG_S" \n\t"
|
|
|
524 " jnz 1b \n\t"
|
|
|
525 :: "r" (src+256), "r" (dest+256), "m" (bias)
|
|
|
526 - : "%esi"
|
|
|
527 + : "%"REG_S
|
|
|
528 );
|
|
|
529 }
|
|
|
530
|
|
|
531 @@ -1288,25 +1288,25 @@
|
|
|
532 asm volatile(
|
|
|
533 "movd %1, %%mm7 \n\t"
|
|
|
534 "punpckldq %1, %%mm7 \n\t"
|
|
|
535 - "movl $-1024, %%esi \n\t"
|
|
|
536 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
537 ".balign 16\n\t"
|
|
|
538 "1: \n\t"
|
|
|
539 - "movq (%0, %%esi), %%mm0 \n\t"
|
|
|
540 - "movq 8(%0, %%esi), %%mm1 \n\t"
|
|
|
541 - "movq 1024(%0, %%esi), %%mm2 \n\t"
|
|
|
542 - "movq 1032(%0, %%esi), %%mm3 \n\t"
|
|
|
543 - "pfadd 2048(%0, %%esi), %%mm0 \n\t"
|
|
|
544 - "pfadd 2056(%0, %%esi), %%mm1 \n\t"
|
|
|
545 + "movq (%0, %%"REG_S"), %%mm0 \n\t"
|
|
|
546 + "movq 8(%0, %%"REG_S"), %%mm1 \n\t"
|
|
|
547 + "movq 1024(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
548 + "movq 1032(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
549 + "pfadd 2048(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
550 + "pfadd 2056(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
551 "pfadd %%mm7, %%mm0 \n\t"
|
|
|
552 "pfadd %%mm7, %%mm1 \n\t"
|
|
|
553 "pfadd %%mm2, %%mm0 \n\t"
|
|
|
554 "pfadd %%mm3, %%mm1 \n\t"
|
|
|
555 - "movq %%mm0, (%0, %%esi) \n\t"
|
|
|
556 - "movq %%mm1, 8(%0, %%esi) \n\t"
|
|
|
557 - "addl $16, %%esi \n\t"
|
|
|
558 + "movq %%mm0, (%0, %%"REG_S") \n\t"
|
|
|
559 + "movq %%mm1, 8(%0, %%"REG_S") \n\t"
|
|
|
560 + "add $16, %%"REG_S" \n\t"
|
|
|
561 " jnz 1b \n\t"
|
|
|
562 :: "r" (samples+256), "m" (bias)
|
|
|
563 - : "%esi"
|
|
|
564 + : "%"REG_S
|
|
|
565 );
|
|
|
566 }
|
|
|
567
|
|
|
568 @@ -1315,27 +1315,27 @@
|
|
|
569 asm volatile(
|
|
|
570 "movd %1, %%mm7 \n\t"
|
|
|
571 "punpckldq %1, %%mm7 \n\t"
|
|
|
572 - "movl $-1024, %%esi \n\t"
|
|
|
573 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
574 ".balign 16\n\t"
|
|
|
575 "1: \n\t"
|
|
|
576 - "movq (%0, %%esi), %%mm0 \n\t"
|
|
|
577 - "movq 8(%0, %%esi), %%mm1 \n\t"
|
|
|
578 - "movq 1024(%0, %%esi), %%mm2 \n\t"
|
|
|
579 - "movq 1032(%0, %%esi), %%mm3 \n\t"
|
|
|
580 - "pfadd 2048(%0, %%esi), %%mm0 \n\t"
|
|
|
581 - "pfadd 2056(%0, %%esi), %%mm1 \n\t"
|
|
|
582 - "pfadd 3072(%0, %%esi), %%mm2 \n\t"
|
|
|
583 - "pfadd 3080(%0, %%esi), %%mm3 \n\t"
|
|
|
584 + "movq (%0, %%"REG_S"), %%mm0 \n\t"
|
|
|
585 + "movq 8(%0, %%"REG_S"), %%mm1 \n\t"
|
|
|
586 + "movq 1024(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
587 + "movq 1032(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
588 + "pfadd 2048(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
589 + "pfadd 2056(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
590 + "pfadd 3072(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
591 + "pfadd 3080(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
592 "pfadd %%mm7, %%mm0 \n\t"
|
|
|
593 "pfadd %%mm7, %%mm1 \n\t"
|
|
|
594 "pfadd %%mm2, %%mm0 \n\t"
|
|
|
595 "pfadd %%mm3, %%mm1 \n\t"
|
|
|
596 - "movq %%mm0, (%0, %%esi) \n\t"
|
|
|
597 - "movq %%mm1, 8(%0, %%esi) \n\t"
|
|
|
598 - "addl $16, %%esi \n\t"
|
|
|
599 + "movq %%mm0, (%0, %%"REG_S") \n\t"
|
|
|
600 + "movq %%mm1, 8(%0, %%"REG_S") \n\t"
|
|
|
601 + "add $16, %%"REG_S" \n\t"
|
|
|
602 " jnz 1b \n\t"
|
|
|
603 :: "r" (samples+256), "m" (bias)
|
|
|
604 - : "%esi"
|
|
|
605 + : "%"REG_S
|
|
|
606 );
|
|
|
607 }
|
|
|
608
|
|
|
609 @@ -1344,29 +1344,29 @@
|
|
|
610 asm volatile(
|
|
|
611 "movd %1, %%mm7 \n\t"
|
|
|
612 "punpckldq %1, %%mm7 \n\t"
|
|
|
613 - "movl $-1024, %%esi \n\t"
|
|
|
614 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
615 ".balign 16\n\t"
|
|
|
616 "1: \n\t"
|
|
|
617 - "movq (%0, %%esi), %%mm0 \n\t"
|
|
|
618 - "movq 8(%0, %%esi), %%mm1 \n\t"
|
|
|
619 - "movq 1024(%0, %%esi), %%mm2 \n\t"
|
|
|
620 - "movq 1032(%0, %%esi), %%mm3 \n\t"
|
|
|
621 - "pfadd 2048(%0, %%esi), %%mm0 \n\t"
|
|
|
622 - "pfadd 2056(%0, %%esi), %%mm1 \n\t"
|
|
|
623 - "pfadd 3072(%0, %%esi), %%mm2 \n\t"
|
|
|
624 - "pfadd 3080(%0, %%esi), %%mm3 \n\t"
|
|
|
625 + "movq (%0, %%"REG_S"), %%mm0 \n\t"
|
|
|
626 + "movq 8(%0, %%"REG_S"), %%mm1 \n\t"
|
|
|
627 + "movq 1024(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
628 + "movq 1032(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
629 + "pfadd 2048(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
630 + "pfadd 2056(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
631 + "pfadd 3072(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
632 + "pfadd 3080(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
633 "pfadd %%mm7, %%mm0 \n\t"
|
|
|
634 "pfadd %%mm7, %%mm1 \n\t"
|
|
|
635 - "pfadd 4096(%0, %%esi), %%mm2 \n\t"
|
|
|
636 - "pfadd 4104(%0, %%esi), %%mm3 \n\t"
|
|
|
637 + "pfadd 4096(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
638 + "pfadd 4104(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
639 "pfadd %%mm2, %%mm0 \n\t"
|
|
|
640 "pfadd %%mm3, %%mm1 \n\t"
|
|
|
641 - "movq %%mm0, (%0, %%esi) \n\t"
|
|
|
642 - "movq %%mm1, 8(%0, %%esi) \n\t"
|
|
|
643 - "addl $16, %%esi \n\t"
|
|
|
644 + "movq %%mm0, (%0, %%"REG_S") \n\t"
|
|
|
645 + "movq %%mm1, 8(%0, %%"REG_S") \n\t"
|
|
|
646 + "add $16, %%"REG_S" \n\t"
|
|
|
647 " jnz 1b \n\t"
|
|
|
648 :: "r" (samples+256), "m" (bias)
|
|
|
649 - : "%esi"
|
|
|
650 + : "%"REG_S
|
|
|
651 );
|
|
|
652 }
|
|
|
653
|
|
|
654 @@ -1375,29 +1375,29 @@
|
|
|
655 asm volatile(
|
|
|
656 "movd %1, %%mm7 \n\t"
|
|
|
657 "punpckldq %1, %%mm7 \n\t"
|
|
|
658 - "movl $-1024, %%esi \n\t"
|
|
|
659 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
660 ".balign 16\n\t"
|
|
|
661 "1: \n\t"
|
|
|
662 - "movq 1024(%0, %%esi), %%mm0 \n\t"
|
|
|
663 - "movq 1032(%0, %%esi), %%mm1 \n\t"
|
|
|
664 + "movq 1024(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
665 + "movq 1032(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
666 "pfadd %%mm7, %%mm0 \n\t" //common
|
|
|
667 "pfadd %%mm7, %%mm1 \n\t" //common
|
|
|
668 - "movq (%0, %%esi), %%mm2 \n\t"
|
|
|
669 - "movq 8(%0, %%esi), %%mm3 \n\t"
|
|
|
670 - "movq 2048(%0, %%esi), %%mm4 \n\t"
|
|
|
671 - "movq 2056(%0, %%esi), %%mm5 \n\t"
|
|
|
672 + "movq (%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
673 + "movq 8(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
674 + "movq 2048(%0, %%"REG_S"), %%mm4\n\t"
|
|
|
675 + "movq 2056(%0, %%"REG_S"), %%mm5\n\t"
|
|
|
676 "pfadd %%mm0, %%mm2 \n\t"
|
|
|
677 "pfadd %%mm1, %%mm3 \n\t"
|
|
|
678 "pfadd %%mm0, %%mm4 \n\t"
|
|
|
679 "pfadd %%mm1, %%mm5 \n\t"
|
|
|
680 - "movq %%mm2, (%0, %%esi) \n\t"
|
|
|
681 - "movq %%mm3, 8(%0, %%esi) \n\t"
|
|
|
682 - "movq %%mm4, 1024(%0, %%esi) \n\t"
|
|
|
683 - "movq %%mm5, 1032(%0, %%esi) \n\t"
|
|
|
684 - "addl $16, %%esi \n\t"
|
|
|
685 + "movq %%mm2, (%0, %%"REG_S") \n\t"
|
|
|
686 + "movq %%mm3, 8(%0, %%"REG_S") \n\t"
|
|
|
687 + "movq %%mm4, 1024(%0, %%"REG_S")\n\t"
|
|
|
688 + "movq %%mm5, 1032(%0, %%"REG_S")\n\t"
|
|
|
689 + "add $16, %%"REG_S" \n\t"
|
|
|
690 " jnz 1b \n\t"
|
|
|
691 :: "r" (samples+256), "m" (bias)
|
|
|
692 - : "%esi"
|
|
|
693 + : "%"REG_S
|
|
|
694 );
|
|
|
695 }
|
|
|
696
|
|
|
697 @@ -1406,29 +1406,29 @@
|
|
|
698 asm volatile(
|
|
|
699 "movd %2, %%mm7 \n\t"
|
|
|
700 "punpckldq %2, %%mm7 \n\t"
|
|
|
701 - "movl $-1024, %%esi \n\t"
|
|
|
702 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
703 ".balign 16\n\t"
|
|
|
704 "1: \n\t"
|
|
|
705 - "movq 1024(%1, %%esi), %%mm0 \n\t"
|
|
|
706 - "movq 1032(%1, %%esi), %%mm1 \n\t"
|
|
|
707 + "movq 1024(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
708 + "movq 1032(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
709 "pfadd %%mm7, %%mm0 \n\t" //common
|
|
|
710 "pfadd %%mm7, %%mm1 \n\t" //common
|
|
|
711 - "movq (%0, %%esi), %%mm2 \n\t"
|
|
|
712 - "movq 8(%0, %%esi), %%mm3 \n\t"
|
|
|
713 - "movq (%1, %%esi), %%mm4 \n\t"
|
|
|
714 - "movq 8(%1, %%esi), %%mm5 \n\t"
|
|
|
715 + "movq (%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
716 + "movq 8(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
717 + "movq (%1, %%"REG_S"), %%mm4 \n\t"
|
|
|
718 + "movq 8(%1, %%"REG_S"), %%mm5 \n\t"
|
|
|
719 "pfadd %%mm0, %%mm2 \n\t"
|
|
|
720 "pfadd %%mm1, %%mm3 \n\t"
|
|
|
721 "pfadd %%mm0, %%mm4 \n\t"
|
|
|
722 "pfadd %%mm1, %%mm5 \n\t"
|
|
|
723 - "movq %%mm2, (%0, %%esi) \n\t"
|
|
|
724 - "movq %%mm3, 8(%0, %%esi) \n\t"
|
|
|
725 - "movq %%mm4, (%1, %%esi) \n\t"
|
|
|
726 - "movq %%mm5, 8(%1, %%esi) \n\t"
|
|
|
727 - "addl $16, %%esi \n\t"
|
|
|
728 + "movq %%mm2, (%0, %%"REG_S") \n\t"
|
|
|
729 + "movq %%mm3, 8(%0, %%"REG_S") \n\t"
|
|
|
730 + "movq %%mm4, (%1, %%"REG_S") \n\t"
|
|
|
731 + "movq %%mm5, 8(%1, %%"REG_S") \n\t"
|
|
|
732 + "add $16, %%"REG_S" \n\t"
|
|
|
733 " jnz 1b \n\t"
|
|
|
734 :: "r" (left+256), "r" (right+256), "m" (bias)
|
|
|
735 - : "%esi"
|
|
|
736 + : "%"REG_S
|
|
|
737 );
|
|
|
738 }
|
|
|
739
|
|
|
740 @@ -1437,15 +1437,15 @@
|
|
|
741 asm volatile(
|
|
|
742 "movd %1, %%mm7 \n\t"
|
|
|
743 "punpckldq %1, %%mm7 \n\t"
|
|
|
