1 /*
2 * Copyright (c) 2012
3 * MIPS Technologies, Inc., California.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
14 * contributors may be used to endorse or promote products derived from
15 * this software without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * Authors: Darko Laus (darko@mips.com)
30 * Djordje Pesut (djordje@mips.com)
31 * Mirjana Vulin (mvulin@mips.com)
32 *
33 * AAC Spectral Band Replication decoding functions optimized for MIPS
34 *
35 * This file is part of FFmpeg.
36 *
37 * FFmpeg is free software; you can redistribute it and/or
38 * modify it under the terms of the GNU Lesser General Public
39 * License as published by the Free Software Foundation; either
40 * version 2.1 of the License, or (at your option) any later version.
41 *
42 * FFmpeg is distributed in the hope that it will be useful,
43 * but WITHOUT ANY WARRANTY; without even the implied warranty of
44 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
45 * Lesser General Public License for more details.
46 *
47 * You should have received a copy of the GNU Lesser General Public
48 * License along with FFmpeg; if not, write to the Free Software
49 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
50 */
51
52 /**
53 * @file
54 * Reference: libavcodec/sbrdsp.c
55 */
56
57 #include "config.h"
58 #include "libavcodec/sbrdsp.h"
59 #include "libavutil/mips/asmdefs.h"
60
61 #if HAVE_INLINE_ASM
62 #if HAVE_MIPSFPU
sbr_qmf_pre_shuffle_mips(float * z)63 static void sbr_qmf_pre_shuffle_mips(float *z)
64 {
65 int Temp1, Temp2, Temp3, Temp4, Temp5, Temp6;
66 float *z1 = &z[66];
67 float *z2 = &z[59];
68 float *z3 = &z[2];
69 float *z4 = z1 + 60;
70
71 /* loop unrolled 5 times */
72 __asm__ volatile (
73 "lui %[Temp6], 0x8000 \n\t"
74 "1: \n\t"
75 "lw %[Temp1], 0(%[z2]) \n\t"
76 "lw %[Temp2], 4(%[z2]) \n\t"
77 "lw %[Temp3], 8(%[z2]) \n\t"
78 "lw %[Temp4], 12(%[z2]) \n\t"
79 "lw %[Temp5], 16(%[z2]) \n\t"
80 "xor %[Temp1], %[Temp1], %[Temp6] \n\t"
81 "xor %[Temp2], %[Temp2], %[Temp6] \n\t"
82 "xor %[Temp3], %[Temp3], %[Temp6] \n\t"
83 "xor %[Temp4], %[Temp4], %[Temp6] \n\t"
84 "xor %[Temp5], %[Temp5], %[Temp6] \n\t"
85 PTR_ADDIU "%[z2], %[z2], -20 \n\t"
86 "sw %[Temp1], 32(%[z1]) \n\t"
87 "sw %[Temp2], 24(%[z1]) \n\t"
88 "sw %[Temp3], 16(%[z1]) \n\t"
89 "sw %[Temp4], 8(%[z1]) \n\t"
90 "sw %[Temp5], 0(%[z1]) \n\t"
91 "lw %[Temp1], 0(%[z3]) \n\t"
92 "lw %[Temp2], 4(%[z3]) \n\t"
93 "lw %[Temp3], 8(%[z3]) \n\t"
94 "lw %[Temp4], 12(%[z3]) \n\t"
95 "lw %[Temp5], 16(%[z3]) \n\t"
96 "sw %[Temp1], 4(%[z1]) \n\t"
97 "sw %[Temp2], 12(%[z1]) \n\t"
98 "sw %[Temp3], 20(%[z1]) \n\t"
99 "sw %[Temp4], 28(%[z1]) \n\t"
100 "sw %[Temp5], 36(%[z1]) \n\t"
101 PTR_ADDIU "%[z3], %[z3], 20 \n\t"
102 PTR_ADDIU "%[z1], %[z1], 40 \n\t"
103 "bne %[z1], %[z4], 1b \n\t"
104 "lw %[Temp1], 132(%[z]) \n\t"
105 "lw %[Temp2], 128(%[z]) \n\t"
106 "lw %[Temp3], 0(%[z]) \n\t"
107 "lw %[Temp4], 4(%[z]) \n\t"
108 "xor %[Temp1], %[Temp1], %[Temp6] \n\t"
109 "sw %[Temp1], 504(%[z]) \n\t"
110 "sw %[Temp2], 508(%[z]) \n\t"
111 "sw %[Temp3], 256(%[z]) \n\t"
112 "sw %[Temp4], 260(%[z]) \n\t"
113
114 : [Temp1]"=&r"(Temp1), [Temp2]"=&r"(Temp2),
115 [Temp3]"=&r"(Temp3), [Temp4]"=&r"(Temp4),
116 [Temp5]"=&r"(Temp5), [Temp6]"=&r"(Temp6),
117 [z1]"+r"(z1), [z2]"+r"(z2), [z3]"+r"(z3)
118 : [z4]"r"(z4), [z]"r"(z)
119 : "memory"
120 );
121 }
122
sbr_qmf_post_shuffle_mips(float W[32][2],const float * z)123 static void sbr_qmf_post_shuffle_mips(float W[32][2], const float *z)
124 {
125 int Temp1, Temp2, Temp3, Temp4, Temp5;
126 float *W_ptr = (float *)W;
127 float *z1 = (float *)z;
128 float *z2 = (float *)&z[60];
129 float *z_end = z1 + 32;
130
131 /* loop unrolled 4 times */
132 __asm__ volatile (
133 "lui %[Temp5], 0x8000 \n\t"
134 "1: \n\t"
135 "lw %[Temp1], 0(%[z2]) \n\t"
136 "lw %[Temp2], 4(%[z2]) \n\t"
137 "lw %[Temp3], 8(%[z2]) \n\t"
138 "lw %[Temp4], 12(%[z2]) \n\t"
139 "xor %[Temp1], %[Temp1], %[Temp5] \n\t"
140 "xor %[Temp2], %[Temp2], %[Temp5] \n\t"
141 "xor %[Temp3], %[Temp3], %[Temp5] \n\t"
142 "xor %[Temp4], %[Temp4], %[Temp5] \n\t"
143 PTR_ADDIU "%[z2], %[z2], -16 \n\t"
144 "sw %[Temp1], 24(%[W_ptr]) \n\t"
145 "sw %[Temp2], 16(%[W_ptr]) \n\t"
146 "sw %[Temp3], 8(%[W_ptr]) \n\t"
147 "sw %[Temp4], 0(%[W_ptr]) \n\t"
148 "lw %[Temp1], 0(%[z1]) \n\t"
149 "lw %[Temp2], 4(%[z1]) \n\t"
150 "lw %[Temp3], 8(%[z1]) \n\t"
151 "lw %[Temp4], 12(%[z1]) \n\t"
152 "sw %[Temp1], 4(%[W_ptr]) \n\t"
153 "sw %[Temp2], 12(%[W_ptr]) \n\t"
154 "sw %[Temp3], 20(%[W_ptr]) \n\t"
155 "sw %[Temp4], 28(%[W_ptr]) \n\t"
156 PTR_ADDIU "%[z1], %[z1], 16 \n\t"
157 PTR_ADDIU "%[W_ptr],%[W_ptr], 32 \n\t"
158 "bne %[z1], %[z_end], 1b \n\t"
159
160 : [Temp1]"=&r"(Temp1), [Temp2]"=&r"(Temp2),
161 [Temp3]"=&r"(Temp3), [Temp4]"=&r"(Temp4),
162 [Temp5]"=&r"(Temp5), [z1]"+r"(z1),
163 [z2]"+r"(z2), [W_ptr]"+r"(W_ptr)
164 : [z_end]"r"(z_end)
165 : "memory"
166 );
167 }
168
169 #if !HAVE_MIPS32R6 && !HAVE_MIPS64R6
sbr_sum64x5_mips(float * z)170 static void sbr_sum64x5_mips(float *z)
171 {
172 int k;
173 float *z1;
174 float f1, f2, f3, f4, f5, f6, f7, f8;
175 for (k = 0; k < 64; k += 8) {
176
177 z1 = &z[k];
178
179 /* loop unrolled 8 times */
180 __asm__ volatile (
181 "lwc1 $f0, 0(%[z1]) \n\t"
182 "lwc1 $f1, 256(%[z1]) \n\t"
183 "lwc1 $f2, 4(%[z1]) \n\t"
184 "lwc1 $f3, 260(%[z1]) \n\t"
185 "lwc1 $f4, 8(%[z1]) \n\t"
186 "add.s %[f1], $f0, $f1 \n\t"
187 "lwc1 $f5, 264(%[z1]) \n\t"
188 "add.s %[f2], $f2, $f3 \n\t"
189 "lwc1 $f6, 12(%[z1]) \n\t"
190 "lwc1 $f7, 268(%[z1]) \n\t"
191 "add.s %[f3], $f4, $f5 \n\t"
192 "lwc1 $f8, 16(%[z1]) \n\t"
193 "lwc1 $f9, 272(%[z1]) \n\t"
194 "add.s %[f4], $f6, $f7 \n\t"
195 "lwc1 $f10, 20(%[z1]) \n\t"
196 "lwc1 $f11, 276(%[z1]) \n\t"
197 "add.s %[f5], $f8, $f9 \n\t"