744 - "movl $-1024, %%esi \n\t"
|
|
|
745 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
746 ".balign 16\n\t"
|
|
|
747 "1: \n\t"
|
|
|
748 - "movq 2048(%0, %%esi), %%mm0 \n\t" // surround
|
|
|
749 - "movq 2056(%0, %%esi), %%mm1 \n\t" // surround
|
|
|
750 - "movq (%0, %%esi), %%mm2 \n\t"
|
|
|
751 - "movq 8(%0, %%esi), %%mm3 \n\t"
|
|
|
752 - "movq 1024(%0, %%esi), %%mm4 \n\t"
|
|
|
753 - "movq 1032(%0, %%esi), %%mm5 \n\t"
|
|
|
754 + "movq 2048(%0, %%"REG_S"), %%mm0\n\t" // surround
|
|
|
755 + "movq 2056(%0, %%"REG_S"), %%mm1\n\t" // surround
|
|
|
756 + "movq (%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
757 + "movq 8(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
758 + "movq 1024(%0, %%"REG_S"), %%mm4\n\t"
|
|
|
759 + "movq 1032(%0, %%"REG_S"), %%mm5\n\t"
|
|
|
760 "pfadd %%mm7, %%mm2 \n\t"
|
|
|
761 "pfadd %%mm7, %%mm3 \n\t"
|
|
|
762 "pfadd %%mm7, %%mm4 \n\t"
|
|
|
763 @@ -1454,14 +1454,14 @@
|
|
|
764 "pfsub %%mm1, %%mm3 \n\t"
|
|
|
765 "pfadd %%mm0, %%mm4 \n\t"
|
|
|
766 "pfadd %%mm1, %%mm5 \n\t"
|
|
|
767 - "movq %%mm2, (%0, %%esi) \n\t"
|
|
|
768 - "movq %%mm3, 8(%0, %%esi) \n\t"
|
|
|
769 - "movq %%mm4, 1024(%0, %%esi) \n\t"
|
|
|
770 - "movq %%mm5, 1032(%0, %%esi) \n\t"
|
|
|
771 - "addl $16, %%esi \n\t"
|
|
|
772 + "movq %%mm2, (%0, %%"REG_S") \n\t"
|
|
|
773 + "movq %%mm3, 8(%0, %%"REG_S") \n\t"
|
|
|
774 + "movq %%mm4, 1024(%0, %%"REG_S")\n\t"
|
|
|
775 + "movq %%mm5, 1032(%0, %%"REG_S")\n\t"
|
|
|
776 + "add $16, %%"REG_S" \n\t"
|
|
|
777 " jnz 1b \n\t"
|
|
|
778 :: "r" (samples+256), "m" (bias)
|
|
|
779 - : "%esi"
|
|
|
780 + : "%"REG_S
|
|
|
781 );
|
|
|
782 }
|
|
|
783
|
|
|
784 @@ -1470,31 +1470,31 @@
|
|
|
785 asm volatile(
|
|
|
786 "movd %1, %%mm7 \n\t"
|
|
|
787 "punpckldq %1, %%mm7 \n\t"
|
|
|
788 - "movl $-1024, %%esi \n\t"
|
|
|
789 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
790 ".balign 16\n\t"
|
|
|
791 "1: \n\t"
|
|
|
792 - "movq 1024(%0, %%esi), %%mm0 \n\t"
|
|
|
793 - "movq 1032(%0, %%esi), %%mm1 \n\t"
|
|
|
794 - "pfadd 3072(%0, %%esi), %%mm0 \n\t"
|
|
|
795 - "pfadd 3080(%0, %%esi), %%mm1 \n\t"
|
|
|
796 + "movq 1024(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
797 + "movq 1032(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
798 + "pfadd 3072(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
799 + "pfadd 3080(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
800 "pfadd %%mm7, %%mm0 \n\t" // common
|
|
|
801 "pfadd %%mm7, %%mm1 \n\t" // common
|
|
|
802 - "movq (%0, %%esi), %%mm2 \n\t"
|
|
|
803 - "movq 8(%0, %%esi), %%mm3 \n\t"
|
|
|
804 - "movq 2048(%0, %%esi), %%mm4 \n\t"
|
|
|
805 - "movq 2056(%0, %%esi), %%mm5 \n\t"
|
|
|
806 + "movq (%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
807 + "movq 8(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
808 + "movq 2048(%0, %%"REG_S"), %%mm4\n\t"
|
|
|
809 + "movq 2056(%0, %%"REG_S"), %%mm5\n\t"
|
|
|
810 "pfadd %%mm0, %%mm2 \n\t"
|
|
|
811 "pfadd %%mm1, %%mm3 \n\t"
|
|
|
812 "pfadd %%mm0, %%mm4 \n\t"
|
|
|
813 "pfadd %%mm1, %%mm5 \n\t"
|
|
|
814 - "movq %%mm2, (%0, %%esi) \n\t"
|
|
|
815 - "movq %%mm3, 8(%0, %%esi) \n\t"
|
|
|
816 - "movq %%mm4, 1024(%0, %%esi) \n\t"
|
|
|
817 - "movq %%mm5, 1032(%0, %%esi) \n\t"
|
|
|
818 - "addl $16, %%esi \n\t"
|
|
|
819 + "movq %%mm2, (%0, %%"REG_S") \n\t"
|
|
|
820 + "movq %%mm3, 8(%0, %%"REG_S") \n\t"
|
|
|
821 + "movq %%mm4, 1024(%0, %%"REG_S")\n\t"
|
|
|
822 + "movq %%mm5, 1032(%0, %%"REG_S")\n\t"
|
|
|
823 + "add $16, %%"REG_S" \n\t"
|
|
|
824 " jnz 1b \n\t"
|
|
|
825 :: "r" (samples+256), "m" (bias)
|
|
|
826 - : "%esi"
|
|
|
827 + : "%"REG_S
|
|
|
828 );
|
|
|
829 }
|
|
|
830
|
|
|
831 @@ -1503,35 +1503,35 @@
|
|
|
832 asm volatile(
|
|
|
833 "movd %1, %%mm7 \n\t"
|
|
|
834 "punpckldq %1, %%mm7 \n\t"
|
|
|
835 - "movl $-1024, %%esi \n\t"
|
|
|
836 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
837 ".balign 16\n\t"
|
|
|
838 "1: \n\t"
|
|
|
839 - "movq 1024(%0, %%esi), %%mm0 \n\t"
|
|
|
840 - "movq 1032(%0, %%esi), %%mm1 \n\t"
|
|
|
841 + "movq 1024(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
842 + "movq 1032(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
843 "pfadd %%mm7, %%mm0 \n\t" // common
|
|
|
844 "pfadd %%mm7, %%mm1 \n\t" // common
|
|
|
845 - "movq (%0, %%esi), %%mm2 \n\t"
|
|
|
846 - "movq 8(%0, %%esi), %%mm3 \n\t"
|
|
|
847 - "movq 2048(%0, %%esi), %%mm4 \n\t"
|
|
|
848 - "movq 2056(%0, %%esi), %%mm5 \n\t"
|
|
|
849 + "movq (%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
850 + "movq 8(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
851 + "movq 2048(%0, %%"REG_S"), %%mm4\n\t"
|
|
|
852 + "movq 2056(%0, %%"REG_S"), %%mm5\n\t"
|
|
|
853 "pfadd %%mm0, %%mm2 \n\t"
|
|
|
854 "pfadd %%mm1, %%mm3 \n\t"
|
|
|
855 "pfadd %%mm0, %%mm4 \n\t"
|
|
|
856 "pfadd %%mm1, %%mm5 \n\t"
|
|
|
857 - "movq 3072(%0, %%esi), %%mm0 \n\t" // surround
|
|
|
858 - "movq 3080(%0, %%esi), %%mm1 \n\t" // surround
|
|
|
859 + "movq 3072(%0, %%"REG_S"), %%mm0\n\t" // surround
|
|
|
860 + "movq 3080(%0, %%"REG_S"), %%mm1\n\t" // surround
|
|
|
861 "pfsub %%mm0, %%mm2 \n\t"
|
|
|
862 "pfsub %%mm1, %%mm3 \n\t"
|
|
|
863 "pfadd %%mm0, %%mm4 \n\t"
|
|
|
864 "pfadd %%mm1, %%mm5 \n\t"
|
|
|
865 - "movq %%mm2, (%0, %%esi) \n\t"
|
|
|
866 - "movq %%mm3, 8(%0, %%esi) \n\t"
|
|
|
867 - "movq %%mm4, 1024(%0, %%esi) \n\t"
|
|
|
868 - "movq %%mm5, 1032(%0, %%esi) \n\t"
|
|
|
869 - "addl $16, %%esi \n\t"
|
|
|
870 + "movq %%mm2, (%0, %%"REG_S") \n\t"
|
|
|
871 + "movq %%mm3, 8(%0, %%"REG_S") \n\t"
|
|
|
872 + "movq %%mm4, 1024(%0, %%"REG_S")\n\t"
|
|
|
873 + "movq %%mm5, 1032(%0, %%"REG_S")\n\t"
|
|
|
874 + "add $16, %%"REG_S" \n\t"
|
|
|
875 " jnz 1b \n\t"
|
|
|
876 :: "r" (samples+256), "m" (bias)
|
|
|
877 - : "%esi"
|
|
|
878 + : "%"REG_S
|
|
|
879 );
|
|
|
880 }
|
|
|
881
|
|
|
882 @@ -1540,17 +1540,17 @@
|
|
|
883 asm volatile(
|
|
|
884 "movd %1, %%mm7 \n\t"
|
|
|
885 "punpckldq %1, %%mm7 \n\t"
|
|
|
886 - "movl $-1024, %%esi \n\t"
|
|
|
887 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
888 ".balign 16\n\t"
|
|
|
889 "1: \n\t"
|
|
|
890 - "movq 2048(%0, %%esi), %%mm0 \n\t"
|
|
|
891 - "movq 2056(%0, %%esi), %%mm1 \n\t"
|
|
|
892 - "pfadd 3072(%0, %%esi), %%mm0 \n\t" // surround
|
|
|
893 - "pfadd 3080(%0, %%esi), %%mm1 \n\t" // surround
|
|
|
894 - "movq (%0, %%esi), %%mm2 \n\t"
|
|
|
895 - "movq 8(%0, %%esi), %%mm3 \n\t"
|
|
|
896 - "movq 1024(%0, %%esi), %%mm4 \n\t"
|
|
|
897 - "movq 1032(%0, %%esi), %%mm5 \n\t"
|
|
|
898 + "movq 2048(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
899 + "movq 2056(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
900 + "pfadd 3072(%0, %%"REG_S"), %%mm0\n\t" // surround
|
|
|
901 + "pfadd 3080(%0, %%"REG_S"), %%mm1\n\t" // surround
|
|
|
902 + "movq (%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
903 + "movq 8(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
904 + "movq 1024(%0, %%"REG_S"), %%mm4\n\t"
|
|
|
905 + "movq 1032(%0, %%"REG_S"), %%mm5\n\t"
|
|
|
906 "pfadd %%mm7, %%mm2 \n\t"
|
|
|
907 "pfadd %%mm7, %%mm3 \n\t"
|
|
|
908 "pfadd %%mm7, %%mm4 \n\t"
|
|
|
909 @@ -1559,14 +1559,14 @@
|
|
|
910 "pfsub %%mm1, %%mm3 \n\t"
|
|
|
911 "pfadd %%mm0, %%mm4 \n\t"
|
|
|
912 "pfadd %%mm1, %%mm5 \n\t"
|
|
|
913 - "movq %%mm2, (%0, %%esi) \n\t"
|
|
|
914 - "movq %%mm3, 8(%0, %%esi) \n\t"
|
|
|
915 - "movq %%mm4, 1024(%0, %%esi) \n\t"
|
|
|
916 - "movq %%mm5, 1032(%0, %%esi) \n\t"
|
|
|
917 - "addl $16, %%esi \n\t"
|
|
|
918 + "movq %%mm2, (%0, %%"REG_S") \n\t"
|
|
|
919 + "movq %%mm3, 8(%0, %%"REG_S") \n\t"
|
|
|
920 + "movq %%mm4, 1024(%0, %%"REG_S")\n\t"
|
|
|
921 + "movq %%mm5, 1032(%0, %%"REG_S")\n\t"
|
|
|
922 + "add $16, %%"REG_S" \n\t"
|
|
|
923 " jnz 1b \n\t"
|
|
|
924 :: "r" (samples+256), "m" (bias)
|
|
|
925 - : "%esi"
|
|
|
926 + : "%"REG_S
|
|
|
927 );
|
|
|
928 }
|
|
|
929
|
|
|
930 @@ -1575,31 +1575,31 @@
|
|
|
931 asm volatile(
|
|
|
932 "movd %1, %%mm7 \n\t"
|
|
|
933 "punpckldq %1, %%mm7 \n\t"
|
|
|
934 - "movl $-1024, %%esi \n\t"
|
|
|
935 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
936 ".balign 16\n\t"
|
|
|
937 "1: \n\t"
|
|
|
938 - "movq 1024(%0, %%esi), %%mm0 \n\t"
|
|
|
939 - "movq 1032(%0, %%esi), %%mm1 \n\t"
|
|
|
940 + "movq 1024(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
941 + "movq 1032(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
942 "pfadd %%mm7, %%mm0 \n\t" // common
|
|
|
943 "pfadd %%mm7, %%mm1 \n\t" // common
|
|
|
944 "movq %%mm0, %%mm2 \n\t" // common
|
|
|
945 "movq %%mm1, %%mm3 \n\t" // common
|
|
|
946 - "pfadd (%0, %%esi), %%mm0 \n\t"
|
|
|
947 - "pfadd 8(%0, %%esi), %%mm1 \n\t"
|
|
|
948 - "pfadd 2048(%0, %%esi), %%mm2 \n\t"
|
|
|
949 - "pfadd 2056(%0, %%esi), %%mm3 \n\t"
|
|
|
950 - "pfadd 3072(%0, %%esi), %%mm0 \n\t"
|
|
|
951 - "pfadd 3080(%0, %%esi), %%mm1 \n\t"
|
|
|
952 - "pfadd 4096(%0, %%esi), %%mm2 \n\t"
|
|
|
953 - "pfadd 4104(%0, %%esi), %%mm3 \n\t"
|
|
|
954 - "movq %%mm0, (%0, %%esi) \n\t"
|
|
|
955 - "movq %%mm1, 8(%0, %%esi) \n\t"
|
|
|
956 - "movq %%mm2, 1024(%0, %%esi) \n\t"
|
|
|
957 - "movq %%mm3, 1032(%0, %%esi) \n\t"
|
|
|
958 - "addl $16, %%esi \n\t"
|
|
|
959 + "pfadd (%0, %%"REG_S"), %%mm0 \n\t"
|
|
|
960 + "pfadd 8(%0, %%"REG_S"), %%mm1 \n\t"
|
|
|
961 + "pfadd 2048(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
962 + "pfadd 2056(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
963 + "pfadd 3072(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
964 + "pfadd 3080(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
965 + "pfadd 4096(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
966 + "pfadd 4104(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
967 + "movq %%mm0, (%0, %%"REG_S") \n\t"
|
|
|
968 + "movq %%mm1, 8(%0, %%"REG_S") \n\t"
|
|
|
969 + "movq %%mm2, 1024(%0, %%"REG_S")\n\t"
|
|
|
970 + "movq %%mm3, 1032(%0, %%"REG_S")\n\t"
|
|
|
971 + "add $16, %%"REG_S" \n\t"
|
|
|
972 " jnz 1b \n\t"
|
|
|
973 :: "r" (samples+256), "m" (bias)
|
|
|
974 - : "%esi"
|
|
|
975 + : "%"REG_S
|
|
|
976 );
|
|
|
977 }
|
|
|
978
|
|
|
979 @@ -1607,23 +1607,23 @@
|
|
|
980 static void mix32toS_3dnow (sample_t * samples, sample_t bias)
|
|
|
981 {
|
|
|
982 asm volatile(
|
|
|
983 - "movl $-1024, %%esi \n\t"
|
|
|
984 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
985 ".balign 16\n\t"
|
|
|
986 "1: \n\t"
|
|
|
987 "movd %1, %%mm7 \n\t"
|
|
|
988 "punpckldq %1, %%mm7 \n\t"
|
|
|
989 - "movq 1024(%0, %%esi), %%mm0 \n\t"
|
|
|
990 - "movq 1032(%0, %%esi), %%mm1 \n\t"
|
|
|
991 - "movq 3072(%0, %%esi), %%mm4 \n\t"
|
|
|
992 - "movq 3080(%0, %%esi), %%mm5 \n\t"
|
|
|
993 + "movq 1024(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
994 + "movq 1032(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
995 + "movq 3072(%0, %%"REG_S"), %%mm4\n\t"
|
|
|
996 + "movq 3080(%0, %%"REG_S"), %%mm5\n\t"
|
|
|
997 "pfadd %%mm7, %%mm0 \n\t" // common
|
|
|
998 "pfadd %%mm7, %%mm1 \n\t" // common
|
|
|
999 - "pfadd 4096(%0, %%esi), %%mm4 \n\t" // surround
|
|
|
1000 - "pfadd 4104(%0, %%esi), %%mm5 \n\t" // surround
|
|
|
1001 - "movq (%0, %%esi), %%mm2 \n\t"
|
|
|
1002 - "movq 8(%0, %%esi), %%mm3 \n\t"
|
|
|
1003 - "movq 2048(%0, %%esi), %%mm6 \n\t"
|
|
|
1004 - "movq 2056(%0, %%esi), %%mm7 \n\t"
|
|
|
1005 + "pfadd 4096(%0, %%"REG_S"), %%mm4\n\t" // surround
|
|
|
1006 + "pfadd 4104(%0, %%"REG_S"), %%mm5\n\t" // surround
|
|
|
1007 + "movq (%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
1008 + "movq 8(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
1009 + "movq 2048(%0, %%"REG_S"), %%mm6\n\t"
|
|
|
1010 + "movq 2056(%0, %%"REG_S"), %%mm7\n\t"
|
|
|
1011 "pfsub %%mm4, %%mm2 \n\t"
|
|
|
1012 "pfsub %%mm5, %%mm3 \n\t"
|
|
|
1013 "pfadd %%mm4, %%mm6 \n\t"
|
|
|
1014 @@ -1632,14 +1632,14 @@
|
|
|
1015 "pfadd %%mm1, %%mm3 \n\t"
|
|
|
1016 "pfadd %%mm0, %%mm6 \n\t"
|
|
|
1017 "pfadd %%mm1, %%mm7 \n\t"
|
|
|
1018 - "movq %%mm2, (%0, %%esi) \n\t"
|
|
|
1019 - "movq %%mm3, 8(%0, %%esi) \n\t"
|
|
|
1020 - "movq %%mm6, 1024(%0, %%esi) \n\t"
|
|
|
1021 - "movq %%mm7, 1032(%0, %%esi) \n\t"
|
|
|
1022 - "addl $16, %%esi \n\t"
|
|
|
1023 + "movq %%mm2, (%0, %%"REG_S") \n\t"
|
|
|
1024 + "movq %%mm3, 8(%0, %%"REG_S") \n\t"
|
|
|
1025 + "movq %%mm6, 1024(%0, %%"REG_S")\n\t"
|
|
|
1026 + "movq %%mm7, 1032(%0, %%"REG_S")\n\t"
|
|
|
1027 + "add $16, %%"REG_S" \n\t"
|
|
|
1028 " jnz 1b \n\t"
|
|
|
1029 :: "r" (samples+256), "m" (bias)
|
|
|
1030 - : "%esi"
|
|
|
1031 + : "%"REG_S
|
|
|
1032 );
|
|
|
1033 }
|
|
|
1034
|
|
|
1035 @@ -1648,29 +1648,29 @@
|
|
|
1036 asm volatile(
|
|
|
1037 "movd %2, %%mm7 \n\t"
|
|
|
1038 "punpckldq %2, %%mm7 \n\t"
|
|
|
1039 - "movl $-1024, %%esi \n\t"
|
|
|
1040 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1041 ".balign 16\n\t"
|
|
|
1042 "1: \n\t"
|
|
|
1043 - "movq (%0, %%esi), %%mm0 \n\t"
|
|
|
1044 - "movq 8(%0, %%esi), %%mm1 \n\t"
|
|
|
1045 - "movq 16(%0, %%esi), %%mm2 \n\t"
|
|
|
1046 - "movq 24(%0, %%esi), %%mm3 \n\t"
|
|
|
1047 - "pfadd 1024(%0, %%esi), %%mm0 \n\t"
|
|
|
1048 - "pfadd 1032(%0, %%esi), %%mm1 \n\t"
|
|
|
1049 - "pfadd 1040(%0, %%esi), %%mm2 \n\t"
|
|
|
1050 - "pfadd 1048(%0, %%esi), %%mm3 \n\t"
|
|
|
1051 + "movq (%0, %%"REG_S"), %%mm0 \n\t"
|
|
|
1052 + "movq 8(%0, %%"REG_S"), %%mm1 \n\t"
|
|
|
1053 + "movq 16(%0, %%"REG_S"), %%mm2 \n\t"
|
|
|
1054 + "movq 24(%0, %%"REG_S"), %%mm3 \n\t"
|
|
|
1055 + "pfadd 1024(%0, %%"REG_S"), %%mm0\n\t"
|
|
|
1056 + "pfadd 1032(%0, %%"REG_S"), %%mm1\n\t"
|
|
|
1057 + "pfadd 1040(%0, %%"REG_S"), %%mm2\n\t"
|
|
|
1058 + "pfadd 1048(%0, %%"REG_S"), %%mm3\n\t"
|
|
|
1059 "pfadd %%mm7, %%mm0 \n\t"
|
|
|
1060 "pfadd %%mm7, %%mm1 \n\t"
|
|
|
1061 "pfadd %%mm7, %%mm2 \n\t"
|
|
|
1062 "pfadd %%mm7, %%mm3 \n\t"
|
|
|
1063 - "movq %%mm0, (%1, %%esi) \n\t"
|
|
|
1064 - "movq %%mm1, 8(%1, %%esi) \n\t"
|
|
|
1065 - "movq %%mm2, 16(%1, %%esi) \n\t"
|
|
|
1066 - "movq %%mm3, 24(%1, %%esi) \n\t"
|
|
|
1067 - "addl $32, %%esi \n\t"
|
|
|
1068 + "movq %%mm0, (%1, %%"REG_S") \n\t"
|
|
|
1069 + "movq %%mm1, 8(%1, %%"REG_S") \n\t"
|
|
|
1070 + "movq %%mm2, 16(%1, %%"REG_S") \n\t"
|
|
|
1071 + "movq %%mm3, 24(%1, %%"REG_S") \n\t"
|
|
|
1072 + "add $32, %%"REG_S" \n\t"
|
|
|
1073 " jnz 1b \n\t"
|
|
|
1074 :: "r" (src+256), "r" (dest+256), "m" (bias)
|
|
|
1075 - : "%esi"
|
|
|
1076 + : "%"REG_S
|
|
|
1077 );
|
|
|
1078 }
|
|
|
1079
|
|
|
1080 @@ -1816,4 +1816,4 @@
|
|
|
1081 __asm __volatile("femms":::"memory");
|
|
|
1082 }
|
|
|
1083
|
|
|
1084 -#endif //ARCH_X86
|
|
|
1085 +#endif // ARCH_X86 || ARCH_X86_64
|
|
|
1086 Index: liba52/imdct.c
|
|
|
1087 ===================================================================
|
|
|
1088 RCS file: /cvsroot/mplayer/main/liba52/imdct.c,v
|
|
|
1089 retrieving revision 1.27
|
|
|
1090 diff -u -r1.27 imdct.c
|
|
|
1091 --- liba52/imdct.c 2 Jun 2005 20:54:02 -0000 1.27
|
|
|
1092 +++ liba52/imdct.c 31 Jul 2005 21:20:09 -0000
|
|
|
1093 @@ -101,7 +101,7 @@
|
|
|
1094 0x03, 0x23, 0x13, 0x33, 0x0b, 0x2b, 0x1b, 0x3b,
|
|
|
1095 0x07, 0x27, 0x17, 0x37, 0x0f, 0x2f, 0x1f, 0x3f};
|
|
|
1096
|
|
|
1097 -#ifdef ARCH_X86
|
|
|
1098 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
1099 // NOTE: SSE needs 16byte alignment or it will segfault
|
|
|
1100 //
|
|
|
1101 static complex_t __attribute__((aligned(16))) buf[128];
|
|
|
1102 @@ -442,8 +442,8 @@
|
|
|
1103 int k;
|
|
|
1104 int p,q;
|
|
|
1105 int m;
|
|
|
1106 - int two_m;
|
|
|
1107 - int two_m_plus_one;
|
|
|
1108 + long two_m;
|
|
|
1109 + long two_m_plus_one;
|
|
|
1110
|
|
|
1111 sample_t tmp_b_i;
|
|
|
1112 sample_t tmp_b_r;
|
|
|
1113 @@ -747,7 +747,7 @@
|
|
|
1114
|
|
|
1115 // Stuff below this line is borrowed from libac3
|
|
|
1116 #include "srfftp.h"
|
|
|
1117 -#ifdef ARCH_X86
|
|
|
1118 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
1119 #ifndef HAVE_3DNOW
|
|
|
1120 #define HAVE_3DNOW 1
|
|
|
1121 #endif
|
|
|
1122 @@ -768,9 +768,9 @@
|
|
|
1123 /* int i,k;
|
|
|
1124 int p,q;*/
|
|
|
1125 int m;
|
|
|
1126 - int two_m;
|
|
|
1127 - int two_m_plus_one;
|
|
|
1128 - int two_m_plus_one_shl3;
|
|
|
1129 + long two_m;
|
|
|
1130 + long two_m_plus_one;
|
|
|
1131 + long two_m_plus_one_shl3;
|
|
|
1132 complex_t *buf_offset;
|
|
|
1133
|
|
|
1134 /* sample_t tmp_a_i;
|
|
|
1135 @@ -788,33 +788,33 @@
|
|
|
1136 /* Pre IFFT complex multiply plus IFFT cmplx conjugate */
|
|
|
1137 /* Bit reversed shuffling */
|
|
|
1138 asm volatile(
|
|
|
1139 - "xorl %%esi, %%esi \n\t"
|
|
|
1140 - "leal "MANGLE(bit_reverse_512)", %%eax \n\t"
|
|
|
1141 - "movl $1008, %%edi \n\t"
|
|
|
1142 - "pushl %%ebp \n\t" //use ebp without telling gcc
|
|
|
1143 + "xor %%"REG_S", %%"REG_S" \n\t"
|
|
|
1144 + "lea "MANGLE(bit_reverse_512)", %%"REG_a"\n\t"
|
|
|
1145 + "mov $1008, %%"REG_D" \n\t"
|
|
|
1146 + "push %%"REG_BP" \n\t" //use ebp without telling gcc
|
|
|
1147 ".balign 16 \n\t"
|
|
|
1148 "1: \n\t"
|
|
|
1149 - "movlps (%0, %%esi), %%xmm0 \n\t" // XXXI
|
|
|
1150 - "movhps 8(%0, %%edi), %%xmm0 \n\t" // RXXI
|
|
|
1151 - "movlps 8(%0, %%esi), %%xmm1 \n\t" // XXXi
|
|
|
1152 - "movhps (%0, %%edi), %%xmm1 \n\t" // rXXi
|
|
|
1153 + "movlps (%0, %%"REG_S"), %%xmm0 \n\t" // XXXI
|
|
|
1154 + "movhps 8(%0, %%"REG_D"), %%xmm0 \n\t" // RXXI
|
|
|
1155 + "movlps 8(%0, %%"REG_S"), %%xmm1 \n\t" // XXXi
|
|
|
1156 + "movhps (%0, %%"REG_D"), %%xmm1 \n\t" // rXXi
|
|
|
1157 "shufps $0x33, %%xmm1, %%xmm0 \n\t" // irIR
|
|
|
1158 - "movaps "MANGLE(sseSinCos1c)"(%%esi), %%xmm2\n\t"
|
|
|
1159 + "movaps "MANGLE(sseSinCos1c)"(%%"REG_S"), %%xmm2\n\t"
|
|
|
1160 "mulps %%xmm0, %%xmm2 \n\t"
|
|
|
1161 "shufps $0xB1, %%xmm0, %%xmm0 \n\t" // riRI
|
|
|
1162 - "mulps "MANGLE(sseSinCos1d)"(%%esi), %%xmm0\n\t"
|
|
|
1163 + "mulps "MANGLE(sseSinCos1d)"(%%"REG_S"), %%xmm0\n\t"
|
|
|
1164 "subps %%xmm0, %%xmm2 \n\t"
|
|
|
1165 - "movzbl (%%eax), %%edx \n\t"
|
|
|
1166 - "movzbl 1(%%eax), %%ebp \n\t"
|
|
|
1167 - "movlps %%xmm2, (%1, %%edx,8) \n\t"
|
|
|
1168 - "movhps %%xmm2, (%1, %%ebp,8) \n\t"
|
|
|
1169 - "addl $16, %%esi \n\t"
|
|
|
1170 - "addl $2, %%eax \n\t" // avoid complex addressing for P4 crap
|
|
|
1171 - "subl $16, %%edi \n\t"
|
|
|
1172 - " jnc 1b \n\t"
|
|
|
1173 - "popl %%ebp \n\t"//no we didnt touch ebp *g*
|
|
|
1174 - :: "b" (data), "c" (buf)
|
|
|
1175 - : "%esi", "%edi", "%eax", "%edx"
|
|
|
1176 + "movzb (%%"REG_a"), %%"REG_d" \n\t"
|
|
|
1177 + "movzb 1(%%"REG_a"), %%"REG_BP" \n\t"
|
|
|
1178 + "movlps %%xmm2, (%1, %%"REG_d", 8) \n\t"
|
|
|
1179 + "movhps %%xmm2, (%1, %%"REG_BP", 8) \n\t"
|
|
|
1180 + "add $16, %%"REG_S" \n\t"
|
|
|
1181 + "add $2, %%"REG_a" \n\t" // avoid complex addressing for P4 crap
|
|
|
1182 + "sub $16, %%"REG_D" \n\t"
|
|
|
1183 + "jnc 1b \n\t"
|
|
|
1184 + "pop %%"REG_BP" \n\t"//no we didnt touch ebp *g*
|
|
16191
|
1185 + :: "b" (data), "c" (buf)
|
|
16174
|
1186 + : "%"REG_S, "%"REG_D, "%"REG_a, "%"REG_d
|
|
|
1187 );
|
|
|
1188
|
|
|
1189
|
|
|
1190 @@ -850,44 +850,44 @@
|
|
|
1191 asm volatile(
|
|
|
1192 "xorps %%xmm1, %%xmm1 \n\t"
|
|
|
1193 "xorps %%xmm2, %%xmm2 \n\t"
|
|
|
1194 - "movl %0, %%esi \n\t"
|
|
|
1195 + "mov %0, %%"REG_S" \n\t"
|
|
|
1196 ".balign 16 \n\t"
|
|
|
1197 "1: \n\t"
|
|
|
1198 - "movlps (%%esi), %%xmm0 \n\t" //buf[p]
|
|
|
1199 - "movlps 8(%%esi), %%xmm1\n\t" //buf[q]
|
|
|
1200 - "movhps (%%esi), %%xmm0 \n\t" //buf[p]
|
|
|
1201 - "movhps 8(%%esi), %%xmm2\n\t" //buf[q]
|
|
|
1202 + "movlps (%%"REG_S"), %%xmm0\n\t" //buf[p]
|
|
|
1203 + "movlps 8(%%"REG_S"), %%xmm1\n\t" //buf[q]
|
|
|
1204 + "movhps (%%"REG_S"), %%xmm0\n\t" //buf[p]
|
|
|
1205 + "movhps 8(%%"REG_S"), %%xmm2\n\t" //buf[q]
|
|
|
1206 "addps %%xmm1, %%xmm0 \n\t"
|
|
|
1207 "subps %%xmm2, %%xmm0 \n\t"
|
|
|
1208 - "movaps %%xmm0, (%%esi) \n\t"
|
|
|
1209 - "addl $16, %%esi \n\t"
|
|
|
1210 - "cmpl %1, %%esi \n\t"
|
|
|
1211 + "movaps %%xmm0, (%%"REG_S")\n\t"
|
|
|
1212 + "add $16, %%"REG_S" \n\t"
|
|
|
1213 + "cmp %1, %%"REG_S" \n\t"
|
|
|
1214 " jb 1b \n\t"
|
|
|
1215 :: "g" (buf), "r" (buf + 128)
|
|
|
1216 - : "%esi"
|
|
|
1217 + : "%"REG_S
|
|
|
1218 );
|
|
|
1219
|
|
|
1220 /* 2. iteration */
|
|
|
1221 // Note w[1]={{1,0}, {0,-1}}
|
|
|
1222 asm volatile(
|
|
|
1223 "movaps "MANGLE(ps111_1)", %%xmm7\n\t" // 1,1,1,-1
|
|
|
1224 - "movl %0, %%esi \n\t"
|
|
|
1225 + "mov %0, %%"REG_S" \n\t"
|
|
|
1226 ".balign 16 \n\t"
|
|
|
1227 "1: \n\t"
|
|
|
1228 - "movaps 16(%%esi), %%xmm2 \n\t" //r2,i2,r3,i3
|
|
|
1229 + "movaps 16(%%"REG_S"), %%xmm2 \n\t" //r2,i2,r3,i3
|
|
|
1230 "shufps $0xB4, %%xmm2, %%xmm2 \n\t" //r2,i2,i3,r3
|
|
|
1231 "mulps %%xmm7, %%xmm2 \n\t" //r2,i2,i3,-r3
|
|
|
1232 - "movaps (%%esi), %%xmm0 \n\t" //r0,i0,r1,i1
|
|
|
1233 - "movaps (%%esi), %%xmm1 \n\t" //r0,i0,r1,i1
|
|
|
1234 + "movaps (%%"REG_S"), %%xmm0 \n\t" //r0,i0,r1,i1
|
|
|
1235 + "movaps (%%"REG_S"), %%xmm1 \n\t" //r0,i0,r1,i1
|
|
|
1236 "addps %%xmm2, %%xmm0 \n\t"
|
|
|
1237 "subps %%xmm2, %%xmm1 \n\t"
|
|
|
1238 - "movaps %%xmm0, (%%esi) \n\t"
|
|
|
1239 - "movaps %%xmm1, 16(%%esi) \n\t"
|
|
|
1240 - "addl $32, %%esi \n\t"
|
|
|
1241 - "cmpl %1, %%esi \n\t"
|
|
|
1242 + "movaps %%xmm0, (%%"REG_S") \n\t"
|
|
|
1243 + "movaps %%xmm1, 16(%%"REG_S") \n\t"
|
|
|
1244 + "add $32, %%"REG_S" \n\t"
|
|
|
1245 + "cmp %1, %%"REG_S" \n\t"
|
|
|
1246 " jb 1b \n\t"
|
|
|
1247 :: "g" (buf), "r" (buf + 128)
|
|
|
1248 - : "%esi"
|
|
|
1249 + : "%"REG_S
|
|
|
1250 );
|
|
|
1251
|
|
|
1252 /* 3. iteration */
|
|
|
1253 @@ -902,11 +902,11 @@
|
|
|
1254 "movaps 16+"MANGLE(sseW2)", %%xmm7\n\t"
|
|
|
1255 "xorps %%xmm5, %%xmm5 \n\t"
|
|
|
1256 "xorps %%xmm2, %%xmm2 \n\t"
|
|
|
1257 - "movl %0, %%esi \n\t"
|
|
|
1258 + "mov %0, %%"REG_S" \n\t"
|
|
|
1259 ".balign 16 \n\t"
|
|
|
1260 "1: \n\t"
|
|
|
1261 - "movaps 32(%%esi), %%xmm2 \n\t" //r4,i4,r5,i5
|
|
|
1262 - "movaps 48(%%esi), %%xmm3 \n\t" //r6,i6,r7,i7
|
|
|
1263 + "movaps 32(%%"REG_S"), %%xmm2 \n\t" //r4,i4,r5,i5
|
|
|
1264 + "movaps 48(%%"REG_S"), %%xmm3 \n\t" //r6,i6,r7,i7
|
|
|
1265 "movaps "MANGLE(sseW2)", %%xmm4 \n\t" //r4,i4,r5,i5
|
|
|
1266 "movaps 32+"MANGLE(sseW2)", %%xmm5\n\t" //r6,i6,r7,i7
|
|
|
1267 "mulps %%xmm2, %%xmm4 \n\t"
|
|
|
1268 @@ -915,8 +915,8 @@
|
|
|
1269 "shufps $0xB1, %%xmm3, %%xmm3 \n\t" //i6,r6,i7,r7
|
|
|
1270 "mulps %%xmm6, %%xmm3 \n\t"
|
|
|
1271 "mulps %%xmm7, %%xmm2 \n\t"
|
|
|
1272 - "movaps (%%esi), %%xmm0 \n\t" //r0,i0,r1,i1
|
|
|
1273 - "movaps 16(%%esi), %%xmm1 \n\t" //r2,i2,r3,i3
|
|
|
1274 + "movaps (%%"REG_S"), %%xmm0 \n\t" //r0,i0,r1,i1
|
|
|
1275 + "movaps 16(%%"REG_S"), %%xmm1 \n\t" //r2,i2,r3,i3
|
|
|
1276 "addps %%xmm4, %%xmm2 \n\t"
|
|
|
1277 "addps %%xmm5, %%xmm3 \n\t"
|
|
|
1278 "movaps %%xmm2, %%xmm4 \n\t"
|
|
|
1279 @@ -925,15 +925,15 @@
|
|
|
1280 "addps %%xmm1, %%xmm3 \n\t"
|
|
|
1281 "subps %%xmm4, %%xmm0 \n\t"
|
|
|
1282 "subps %%xmm5, %%xmm1 \n\t"
|
|
|
1283 - "movaps %%xmm2, (%%esi) \n\t"
|
|
|
1284 - "movaps %%xmm3, 16(%%esi) \n\t"
|
|
|
1285 - "movaps %%xmm0, 32(%%esi) \n\t"
|
|
|
1286 - "movaps %%xmm1, 48(%%esi) \n\t"
|
|
|
1287 - "addl $64, %%esi \n\t"
|
|
|
1288 - "cmpl %1, %%esi \n\t"
|
|
|
1289 + "movaps %%xmm2, (%%"REG_S") \n\t"
|
|
|
1290 + "movaps %%xmm3, 16(%%"REG_S") \n\t"
|
|
|
1291 + "movaps %%xmm0, 32(%%"REG_S") \n\t"
|
|
|
1292 + "movaps %%xmm1, 48(%%"REG_S") \n\t"
|
|
|
1293 + "add $64, %%"REG_S" \n\t"
|
|
|
1294 + "cmp %1, %%"REG_S" \n\t"
|
|
|
1295 " jb 1b \n\t"
|
|
|
1296 :: "g" (buf), "r" (buf + 128)
|
|
|
1297 - : "%esi"
|
|
|
1298 + : "%"REG_S
|
|
|
1299 );
|
|
|
1300
|
|
|
1301 /* 4-7. iterations */
|
|
|
1302 @@ -943,52 +943,52 @@
|
|
|
1303 two_m_plus_one_shl3 = (two_m_plus_one<<3);
|
|
|
1304 buf_offset = buf+128;
|
|
|
1305 asm volatile(
|
|
|
1306 - "movl %0, %%esi \n\t"
|
|
|
1307 + "mov %0, %%"REG_S" \n\t"
|
|
|
1308 ".balign 16 \n\t"
|
|
|
1309 "1: \n\t"
|
|
|
1310 - "xorl %%edi, %%edi \n\t" // k
|
|
|
1311 - "leal (%%esi, %3), %%edx \n\t"
|
|
|
1312 + "xor %%"REG_D", %%"REG_D" \n\t" // k
|
|
|
1313 + "lea (%%"REG_S", %3), %%"REG_d" \n\t"
|
|
|
1314 "2: \n\t"
|
|
|
1315 - "movaps (%%edx, %%edi), %%xmm1 \n\t"
|
|
|
1316 - "movaps (%4, %%edi, 2), %%xmm2 \n\t"
|
|
|
1317 + "movaps (%%"REG_d", %%"REG_D"), %%xmm1 \n\t"
|
|
|
1318 + "movaps (%4, %%"REG_D", 2), %%xmm2 \n\t"
|
|
|
1319 "mulps %%xmm1, %%xmm2 \n\t"
|
|
|
1320 "shufps $0xB1, %%xmm1, %%xmm1 \n\t"
|
|
|
1321 - "mulps 16(%4, %%edi, 2), %%xmm1 \n\t"
|
|
|
1322 - "movaps (%%esi, %%edi), %%xmm0 \n\t"
|
|
|
1323 + "mulps 16(%4, %%"REG_D", 2), %%xmm1 \n\t"
|
|
|
1324 + "movaps (%%"REG_S", %%"REG_D"), %%xmm0 \n\t"
|
|
|
1325 "addps %%xmm2, %%xmm1 \n\t"
|
|
|
1326 "movaps %%xmm1, %%xmm2 \n\t"
|
|
|
1327 "addps %%xmm0, %%xmm1 \n\t"
|
|
|
1328 "subps %%xmm2, %%xmm0 \n\t"
|
|
|
1329 - "movaps %%xmm1, (%%esi, %%edi) \n\t"
|
|
|
1330 - "movaps %%xmm0, (%%edx, %%edi) \n\t"
|
|
|
1331 - "addl $16, %%edi \n\t"
|
|
|
1332 - "cmpl %3, %%edi \n\t" //FIXME (opt) count against 0
|
|
|
1333 - " jb 2b \n\t"
|
|
|
1334 - "addl %2, %%esi \n\t"
|
|
|
1335 - "cmpl %1, %%esi \n\t"
|
|
|
1336 + "movaps %%xmm1, (%%"REG_S", %%"REG_D") \n\t"
|
|
|
1337 + "movaps %%xmm0, (%%"REG_d", %%"REG_D") \n\t"
|
|
|
1338 + "add $16, %%"REG_D" \n\t"
|
|
|
1339 + "cmp %3, %%"REG_D" \n\t" //FIXME (opt) count against 0
|
|
|
1340 + "jb 2b \n\t"
|
|
|
1341 + "add %2, %%"REG_S" \n\t"
|
|
|
1342 + "cmp %1, %%"REG_S" \n\t"
|
|
|
1343 " jb 1b \n\t"
|
|
|
1344 :: "g" (buf), "m" (buf_offset), "m" (two_m_plus_one_shl3), "r" (two_m<<3),
|
|
|
1345 "r" (sseW[m])
|
|
|
1346 - : "%esi", "%edi", "%edx"
|
|
|
1347 + : "%"REG_S, "%"REG_D, "%"REG_d
|
|
|
1348 );
|
|
|
1349 }
|
|
|
1350
|
|
|
1351 /* Post IFFT complex multiply plus IFFT complex conjugate*/
|
|
|
1352 asm volatile(
|
|
|
1353 - "movl $-1024, %%esi \n\t"
|
|
|
1354 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1355 ".balign 16 \n\t"
|
|
|
1356 "1: \n\t"
|
|
|
1357 - "movaps (%0, %%esi), %%xmm0 \n\t"
|
|
|
1358 - "movaps (%0, %%esi), %%xmm1 \n\t"
|
|
|
1359 + "movaps (%0, %%"REG_S"), %%xmm0 \n\t"
|
|
|
1360 + "movaps (%0, %%"REG_S"), %%xmm1 \n\t"
|
|
|
1361 "shufps $0xB1, %%xmm0, %%xmm0 \n\t"
|
|
|
1362 - "mulps 1024+"MANGLE(sseSinCos1c)"(%%esi), %%xmm1\n\t"
|
|
|
1363 - "mulps 1024+"MANGLE(sseSinCos1d)"(%%esi), %%xmm0\n\t"
|
|
|
1364 + "mulps 1024+"MANGLE(sseSinCos1c)"(%%"REG_S"), %%xmm1\n\t"
|
|
|
1365 + "mulps 1024+"MANGLE(sseSinCos1d)"(%%"REG_S"), %%xmm0\n\t"
|
|
|
1366 "addps %%xmm1, %%xmm0 \n\t"
|
|
|
1367 - "movaps %%xmm0, (%0, %%esi) \n\t"
|
|
|
1368 - "addl $16, %%esi \n\t"
|
|
|
1369 + "movaps %%xmm0, (%0, %%"REG_S") \n\t"
|
|
|
1370 + "add $16, %%"REG_S" \n\t"
|
|
|
1371 " jnz 1b \n\t"
|
|
|
1372 :: "r" (buf+128)
|
|
|
1373 - : "%esi"
|
|
|
1374 + : "%"REG_S
|
|
|
1375 );
|
|
|
1376
|
|
|
1377
|
|
|
1378 @@ -998,54 +998,54 @@
|
|
|
1379
|
|
|
1380 /* Window and convert to real valued signal */
|
|
|
1381 asm volatile(
|
|
|
1382 - "xorl %%edi, %%edi \n\t" // 0
|
|
|
1383 - "xorl %%esi, %%esi \n\t" // 0
|
|
|
1384 + "xor %%"REG_D", %%"REG_D" \n\t" // 0
|
|
|
1385 + "xor %%"REG_S", %%"REG_S" \n\t" // 0
|
|
|
1386 "movss %3, %%xmm2 \n\t" // bias
|
|
|
1387 "shufps $0x00, %%xmm2, %%xmm2 \n\t" // bias, bias, ...
|
|
|
1388 ".balign 16 \n\t"
|
|
|
1389 "1: \n\t"
|
|
|
1390 - "movlps (%0, %%esi), %%xmm0 \n\t" // ? ? A ?
|
|
|
1391 - "movlps 8(%0, %%esi), %%xmm1 \n\t" // ? ? C ?
|
|
|
1392 - "movhps -16(%0, %%edi), %%xmm1 \n\t" // ? D C ?
|
|
|
1393 - "movhps -8(%0, %%edi), %%xmm0 \n\t" // ? B A ?
|
|
|
1394 + "movlps (%0, %%"REG_S"), %%xmm0 \n\t" // ? ? A ?
|
|
|
1395 + "movlps 8(%0, %%"REG_S"), %%xmm1 \n\t" // ? ? C ?
|
|
|
1396 + "movhps -16(%0, %%"REG_D"), %%xmm1 \n\t" // ? D C ?
|
|
|
1397 + "movhps -8(%0, %%"REG_D"), %%xmm0 \n\t" // ? B A ?
|
|
|
1398 "shufps $0x99, %%xmm1, %%xmm0 \n\t" // D C B A
|
|
|
1399 - "mulps "MANGLE(sseWindow)"(%%esi), %%xmm0\n\t"
|
|
|
1400 - "addps (%2, %%esi), %%xmm0 \n\t"
|
|
|
1401 + "mulps "MANGLE(sseWindow)"(%%"REG_S"), %%xmm0\n\t"
|
|
|
1402 + "addps (%2, %%"REG_S"), %%xmm0 \n\t"
|
|
|
1403 "addps %%xmm2, %%xmm0 \n\t"
|
|
|
1404 - "movaps %%xmm0, (%1, %%esi) \n\t"
|
|
|
1405 - "addl $16, %%esi \n\t"
|
|
|
1406 - "subl $16, %%edi \n\t"
|
|
|
1407 - "cmpl $512, %%esi \n\t"
|
|
|
1408 + "movaps %%xmm0, (%1, %%"REG_S") \n\t"
|
|
|
1409 + "add $16, %%"REG_S" \n\t"
|
|
|
1410 + "sub $16, %%"REG_D" \n\t"
|
|
|
1411 + "cmp $512, %%"REG_S" \n\t"
|
|
|
1412 " jb 1b \n\t"
|
|
|
1413 :: "r" (buf+64), "r" (data_ptr), "r" (delay_ptr), "m" (bias)
|
|
|
1414 - : "%esi", "%edi"
|
|
|
1415 + : "%"REG_S, "%"REG_D
|
|
|
1416 );
|
|
|
1417 data_ptr+=128;
|
|
|
1418 delay_ptr+=128;
|
|
|
1419 // window_ptr+=128;
|
|
|
1420
|
|
|
1421 asm volatile(
|
|
|
1422 - "movl $1024, %%edi \n\t" // 512
|
|
|
1423 - "xorl %%esi, %%esi \n\t" // 0
|
|
|
1424 + "mov $1024, %%"REG_D" \n\t" // 512
|
|
|
1425 + "xor %%"REG_S", %%"REG_S" \n\t" // 0
|
|
|
1426 "movss %3, %%xmm2 \n\t" // bias
|
|
|
1427 "shufps $0x00, %%xmm2, %%xmm2 \n\t" // bias, bias, ...
|
|
|
1428 ".balign 16 \n\t"
|
|
|
1429 "1: \n\t"
|
|
|
1430 - "movlps (%0, %%esi), %%xmm0 \n\t" // ? ? ? A
|
|
|
1431 - "movlps 8(%0, %%esi), %%xmm1 \n\t" // ? ? ? C
|
|
|
1432 - "movhps -16(%0, %%edi), %%xmm1 \n\t" // D ? ? C
|
|
|
1433 - "movhps -8(%0, %%edi), %%xmm0 \n\t" // B ? ? A
|
|
|
1434 + "movlps (%0, %%"REG_S"), %%xmm0 \n\t" // ? ? ? A
|
|
|
1435 + "movlps 8(%0, %%"REG_S"), %%xmm1 \n\t" // ? ? ? C
|
|
|
1436 + "movhps -16(%0, %%"REG_D"), %%xmm1 \n\t" // D ? ? C
|
|
|
1437 + "movhps -8(%0, %%"REG_D"), %%xmm0 \n\t" // B ? ? A
|
|
|
1438 "shufps $0xCC, %%xmm1, %%xmm0 \n\t" // D C B A
|
|
|
1439 - "mulps 512+"MANGLE(sseWindow)"(%%esi), %%xmm0\n\t"
|
|
|
1440 - "addps (%2, %%esi), %%xmm0 \n\t"
|
|
|
1441 + "mulps 512+"MANGLE(sseWindow)"(%%"REG_S"), %%xmm0\n\t"
|
|
|
1442 + "addps (%2, %%"REG_S"), %%xmm0 \n\t"
|
|
|
1443 "addps %%xmm2, %%xmm0 \n\t"
|
|
|
1444 - "movaps %%xmm0, (%1, %%esi) \n\t"
|
|
|
1445 - "addl $16, %%esi \n\t"
|
|
|
1446 - "subl $16, %%edi \n\t"
|
|
|
1447 - "cmpl $512, %%esi \n\t"
|
|
|
1448 + "movaps %%xmm0, (%1, %%"REG_S") \n\t"
|
|
|
1449 + "add $16, %%"REG_S" \n\t"
|
|
|
1450 + "sub $16, %%"REG_D" \n\t"
|
|
|
1451 + "cmp $512, %%"REG_S" \n\t"
|
|
|
1452 " jb 1b \n\t"
|
|
|
1453 :: "r" (buf), "r" (data_ptr), "r" (delay_ptr), "m" (bias)
|
|
|
1454 - : "%esi", "%edi"
|
|
|
1455 + : "%"REG_S, "%"REG_D
|
|
|
1456 );
|
|
|
1457 data_ptr+=128;
|
|
|
1458 // window_ptr+=128;
|
|
|
1459 @@ -1054,48 +1054,48 @@
|
|
|
1460 delay_ptr = delay;
|
|
|
1461
|
|
|
1462 asm volatile(
|
|
|
1463 - "xorl %%edi, %%edi \n\t" // 0
|
|
|
1464 - "xorl %%esi, %%esi \n\t" // 0
|
|
|
1465 + "xor %%"REG_D", %%"REG_D" \n\t" // 0
|
|
|
1466 + "xor %%"REG_S", %%"REG_S" \n\t" // 0
|
|
|
1467 ".balign 16 \n\t"
|
|
|
1468 "1: \n\t"
|
|
|
1469 - "movlps (%0, %%esi), %%xmm0 \n\t" // ? ? ? A
|
|
|
1470 - "movlps 8(%0, %%esi), %%xmm1 \n\t" // ? ? ? C
|
|
|
1471 - "movhps -16(%0, %%edi), %%xmm1 \n\t" // D ? ? C
|
|
|
1472 - "movhps -8(%0, %%edi), %%xmm0 \n\t" // B ? ? A
|
|
|
1473 + "movlps (%0, %%"REG_S"), %%xmm0 \n\t" // ? ? ? A
|
|
|
1474 + "movlps 8(%0, %%"REG_S"), %%xmm1 \n\t" // ? ? ? C
|
|
|
1475 + "movhps -16(%0, %%"REG_D"), %%xmm1 \n\t" // D ? ? C
|
|
|
1476 + "movhps -8(%0, %%"REG_D"), %%xmm0 \n\t" // B ? ? A
|
|
|
1477 "shufps $0xCC, %%xmm1, %%xmm0 \n\t" // D C B A
|
|
|
1478 - "mulps 1024+"MANGLE(sseWindow)"(%%esi), %%xmm0\n\t"
|
|
|
1479 - "movaps %%xmm0, (%1, %%esi) \n\t"
|
|
|
1480 - "addl $16, %%esi \n\t"
|
|
|
1481 - "subl $16, %%edi \n\t"
|
|
|
1482 - "cmpl $512, %%esi \n\t"
|
|
|
1483 + "mulps 1024+"MANGLE(sseWindow)"(%%"REG_S"), %%xmm0\n\t"
|
|
|
1484 + "movaps %%xmm0, (%1, %%"REG_S") \n\t"
|
|
|
1485 + "add $16, %%"REG_S" \n\t"
|
|
|
1486 + "sub $16, %%"REG_D" \n\t"
|
|
|
1487 + "cmp $512, %%"REG_S" \n\t"
|
|
|
1488 " jb 1b \n\t"
|
|
|
1489 :: "r" (buf+64), "r" (delay_ptr)
|
|
|
1490 - : "%esi", "%edi"
|
|
|
1491 + : "%"REG_S, "%"REG_D
|
|
|
1492 );
|
|
|
1493 delay_ptr+=128;
|
|
|
1494 // window_ptr-=128;
|
|
|
1495
|
|
|
1496 asm volatile(
|
|
|
1497 - "movl $1024, %%edi \n\t" // 1024
|
|
|
1498 - "xorl %%esi, %%esi \n\t" // 0
|
|
|
1499 + "mov $1024, %%"REG_D" \n\t" // 1024
|
|
|
1500 + "xor %%"REG_S", %%"REG_S" \n\t" // 0
|
|
|
1501 ".balign 16 \n\t"
|
|
|
1502 "1: \n\t"
|
|
|
1503 - "movlps (%0, %%esi), %%xmm0 \n\t" // ? ? A ?
|
|
|
1504 - "movlps 8(%0, %%esi), %%xmm1 \n\t" // ? ? C ?
|
|
|
1505 - "movhps -16(%0, %%edi), %%xmm1 \n\t" // ? D C ?
|
|
|
1506 - "movhps -8(%0, %%edi), %%xmm0 \n\t" // ? B A ?
|
|
|
1507 + "movlps (%0, %%"REG_S"), %%xmm0 \n\t" // ? ? A ?
|
|
|
1508 + "movlps 8(%0, %%"REG_S"), %%xmm1 \n\t" // ? ? C ?
|
|
|
1509 + "movhps -16(%0, %%"REG_D"), %%xmm1 \n\t" // ? D C ?
|
|
|
1510 + "movhps -8(%0, %%"REG_D"), %%xmm0 \n\t" // ? B A ?
|
|
|
1511 "shufps $0x99, %%xmm1, %%xmm0 \n\t" // D C B A
|
|
|
1512 - "mulps 1536+"MANGLE(sseWindow)"(%%esi), %%xmm0\n\t"
|
|
|
1513 - "movaps %%xmm0, (%1, %%esi) \n\t"
|
|
|
1514 - "addl $16, %%esi \n\t"
|
|
|
1515 - "subl $16, %%edi \n\t"
|
|
|
1516 - "cmpl $512, %%esi \n\t"
|
|
|
1517 + "mulps 1536+"MANGLE(sseWindow)"(%%"REG_S"), %%xmm0\n\t"
|
|
|
1518 + "movaps %%xmm0, (%1, %%"REG_S") \n\t"
|
|
|
1519 + "add $16, %%"REG_S" \n\t"
|
|
|
1520 + "sub $16, %%"REG_D" \n\t"
|
|
|
1521 + "cmp $512, %%"REG_S" \n\t"
|
|
|
1522 " jb 1b \n\t"
|
|
|
1523 :: "r" (buf), "r" (delay_ptr)
|
|
|
1524 - : "%esi", "%edi"
|
|
|
1525 + : "%"REG_S, "%"REG_D
|
|
|
1526 );
|
|
|
1527 }
|
|
|
1528 -#endif //arch_x86
|
|
|
1529 +#endif // ARCH_X86 || ARCH_X86_64
|
|
|
1530
|
|
|
1531 void
|
|
|
1532 imdct_do_256(sample_t data[],sample_t delay[],sample_t bias)
|
|
|
1533 @@ -1242,7 +1242,7 @@
|
|
|
1534 xcos1[i] = -cos ((M_PI / 2048) * (8 * i + 1));
|
|
|
1535 xsin1[i] = -sin ((M_PI / 2048) * (8 * i + 1));
|
|
|
1536 }
|
|
|
1537 -#ifdef ARCH_X86
|
|
|
1538 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
1539 for (i = 0; i < 128; i++) {
|
|
|
1540 sseSinCos1c[2*i+0]= xcos1[i];
|
|
|
1541 sseSinCos1c[2*i+1]= -xcos1[i];
|
|
|
1542 @@ -1264,7 +1264,7 @@
|
|
|
1543 w[i][k].imag = sin (-M_PI * k / j);
|
|
|
1544 }
|
|
|
1545 }
|
|
|
1546 -#ifdef ARCH_X86
|
|
|
1547 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
1548 for (i = 1; i < 7; i++) {
|
|
|
1549 j = 1 << i;
|
|
|
1550 for (k = 0; k < j; k+=2) {
|
|
|
1551 @@ -1307,10 +1307,10 @@
|
|
|
1552 sseWindow[384 + 2*i+0]= imdct_window[126 - 2*i+1];
|
|
|
1553 sseWindow[384 + 2*i+1]= -imdct_window[126 - 2*i+0];
|
|
|
1554 }
|
|
|
1555 -#endif // arch_x86
|
|
|
1556 +#endif // ARCH_X86 || ARCH_X86_64
|
|
|
1557
|
|
|
1558 imdct_512 = imdct_do_512;
|
|
|
1559 -#ifdef ARCH_X86
|
|
|
1560 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
1561 if(mm_accel & MM_ACCEL_X86_SSE)
|
|
|
1562 {
|
|
|
1563 fprintf (stderr, "Using SSE optimized IMDCT transform\n");
|
|
|
1564 @@ -1329,7 +1329,7 @@
|
|
|
1565 imdct_512 = imdct_do_512_3dnow;
|
|
|
1566 }
|
|
|
1567 else
|
|
|
1568 -#endif // arch_x86
|
|
|
1569 +#endif // ARCH_X86 || ARCH_X86_64
|
|
|
1570 #ifdef HAVE_ALTIVEC
|
|
|
1571 if (mm_accel & MM_ACCEL_PPC_ALTIVEC)
|
|
|
1572 {
|
|
|
1573 Index: liba52/resample.c
|
|
|
1574 ===================================================================
|
|
|
1575 RCS file: /cvsroot/mplayer/main/liba52/resample.c,v
|
|
|
1576 retrieving revision 1.16
|
|
|
1577 diff -u -r1.16 resample.c
|
|
|
1578 --- liba52/resample.c 25 Jan 2004 18:29:11 -0000 1.16
|
|
|
1579 +++ liba52/resample.c 31 Jul 2005 21:20:10 -0000
|
|
|
1580 @@ -15,7 +15,7 @@
|
|
|
1581
|
|
|
1582 #include "resample_c.c"
|
|
|
1583
|
|
|
1584 -#ifdef ARCH_X86
|
|
|
1585 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
1586 #include "resample_mmx.c"
|
|
|
1587 #endif
|
|
|
1588
|
|
|
1589 @@ -26,7 +26,7 @@
|
|
|
1590 void* a52_resample_init(uint32_t mm_accel,int flags,int chans){
|
|
|
1591 void* tmp;
|
|
|
1592
|
|
|
1593 -#ifdef ARCH_X86
|
|
|
1594 +#if defined(ARCH_X86) || defined(ARCH_X86_64)
|
|
|
1595 if(mm_accel&MM_ACCEL_X86_MMX){
|
|
|
1596 tmp=a52_resample_MMX(flags,chans);
|
|
|
1597 if(tmp){
|
|
|
1598 Index: liba52/resample_mmx.c
|
|
|
1599 ===================================================================
|
|
|
1600 RCS file: /cvsroot/mplayer/main/liba52/resample_mmx.c,v
|
|
|
1601 retrieving revision 1.17
|
|
|
1602 diff -u -r1.17 resample_mmx.c
|
|
|
1603 --- liba52/resample_mmx.c 26 Apr 2004 19:47:50 -0000 1.17
|
|
|
1604 +++ liba52/resample_mmx.c 31 Jul 2005 21:20:10 -0000
|
|
|
1605 @@ -7,6 +7,9 @@
|
|
|
1606 and it would mean (C / MMX2 / MMX / 3DNOW) versions
|
|
|
1607 */
|
|
|
1608
|
|
|
1609 +#include "a52_internal.h"
|
|
|
1610 +
|
|
|
1611 +
|
|
|
1612 static uint64_t attribute_used __attribute__((aligned(8))) magicF2W= 0x43c0000043c00000LL;
|
|
|
1613 static uint64_t attribute_used __attribute__((aligned(8))) wm1010= 0xFFFF0000FFFF0000LL;
|
|
|
1614 static uint64_t attribute_used __attribute__((aligned(8))) wm0101= 0x0000FFFF0000FFFFLL;
|
|
|
1615 @@ -15,36 +18,36 @@
|
|
|
1616 static int a52_resample_MONO_to_5_MMX(float * _f, int16_t * s16){
|
|
|
1617 int32_t * f = (int32_t *) _f;
|
|
|
1618 asm volatile(
|
|
|
1619 - "movl $-512, %%esi \n\t"
|
|
|
1620 + "mov $-512, %%"REG_S" \n\t"
|
|
|
1621 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
1622 "movq "MANGLE(wm1100)", %%mm3 \n\t"
|
|
|
1623 "movq "MANGLE(wm0101)", %%mm4 \n\t"
|
|
|
1624 "movq "MANGLE(wm1010)", %%mm5 \n\t"
|
|
|
1625 "pxor %%mm6, %%mm6 \n\t"
|
|
|
1626 "1: \n\t"
|
|
|
1627 - "movq (%1, %%esi, 2), %%mm0 \n\t"
|
|
|
1628 - "movq 8(%1, %%esi, 2), %%mm1 \n\t"
|
|
|
1629 - "leal (%%esi, %%esi, 4), %%edi \n\t"
|
|
|
1630 + "movq (%1, %%"REG_S", 2), %%mm0 \n\t"
|
|
|
1631 + "movq 8(%1, %%"REG_S", 2), %%mm1\n\t"
|
|
|
1632 + "lea (%%"REG_S", %%"REG_S", 4), %%"REG_D"\n\t"
|
|
|
1633 "psubd %%mm7, %%mm0 \n\t"
|
|
|
1634 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1635 "packssdw %%mm1, %%mm0 \n\t"
|
|
|
1636 "movq %%mm0, %%mm1 \n\t"
|
|
|
1637 "pand %%mm4, %%mm0 \n\t"
|
|
|
1638 "pand %%mm5, %%mm1 \n\t"
|
|
|
1639 - "movq %%mm6, (%0, %%edi) \n\t" // 0 0 0 0
|
|
|
1640 - "movd %%mm0, 8(%0, %%edi) \n\t" // A 0
|
|
|
1641 + "movq %%mm6, (%0, %%"REG_D") \n\t" // 0 0 0 0
|
|
|
1642 + "movd %%mm0, 8(%0, %%"REG_D") \n\t" // A 0
|
|
|
1643 "pand %%mm3, %%mm0 \n\t"
|
|
|
1644 - "movd %%mm6, 12(%0, %%edi) \n\t" // 0 0
|
|
|
1645 - "movd %%mm1, 16(%0, %%edi) \n\t" // 0 B
|
|
|
1646 + "movd %%mm6, 12(%0, %%"REG_D") \n\t" // 0 0
|
|
|
1647 + "movd %%mm1, 16(%0, %%"REG_D") \n\t" // 0 B
|
|
|
1648 "pand %%mm3, %%mm1 \n\t"
|
|
|
1649 - "movd %%mm6, 20(%0, %%edi) \n\t" // 0 0
|
|
|
1650 - "movq %%mm0, 24(%0, %%edi) \n\t" // 0 0 C 0
|
|
|
1651 - "movq %%mm1, 32(%0, %%edi) \n\t" // 0 0 0 B
|
|
|
1652 - "addl $8, %%esi \n\t"
|
|
|
1653 + "movd %%mm6, 20(%0, %%"REG_D") \n\t" // 0 0
|
|
|
1654 + "movq %%mm0, 24(%0, %%"REG_D") \n\t" // 0 0 C 0
|
|
|
1655 + "movq %%mm1, 32(%0, %%"REG_D") \n\t" // 0 0 0 B
|
|
|
1656 + "add $8, %%"REG_S" \n\t"
|
|
|
1657 " jnz 1b \n\t"
|
|
|
1658 "emms \n\t"
|
|
|
1659 :: "r" (s16+1280), "r" (f+256)
|
|
|
1660 - :"%esi", "%edi", "memory"
|
|
|
1661 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
1662 );
|
|
|
1663 return 5*256;
|
|
|
1664 }
|
|
|
1665 @@ -54,29 +57,29 @@
|
|
|
1666 /* benchmark scores are 0.3% better with SSE but we would need to set bias=0 and premultiply it
|
|
|
1667 #ifdef HAVE_SSE
|
|
|
1668 asm volatile(
|
|
|
1669 - "movl $-1024, %%esi \n\t"
|
|
|
1670 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1671 "1: \n\t"
|
|
|
1672 - "cvtps2pi (%1, %%esi), %%mm0 \n\t"
|
|
|
1673 - "cvtps2pi 1024(%1, %%esi), %%mm2\n\t"
|
|
|
1674 + "cvtps2pi (%1, %%"REG_S"), %%mm0\n\t"
|
|
|
1675 + "cvtps2pi 1024(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1676 "movq %%mm0, %%mm1 \n\t"
|
|
|
1677 "punpcklwd %%mm2, %%mm0 \n\t"
|
|
|
1678 "punpckhwd %%mm2, %%mm1 \n\t"
|
|
|
1679 - "movq %%mm0, (%0, %%esi) \n\t"
|
|
|
1680 - "movq %%mm1, 8(%0, %%esi) \n\t"
|
|
|
1681 - "addl $16, %%esi \n\t"
|
|
|
1682 + "movq %%mm0, (%0, %%"REG_S") \n\t"
|
|
|
1683 + "movq %%mm1, 8(%0, %%"REG_S") \n\t"
|
|
|
1684 + "add $16, %%"REG_S" \n\t"
|
|
|
1685 " jnz 1b \n\t"
|
|
|
1686 "emms \n\t"
|
|
|
1687 :: "r" (s16+512), "r" (f+256)
|
|
|
1688 - :"%esi", "memory"
|
|
|
1689 + :"%"REG_S, "memory"
|
|
|
1690 );*/
|
|
|
1691 asm volatile(
|
|
|
1692 - "movl $-1024, %%esi \n\t"
|
|
|
1693 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1694 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
1695 "1: \n\t"
|
|
|
1696 - "movq (%1, %%esi), %%mm0 \n\t"
|
|
|
1697 - "movq 8(%1, %%esi), %%mm1 \n\t"
|
|
|
1698 - "movq 1024(%1, %%esi), %%mm2 \n\t"
|
|
|
1699 - "movq 1032(%1, %%esi), %%mm3 \n\t"
|
|
|
1700 + "movq (%1, %%"REG_S"), %%mm0 \n\t"
|
|
|
1701 + "movq 8(%1, %%"REG_S"), %%mm1 \n\t"
|
|
|
1702 + "movq 1024(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1703 + "movq 1032(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1704 "psubd %%mm7, %%mm0 \n\t"
|
|
|
1705 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1706 "psubd %%mm7, %%mm2 \n\t"
|
|
|
1707 @@ -86,13 +89,13 @@
|
|
|
1708 "movq %%mm0, %%mm1 \n\t"
|
|
|
1709 "punpcklwd %%mm2, %%mm0 \n\t"
|
|
|
1710 "punpckhwd %%mm2, %%mm1 \n\t"
|
|
|
1711 - "movq %%mm0, (%0, %%esi) \n\t"
|
|
|
1712 - "movq %%mm1, 8(%0, %%esi) \n\t"
|
|
|
1713 - "addl $16, %%esi \n\t"
|
|
|
1714 + "movq %%mm0, (%0, %%"REG_S") \n\t"
|
|
|
1715 + "movq %%mm1, 8(%0, %%"REG_S") \n\t"
|
|
|
1716 + "add $16, %%"REG_S" \n\t"
|
|
|
1717 " jnz 1b \n\t"
|
|
|
1718 "emms \n\t"
|
|
|
1719 :: "r" (s16+512), "r" (f+256)
|
|
|
1720 - :"%esi", "memory"
|
|
|
1721 + :"%"REG_S, "memory"
|
|
|
1722 );
|
|
|
1723 return 2*256;
|
|
|
1724 }
|
|
|
1725 @@ -100,23 +103,23 @@
|
|
|
1726 static int a52_resample_3F_to_5_MMX(float * _f, int16_t * s16){
|
|
|
1727 int32_t * f = (int32_t *) _f;
|
|
|
1728 asm volatile(
|
|
|
1729 - "movl $-1024, %%esi \n\t"
|
|
|
1730 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1731 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
1732 "pxor %%mm6, %%mm6 \n\t"
|
|
|
1733 "movq %%mm7, %%mm5 \n\t"
|
|
|
1734 "punpckldq %%mm6, %%mm5 \n\t"
|
|
|
1735 "1: \n\t"
|
|
|
1736 - "movd (%1, %%esi), %%mm0 \n\t"
|
|
|
1737 - "punpckldq 2048(%1, %%esi), %%mm0\n\t"
|
|
|
1738 - "movd 1024(%1, %%esi), %%mm1 \n\t"
|
|
|
1739 - "punpckldq 4(%1, %%esi), %%mm1 \n\t"
|
|
|
1740 - "movd 2052(%1, %%esi), %%mm2 \n\t"
|
|
|
1741 + "movd (%1, %%"REG_S"), %%mm0 \n\t"
|
|
|
1742 + "punpckldq 2048(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
1743 + "movd 1024(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1744 + "punpckldq 4(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1745 + "movd 2052(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1746 "movq %%mm7, %%mm3 \n\t"
|
|
|
1747 - "punpckldq 1028(%1, %%esi), %%mm3\n\t"
|
|
|
1748 - "movd 8(%1, %%esi), %%mm4 \n\t"
|
|
|
1749 - "punpckldq 2056(%1, %%esi), %%mm4\n\t"
|
|
|
1750 - "leal (%%esi, %%esi, 4), %%edi \n\t"
|
|
|
1751 - "sarl $1, %%edi \n\t"
|
|
|
1752 + "punpckldq 1028(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1753 + "movd 8(%1, %%"REG_S"), %%mm4 \n\t"
|
|
|
1754 + "punpckldq 2056(%1, %%"REG_S"), %%mm4\n\t"
|
|
|
1755 + "lea (%%"REG_S", %%"REG_S", 4), %%"REG_D"\n\t"
|
|
|
1756 + "sar $1, %%"REG_D" \n\t"
|
|
|
1757 "psubd %%mm7, %%mm0 \n\t"
|
|
|
1758 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1759 "psubd %%mm5, %%mm2 \n\t"
|
|
|
1760 @@ -125,29 +128,28 @@
|
|
|
1761 "packssdw %%mm6, %%mm0 \n\t"
|
|
|
1762 "packssdw %%mm2, %%mm1 \n\t"
|
|
|
1763 "packssdw %%mm4, %%mm3 \n\t"
|
|
|
1764 - "movq %%mm0, (%0, %%edi) \n\t"
|
|
|
1765 - "movq %%mm1, 8(%0, %%edi) \n\t"
|
|
|
1766 - "movq %%mm3, 16(%0, %%edi) \n\t"
|
|
|
1767 -
|
|
|
1768 - "movd 1032(%1, %%esi), %%mm1 \n\t"
|
|
|
1769 - "punpckldq 12(%1, %%esi), %%mm1\n\t"
|
|
|
1770 - "movd 2060(%1, %%esi), %%mm2 \n\t"
|
|
|
1771 + "movq %%mm0, (%0, %%"REG_D") \n\t"
|
|
|
1772 + "movq %%mm1, 8(%0, %%"REG_D") \n\t"
|
|
|
1773 + "movq %%mm3, 16(%0, %%"REG_D") \n\t"
|
|
|
1774 + "movd 1032(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1775 + "punpckldq 12(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1776 + "movd 2060(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1777 "movq %%mm7, %%mm3 \n\t"
|
|
|
1778 - "punpckldq 1036(%1, %%esi), %%mm3\n\t"
|
|
|
1779 + "punpckldq 1036(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1780 "pxor %%mm0, %%mm0 \n\t"
|
|
|
1781 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1782 "psubd %%mm5, %%mm2 \n\t"
|
|
|
1783 "psubd %%mm7, %%mm3 \n\t"
|
|
|
1784 "packssdw %%mm1, %%mm0 \n\t"
|
|
|
1785 "packssdw %%mm3, %%mm2 \n\t"
|
|
|
1786 - "movq %%mm0, 24(%0, %%edi) \n\t"
|
|
|
1787 - "movq %%mm2, 32(%0, %%edi) \n\t"
|
|
|
1788 + "movq %%mm0, 24(%0, %%"REG_D") \n\t"
|
|
|
1789 + "movq %%mm2, 32(%0, %%"REG_D") \n\t"
|
|
|
1790
|
|
|
1791 - "addl $16, %%esi \n\t"
|
|
|
1792 + "add $16, %%"REG_S" \n\t"
|
|
|
1793 " jnz 1b \n\t"
|
|
|
1794 "emms \n\t"
|
|
|
1795 :: "r" (s16+1280), "r" (f+256)
|
|
|
1796 - :"%esi", "%edi", "memory"
|
|
|
1797 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
1798 );
|
|
|
1799 return 5*256;
|
|
|
1800 }
|
|
|
1801 @@ -155,23 +157,23 @@
|
|
|
1802 static int a52_resample_2F_2R_to_4_MMX(float * _f, int16_t * s16){
|
|
|
1803 int32_t * f = (int32_t *) _f;
|
|
|
1804 asm volatile(
|
|
|
1805 - "movl $-1024, %%esi \n\t"
|
|
|
1806 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1807 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
1808 "1: \n\t"
|
|
|
1809 - "movq (%1, %%esi), %%mm0 \n\t"
|
|
|
1810 - "movq 8(%1, %%esi), %%mm1 \n\t"
|
|
|
1811 - "movq 1024(%1, %%esi), %%mm2 \n\t"
|
|
|
1812 - "movq 1032(%1, %%esi), %%mm3 \n\t"
|
|
|
1813 + "movq (%1, %%"REG_S"), %%mm0 \n\t"
|
|
|
1814 + "movq 8(%1, %%"REG_S"), %%mm1 \n\t"
|
|
|
1815 + "movq 1024(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1816 + "movq 1032(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1817 "psubd %%mm7, %%mm0 \n\t"
|
|
|
1818 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1819 "psubd %%mm7, %%mm2 \n\t"
|
|
|
1820 "psubd %%mm7, %%mm3 \n\t"
|
|
|
1821 "packssdw %%mm1, %%mm0 \n\t"
|
|
|
1822 "packssdw %%mm3, %%mm2 \n\t"
|
|
|
1823 - "movq 2048(%1, %%esi), %%mm3 \n\t"
|
|
|
1824 - "movq 2056(%1, %%esi), %%mm4 \n\t"
|
|
|
1825 - "movq 3072(%1, %%esi), %%mm5 \n\t"
|
|
|
1826 - "movq 3080(%1, %%esi), %%mm6 \n\t"
|
|
|
1827 + "movq 2048(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1828 + "movq 2056(%1, %%"REG_S"), %%mm4\n\t"
|
|
|
1829 + "movq 3072(%1, %%"REG_S"), %%mm5\n\t"
|
|
|
1830 + "movq 3080(%1, %%"REG_S"), %%mm6\n\t"
|
|
|
1831 "psubd %%mm7, %%mm3 \n\t"
|
|
|
1832 "psubd %%mm7, %%mm4 \n\t"
|
|
|
1833 "psubd %%mm7, %%mm5 \n\t"
|
|
|
1834 @@ -190,15 +192,15 @@
|
|
|
1835 "punpckhdq %%mm3, %%mm2 \n\t"
|
|
|
1836 "punpckldq %%mm4, %%mm1 \n\t"
|
|
|
1837 "punpckhdq %%mm4, %%mm5 \n\t"
|
|
|
1838 - "movq %%mm0, (%0, %%esi,2) \n\t"
|
|
|
1839 - "movq %%mm2, 8(%0, %%esi,2) \n\t"
|
|
|
1840 - "movq %%mm1, 16(%0, %%esi,2) \n\t"
|
|
|
1841 - "movq %%mm5, 24(%0, %%esi,2) \n\t"
|
|
|
1842 - "addl $16, %%esi \n\t"
|
|
|
1843 + "movq %%mm0, (%0, %%"REG_S",2) \n\t"
|
|
|
1844 + "movq %%mm2, 8(%0, %%"REG_S",2) \n\t"
|
|
|
1845 + "movq %%mm1, 16(%0, %%"REG_S",2)\n\t"
|
|
|
1846 + "movq %%mm5, 24(%0, %%"REG_S",2)\n\t"
|
|
|
1847 + "add $16, %%"REG_S" \n\t"
|
|
|
1848 " jnz 1b \n\t"
|
|
|
1849 "emms \n\t"
|
|
|
1850 :: "r" (s16+1024), "r" (f+256)
|
|
|
1851 - :"%esi", "memory"
|
|
|
1852 + :"%"REG_S, "memory"
|
|
|
1853 );
|
|
|
1854 return 4*256;
|
|
|
1855 }
|
|
|
1856 @@ -206,23 +208,23 @@
|
|
|
1857 static int a52_resample_3F_2R_to_5_MMX(float * _f, int16_t * s16){
|
|
|
1858 int32_t * f = (int32_t *) _f;
|
|
|
1859 asm volatile(
|
|
|
1860 - "movl $-1024, %%esi \n\t"
|
|
|
1861 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1862 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
1863 "1: \n\t"
|
|
|
1864 - "movd (%1, %%esi), %%mm0 \n\t"
|
|
|
1865 - "punpckldq 2048(%1, %%esi), %%mm0\n\t"
|
|
|
1866 - "movd 3072(%1, %%esi), %%mm1 \n\t"
|
|
|
1867 - "punpckldq 4096(%1, %%esi), %%mm1\n\t"
|
|
|
1868 - "movd 1024(%1, %%esi), %%mm2 \n\t"
|
|
|
1869 - "punpckldq 4(%1, %%esi), %%mm2 \n\t"
|
|
|
1870 - "movd 2052(%1, %%esi), %%mm3 \n\t"
|
|
|
1871 - "punpckldq 3076(%1, %%esi), %%mm3\n\t"
|
|
|
1872 - "movd 4100(%1, %%esi), %%mm4 \n\t"
|
|
|
1873 - "punpckldq 1028(%1, %%esi), %%mm4\n\t"
|
|
|
1874 - "movd 8(%1, %%esi), %%mm5 \n\t"
|
|
|
1875 - "punpckldq 2056(%1, %%esi), %%mm5\n\t"
|
|
|
1876 - "leal (%%esi, %%esi, 4), %%edi \n\t"
|
|
|
1877 - "sarl $1, %%edi \n\t"
|
|
|
1878 + "movd (%1, %%"REG_S"), %%mm0 \n\t"
|
|
|
1879 + "punpckldq 2048(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
1880 + "movd 3072(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1881 + "punpckldq 4096(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1882 + "movd 1024(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1883 + "punpckldq 4(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1884 + "movd 2052(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1885 + "punpckldq 3076(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1886 + "movd 4100(%1, %%"REG_S"), %%mm4\n\t"
|
|
|
1887 + "punpckldq 1028(%1, %%"REG_S"), %%mm4\n\t"
|
|
|
1888 + "movd 8(%1, %%"REG_S"), %%mm5 \n\t"
|
|
|
1889 + "punpckldq 2056(%1, %%"REG_S"), %%mm5\n\t"
|
|
|
1890 + "lea (%%"REG_S", %%"REG_S", 4), %%"REG_D"\n\t"
|
|
|
1891 + "sar $1, %%"REG_D" \n\t"
|
|
|
1892 "psubd %%mm7, %%mm0 \n\t"
|
|
|
1893 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1894 "psubd %%mm7, %%mm2 \n\t"
|
|
|
1895 @@ -232,32 +234,32 @@
|
|
|
1896 "packssdw %%mm1, %%mm0 \n\t"
|
|
|
1897 "packssdw %%mm3, %%mm2 \n\t"
|
|
|
1898 "packssdw %%mm5, %%mm4 \n\t"
|
|
|
1899 - "movq %%mm0, (%0, %%edi) \n\t"
|
|
|
1900 - "movq %%mm2, 8(%0, %%edi) \n\t"
|
|
|
1901 - "movq %%mm4, 16(%0, %%edi) \n\t"
|
|
|
1902 + "movq %%mm0, (%0, %%"REG_D") \n\t"
|
|
|
1903 + "movq %%mm2, 8(%0, %%"REG_D") \n\t"
|
|
|
1904 + "movq %%mm4, 16(%0, %%"REG_D") \n\t"
|
|
|
1905
|
|
|
1906 - "movd 3080(%1, %%esi), %%mm0 \n\t"
|
|
|
1907 - "punpckldq 4104(%1, %%esi), %%mm0\n\t"
|
|
|
1908 - "movd 1032(%1, %%esi), %%mm1 \n\t"
|
|
|
1909 - "punpckldq 12(%1, %%esi), %%mm1\n\t"
|
|
|
1910 - "movd 2060(%1, %%esi), %%mm2 \n\t"
|
|
|
1911 - "punpckldq 3084(%1, %%esi), %%mm2\n\t"
|
|
|
1912 - "movd 4108(%1, %%esi), %%mm3 \n\t"
|
|
|
1913 - "punpckldq 1036(%1, %%esi), %%mm3\n\t"
|
|
|
1914 + "movd 3080(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
1915 + "punpckldq 4104(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
1916 + "movd 1032(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1917 + "punpckldq 12(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1918 + "movd 2060(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1919 + "punpckldq 3084(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
1920 + "movd 4108(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1921 + "punpckldq 1036(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
1922 "psubd %%mm7, %%mm0 \n\t"
|
|
|
1923 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1924 "psubd %%mm7, %%mm2 \n\t"
|
|
|
1925 "psubd %%mm7, %%mm3 \n\t"
|
|
|
1926 "packssdw %%mm1, %%mm0 \n\t"
|
|
|
1927 "packssdw %%mm3, %%mm2 \n\t"
|
|
|
1928 - "movq %%mm0, 24(%0, %%edi) \n\t"
|
|
|
1929 - "movq %%mm2, 32(%0, %%edi) \n\t"
|
|
|
1930 + "movq %%mm0, 24(%0, %%"REG_D") \n\t"
|
|
|
1931 + "movq %%mm2, 32(%0, %%"REG_D") \n\t"
|
|
|
1932
|
|
|
1933 - "addl $16, %%esi \n\t"
|
|
|
1934 + "add $16, %%"REG_S" \n\t"
|
|
|
1935 " jnz 1b \n\t"
|
|
|
1936 "emms \n\t"
|
|
|
1937 :: "r" (s16+1280), "r" (f+256)
|
|
|
1938 - :"%esi", "%edi", "memory"
|
|
|
1939 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
1940 );
|
|
|
1941 return 5*256;
|
|
|
1942 }
|
|
|
1943 @@ -265,14 +267,14 @@
|
|
|
1944 static int a52_resample_MONO_LFE_to_6_MMX(float * _f, int16_t * s16){
|
|
|
1945 int32_t * f = (int32_t *) _f;
|
|
|
1946 asm volatile(
|
|
|
1947 - "movl $-1024, %%esi \n\t"
|
|
|
1948 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
1949 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
1950 "pxor %%mm6, %%mm6 \n\t"
|
|
|
1951 "1: \n\t"
|
|
|
1952 - "movq 1024(%1, %%esi), %%mm0 \n\t"
|
|
|
1953 - "movq 1032(%1, %%esi), %%mm1 \n\t"
|
|
|
1954 - "movq (%1, %%esi), %%mm2 \n\t"
|
|
|
1955 - "movq 8(%1, %%esi), %%mm3 \n\t"
|
|
|
1956 + "movq 1024(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
1957 + "movq 1032(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
1958 + "movq (%1, %%"REG_S"), %%mm2 \n\t"
|
|
|
1959 + "movq 8(%1, %%"REG_S"), %%mm3 \n\t"
|
|
|
1960 "psubd %%mm7, %%mm0 \n\t"
|
|
|
1961 "psubd %%mm7, %%mm1 \n\t"
|
|
|
1962 "psubd %%mm7, %%mm2 \n\t"
|
|
|
1963 @@ -282,22 +284,22 @@
|
|
|
1964 "movq %%mm0, %%mm1 \n\t"
|
|
|
1965 "punpcklwd %%mm2, %%mm0 \n\t"
|
|
|
1966 "punpckhwd %%mm2, %%mm1 \n\t"
|
|
|
1967 - "leal (%%esi, %%esi, 2), %%edi \n\t"
|
|
|
1968 - "movq %%mm6, (%0, %%edi) \n\t"
|
|
|
1969 - "movd %%mm0, 8(%0, %%edi) \n\t"
|
|
|
1970 + "lea (%%"REG_S", %%"REG_S", 2), %%"REG_D"\n\t"
|
|
|
1971 + "movq %%mm6, (%0, %%"REG_D") \n\t"
|
|
|
1972 + "movd %%mm0, 8(%0, %%"REG_D") \n\t"
|
|
|
1973 "punpckhdq %%mm0, %%mm0 \n\t"
|
|
|
1974 - "movq %%mm6, 12(%0, %%edi) \n\t"
|
|
|
1975 - "movd %%mm0, 20(%0, %%edi) \n\t"
|
|
|
1976 - "movq %%mm6, 24(%0, %%edi) \n\t"
|
|
|
1977 - "movd %%mm1, 32(%0, %%edi) \n\t"
|
|
|
1978 + "movq %%mm6, 12(%0, %%"REG_D") \n\t"
|
|
|
1979 + "movd %%mm0, 20(%0, %%"REG_D") \n\t"
|
|
|
1980 + "movq %%mm6, 24(%0, %%"REG_D") \n\t"
|
|
|
1981 + "movd %%mm1, 32(%0, %%"REG_D") \n\t"
|
|
|
1982 "punpckhdq %%mm1, %%mm1 \n\t"
|
|
|
1983 - "movq %%mm6, 36(%0, %%edi) \n\t"
|
|
|
1984 - "movd %%mm1, 44(%0, %%edi) \n\t"
|
|
|
1985 - "addl $16, %%esi \n\t"
|
|
|
1986 + "movq %%mm6, 36(%0, %%"REG_D") \n\t"
|
|
|
1987 + "movd %%mm1, 44(%0, %%"REG_D") \n\t"
|
|
|
1988 + "add $16, %%"REG_S" \n\t"
|
|
|
1989 " jnz 1b \n\t"
|
|
|
1990 "emms \n\t"
|
|
|
1991 :: "r" (s16+1536), "r" (f+256)
|
|
|
1992 - :"%esi", "%edi", "memory"
|
|
|
1993 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
1994 );
|
|
|
1995 return 6*256;
|
|
|
1996 }
|
|
|
1997 @@ -305,17 +307,17 @@
|
|
|
1998 static int a52_resample_STEREO_LFE_to_6_MMX(float * _f, int16_t * s16){
|
|
|
1999 int32_t * f = (int32_t *) _f;
|
|
|
2000 asm volatile(
|
|
|
2001 - "movl $-1024, %%esi \n\t"
|
|
|
2002 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
2003 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
2004 "pxor %%mm6, %%mm6 \n\t"
|
|
|
2005 "1: \n\t"
|
|
|
2006 - "movq 1024(%1, %%esi), %%mm0 \n\t"
|
|
|
2007 - "movq 2048(%1, %%esi), %%mm1 \n\t"
|
|
|
2008 - "movq (%1, %%esi), %%mm5 \n\t"
|
|
|
2009 + "movq 1024(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
2010 + "movq 2048(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
2011 + "movq (%1, %%"REG_S"), %%mm5 \n\t"
|
|
|
2012 "psubd %%mm7, %%mm0 \n\t"
|
|
|
2013 "psubd %%mm7, %%mm1 \n\t"
|
|
|
2014 "psubd %%mm7, %%mm5 \n\t"
|
|
|
2015 - "leal (%%esi, %%esi, 2), %%edi \n\t"
|
|
|
2016 + "lea (%%"REG_S", %%"REG_S", 2), %%"REG_D"\n\t"
|
|
|
2017
|
|
|
2018 "pxor %%mm4, %%mm4 \n\t"
|
|
|
2019 "packssdw %%mm5, %%mm0 \n\t" // FfAa
|
|
|
2020 @@ -327,15 +329,15 @@
|
|
|
2021 "punpckldq %%mm6, %%mm0 \n\t" // 00ba
|
|
|
2022 "punpckhdq %%mm1, %%mm3 \n\t" // BAf0
|
|
|
2023
|
|
|
2024 - "movq %%mm0, (%0, %%edi) \n\t" // 00ba
|
|
|
2025 + "movq %%mm0, (%0, %%"REG_D") \n\t" // 00ba
|
|
|
2026 "punpckhdq %%mm4, %%mm0 \n\t" // F000
|
|
|
2027 - "movq %%mm3, 8(%0, %%edi) \n\t" // BAf0
|
|
|
2028 - "movq %%mm0, 16(%0, %%edi) \n\t" // F000
|
|
|
2029 - "addl $8, %%esi \n\t"
|
|
|
2030 + "movq %%mm3, 8(%0, %%"REG_D") \n\t" // BAf0
|
|
|
2031 + "movq %%mm0, 16(%0, %%"REG_D") \n\t" // F000
|
|
|
2032 + "add $8, %%"REG_S" \n\t"
|
|
|
2033 " jnz 1b \n\t"
|
|
|
2034 "emms \n\t"
|
|
|
2035 :: "r" (s16+1536), "r" (f+256)
|
|
|
2036 - :"%esi", "%edi", "memory"
|
|
|
2037 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
2038 );
|
|
|
2039 return 6*256;
|
|
|
2040 }
|
|
|
2041 @@ -343,19 +345,19 @@
|
|
|
2042 static int a52_resample_3F_LFE_to_6_MMX(float * _f, int16_t * s16){
|
|
|
2043 int32_t * f = (int32_t *) _f;
|
|
|
2044 asm volatile(
|
|
|
2045 - "movl $-1024, %%esi \n\t"
|
|
|
2046 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
2047 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
2048 "pxor %%mm6, %%mm6 \n\t"
|
|
|
2049 "1: \n\t"
|
|
|
2050 - "movq 1024(%1, %%esi), %%mm0 \n\t"
|
|
|
2051 - "movq 3072(%1, %%esi), %%mm1 \n\t"
|
|
|
2052 - "movq 2048(%1, %%esi), %%mm4 \n\t"
|
|
|
2053 - "movq (%1, %%esi), %%mm5 \n\t"
|
|
|
2054 + "movq 1024(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
2055 + "movq 3072(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
2056 + "movq 2048(%1, %%"REG_S"), %%mm4\n\t"
|
|
|
2057 + "movq (%1, %%"REG_S"), %%mm5 \n\t"
|
|
|
2058 "psubd %%mm7, %%mm0 \n\t"
|
|
|
2059 "psubd %%mm7, %%mm1 \n\t"
|
|
|
2060 "psubd %%mm7, %%mm4 \n\t"
|
|
|
2061 "psubd %%mm7, %%mm5 \n\t"
|
|
|
2062 - "leal (%%esi, %%esi, 2), %%edi \n\t"
|
|
|
2063 + "lea (%%"REG_S", %%"REG_S", 2), %%"REG_D"\n\t"
|
|
|
2064
|
|
|
2065 "packssdw %%mm4, %%mm0 \n\t" // EeAa
|
|
|
2066 "packssdw %%mm5, %%mm1 \n\t" // FfBb
|
|
|
2067 @@ -366,16 +368,16 @@
|
|
|
2068 "punpckldq %%mm6, %%mm0 \n\t" // 00ba
|
|
|
2069 "punpckhdq %%mm1, %%mm1 \n\t" // BABA
|
|
|
2070
|
|
|
2071 - "movq %%mm0, (%0, %%edi) \n\t"
|
|
|
2072 + "movq %%mm0, (%0, %%"REG_D") \n\t"
|
|
|
2073 "punpckhdq %%mm2, %%mm0 \n\t" // FE00
|
|
|
2074 "punpckldq %%mm1, %%mm2 \n\t" // BAfe
|
|
|
2075 - "movq %%mm2, 8(%0, %%edi) \n\t"
|
|
|
2076 - "movq %%mm0, 16(%0, %%edi) \n\t"
|
|
|
2077 - "addl $8, %%esi \n\t"
|
|
|
2078 + "movq %%mm2, 8(%0, %%"REG_D") \n\t"
|
|
|
2079 + "movq %%mm0, 16(%0, %%"REG_D") \n\t"
|
|
|
2080 + "add $8, %%"REG_S" \n\t"
|
|
|
2081 " jnz 1b \n\t"
|
|
|
2082 "emms \n\t"
|
|
|
2083 :: "r" (s16+1536), "r" (f+256)
|
|
|
2084 - :"%esi", "%edi", "memory"
|
|
|
2085 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
2086 );
|
|
|
2087 return 6*256;
|
|
|
2088 }
|
|
|
2089 @@ -383,21 +385,21 @@
|
|
|
2090 static int a52_resample_2F_2R_LFE_to_6_MMX(float * _f, int16_t * s16){
|
|
|
2091 int32_t * f = (int32_t *) _f;
|
|
|
2092 asm volatile(
|
|
|
2093 - "movl $-1024, %%esi \n\t"
|
|
|
2094 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
2095 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
2096 // "pxor %%mm6, %%mm6 \n\t"
|
|
|
2097 "1: \n\t"
|
|
|
2098 - "movq 1024(%1, %%esi), %%mm0 \n\t"
|
|
|
2099 - "movq 2048(%1, %%esi), %%mm1 \n\t"
|
|
|
2100 - "movq 3072(%1, %%esi), %%mm2 \n\t"
|
|
|
2101 - "movq 4096(%1, %%esi), %%mm3 \n\t"
|
|
|
2102 - "movq (%1, %%esi), %%mm5 \n\t"
|
|
|
2103 + "movq 1024(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
2104 + "movq 2048(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
2105 + "movq 3072(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
2106 + "movq 4096(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
2107 + "movq (%1, %%"REG_S"), %%mm5 \n\t"
|
|
|
2108 "psubd %%mm7, %%mm0 \n\t"
|
|
|
2109 "psubd %%mm7, %%mm1 \n\t"
|
|
|
2110 "psubd %%mm7, %%mm2 \n\t"
|
|
|
2111 "psubd %%mm7, %%mm3 \n\t"
|
|
|
2112 "psubd %%mm7, %%mm5 \n\t"
|
|
|
2113 - "leal (%%esi, %%esi, 2), %%edi \n\t"
|
|
|
2114 + "lea (%%"REG_S", %%"REG_S", 2), %%"REG_D"\n\t"
|
|
|
2115
|
|
|
2116 "packssdw %%mm2, %%mm0 \n\t" // CcAa
|
|
|
2117 "packssdw %%mm3, %%mm1 \n\t" // DdBb
|
|
|
2118 @@ -414,14 +416,14 @@
|
|
|
2119 "punpckldq %%mm1, %%mm4 \n\t" // BAf0
|
|
|
2120 "punpckhdq %%mm3, %%mm2 \n\t" // F0DC
|
|
|
2121
|
|
|
2122 - "movq %%mm0, (%0, %%edi) \n\t"
|
|
|
2123 - "movq %%mm4, 8(%0, %%edi) \n\t"
|
|
|
2124 - "movq %%mm2, 16(%0, %%edi) \n\t"
|
|
|
2125 - "addl $8, %%esi \n\t"
|
|
|
2126 + "movq %%mm0, (%0, %%"REG_D") \n\t"
|
|
|
2127 + "movq %%mm4, 8(%0, %%"REG_D") \n\t"
|
|
|
2128 + "movq %%mm2, 16(%0, %%"REG_D") \n\t"
|
|
|
2129 + "add $8, %%"REG_S" \n\t"
|
|
|
2130 " jnz 1b \n\t"
|
|
|
2131 "emms \n\t"
|
|
|
2132 :: "r" (s16+1536), "r" (f+256)
|
|
|
2133 - :"%esi", "%edi", "memory"
|
|
|
2134 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
2135 );
|
|
|
2136 return 6*256;
|
|
|
2137 }
|
|
|
2138 @@ -429,23 +431,23 @@
|
|
|
2139 static int a52_resample_3F_2R_LFE_to_6_MMX(float * _f, int16_t * s16){
|
|
|
2140 int32_t * f = (int32_t *) _f;
|
|
|
2141 asm volatile(
|
|
|
2142 - "movl $-1024, %%esi \n\t"
|
|
|
2143 + "mov $-1024, %%"REG_S" \n\t"
|
|
|
2144 "movq "MANGLE(magicF2W)", %%mm7 \n\t"
|
|
|
2145 // "pxor %%mm6, %%mm6 \n\t"
|
|
|
2146 "1: \n\t"
|
|
|
2147 - "movq 1024(%1, %%esi), %%mm0 \n\t"
|
|
|
2148 - "movq 3072(%1, %%esi), %%mm1 \n\t"
|
|
|
2149 - "movq 4096(%1, %%esi), %%mm2 \n\t"
|
|
|
2150 - "movq 5120(%1, %%esi), %%mm3 \n\t"
|
|
|
2151 - "movq 2048(%1, %%esi), %%mm4 \n\t"
|
|
|
2152 - "movq (%1, %%esi), %%mm5 \n\t"
|
|
|
2153 + "movq 1024(%1, %%"REG_S"), %%mm0\n\t"
|
|
|
2154 + "movq 3072(%1, %%"REG_S"), %%mm1\n\t"
|
|
|
2155 + "movq 4096(%1, %%"REG_S"), %%mm2\n\t"
|
|
|
2156 + "movq 5120(%1, %%"REG_S"), %%mm3\n\t"
|
|
|
2157 + "movq 2048(%1, %%"REG_S"), %%mm4\n\t"
|
|
|
2158 + "movq (%1, %%"REG_S"), %%mm5 \n\t"
|
|
|
2159 "psubd %%mm7, %%mm0 \n\t"
|
|
|
2160 "psubd %%mm7, %%mm1 \n\t"
|
|
|
2161 "psubd %%mm7, %%mm2 \n\t"
|
|
|
2162 "psubd %%mm7, %%mm3 \n\t"
|
|
|
2163 "psubd %%mm7, %%mm4 \n\t"
|
|
|
2164 "psubd %%mm7, %%mm5 \n\t"
|
|
|
2165 - "leal (%%esi, %%esi, 2), %%edi \n\t"
|
|
|
2166 + "lea (%%"REG_S", %%"REG_S", 2), %%"REG_D"\n\t"
|
|
|
2167
|
|
|
2168 "packssdw %%mm2, %%mm0 \n\t" // CcAa
|
|
|
2169 "packssdw %%mm3, %%mm1 \n\t" // DdBb
|
|
|
2170 @@ -462,14 +464,14 @@
|
|
|
2171 "punpckldq %%mm1, %%mm4 \n\t" // BAfe
|
|
|
2172 "punpckhdq %%mm3, %%mm2 \n\t" // FEDC
|
|
|
2173
|
|
|
2174 - "movq %%mm0, (%0, %%edi) \n\t"
|
|
|
2175 - "movq %%mm4, 8(%0, %%edi) \n\t"
|
|
|
2176 - "movq %%mm2, 16(%0, %%edi) \n\t"
|
|
|
2177 - "addl $8, %%esi \n\t"
|
|
|
2178 + "movq %%mm0, (%0, %%"REG_D") \n\t"
|
|
|
2179 + "movq %%mm4, 8(%0, %%"REG_D") \n\t"
|
|
|
2180 + "movq %%mm2, 16(%0, %%"REG_D") \n\t"
|
|
|
2181 + "add $8, %%"REG_S" \n\t"
|
|
|
2182 " jnz 1b \n\t"
|
|
|
2183 "emms \n\t"
|
|
|
2184 :: "r" (s16+1536), "r" (f+256)
|
|
|
2185 - :"%esi", "%edi", "memory"
|
|
|
2186 + :"%"REG_S, "%"REG_D, "memory"
|
|
|
2187 );
|
|
|
2188 return 6*256;
|
|
|
2189 }
|