198 "lwc1 $f12, 24(%[z1]) \n\t"
199 "lwc1 $f13, 280(%[z1]) \n\t"
200 "add.s %[f6], $f10, $f11 \n\t"
201 "lwc1 $f14, 28(%[z1]) \n\t"
202 "lwc1 $f15, 284(%[z1]) \n\t"
203 "add.s %[f7], $f12, $f13 \n\t"
204 "lwc1 $f0, 512(%[z1]) \n\t"
205 "lwc1 $f1, 516(%[z1]) \n\t"
206 "add.s %[f8], $f14, $f15 \n\t"
207 "lwc1 $f2, 520(%[z1]) \n\t"
208 "add.s %[f1], %[f1], $f0 \n\t"
209 "add.s %[f2], %[f2], $f1 \n\t"
210 "lwc1 $f3, 524(%[z1]) \n\t"
211 "add.s %[f3], %[f3], $f2 \n\t"
212 "lwc1 $f4, 528(%[z1]) \n\t"
213 "lwc1 $f5, 532(%[z1]) \n\t"
214 "add.s %[f4], %[f4], $f3 \n\t"
215 "lwc1 $f6, 536(%[z1]) \n\t"
216 "add.s %[f5], %[f5], $f4 \n\t"
217 "add.s %[f6], %[f6], $f5 \n\t"
218 "lwc1 $f7, 540(%[z1]) \n\t"
219 "add.s %[f7], %[f7], $f6 \n\t"
220 "lwc1 $f0, 768(%[z1]) \n\t"
221 "lwc1 $f1, 772(%[z1]) \n\t"
222 "add.s %[f8], %[f8], $f7 \n\t"
223 "lwc1 $f2, 776(%[z1]) \n\t"
224 "add.s %[f1], %[f1], $f0 \n\t"
225 "add.s %[f2], %[f2], $f1 \n\t"
226 "lwc1 $f3, 780(%[z1]) \n\t"
227 "add.s %[f3], %[f3], $f2 \n\t"
228 "lwc1 $f4, 784(%[z1]) \n\t"
229 "lwc1 $f5, 788(%[z1]) \n\t"
230 "add.s %[f4], %[f4], $f3 \n\t"
231 "lwc1 $f6, 792(%[z1]) \n\t"
232 "add.s %[f5], %[f5], $f4 \n\t"
233 "add.s %[f6], %[f6], $f5 \n\t"
234 "lwc1 $f7, 796(%[z1]) \n\t"
235 "add.s %[f7], %[f7], $f6 \n\t"
236 "lwc1 $f0, 1024(%[z1]) \n\t"
237 "lwc1 $f1, 1028(%[z1]) \n\t"
238 "add.s %[f8], %[f8], $f7 \n\t"
239 "lwc1 $f2, 1032(%[z1]) \n\t"
240 "add.s %[f1], %[f1], $f0 \n\t"
241 "add.s %[f2], %[f2], $f1 \n\t"
242 "lwc1 $f3, 1036(%[z1]) \n\t"
243 "add.s %[f3], %[f3], $f2 \n\t"
244 "lwc1 $f4, 1040(%[z1]) \n\t"
245 "lwc1 $f5, 1044(%[z1]) \n\t"
246 "add.s %[f4], %[f4], $f3 \n\t"
247 "lwc1 $f6, 1048(%[z1]) \n\t"
248 "add.s %[f5], %[f5], $f4 \n\t"
249 "add.s %[f6], %[f6], $f5 \n\t"
250 "lwc1 $f7, 1052(%[z1]) \n\t"
251 "add.s %[f7], %[f7], $f6 \n\t"
252 "swc1 %[f1], 0(%[z1]) \n\t"
253 "swc1 %[f2], 4(%[z1]) \n\t"
254 "add.s %[f8], %[f8], $f7 \n\t"
255 "swc1 %[f3], 8(%[z1]) \n\t"
256 "swc1 %[f4], 12(%[z1]) \n\t"
257 "swc1 %[f5], 16(%[z1]) \n\t"
258 "swc1 %[f6], 20(%[z1]) \n\t"
259 "swc1 %[f7], 24(%[z1]) \n\t"
260 "swc1 %[f8], 28(%[z1]) \n\t"
261
262 : [f1]"=&f"(f1), [f2]"=&f"(f2), [f3]"=&f"(f3),
263 [f4]"=&f"(f4), [f5]"=&f"(f5), [f6]"=&f"(f6),
264 [f7]"=&f"(f7), [f8]"=&f"(f8)
265 : [z1]"r"(z1)
266 : "$f0", "$f1", "$f2", "$f3", "$f4", "$f5",
267 "$f6", "$f7", "$f8", "$f9", "$f10", "$f11",
268 "$f12", "$f13", "$f14", "$f15",
269 "memory"
270 );
271 }
272 }
273
sbr_sum_square_mips(float (* x)[2],int n)274 static float sbr_sum_square_mips(float (*x)[2], int n)
275 {
276 float sum0 = 0.0f, sum1 = 0.0f;
277 float *p_x;
278 float temp0, temp1, temp2, temp3;
279 float *loop_end;
280 p_x = &x[0][0];
281 loop_end = p_x + (n >> 1)*4 - 4;
282
283 __asm__ volatile (
284 ".set push \n\t"
285 ".set noreorder \n\t"
286 "lwc1 %[temp0], 0(%[p_x]) \n\t"
287 "lwc1 %[temp1], 4(%[p_x]) \n\t"
288 "lwc1 %[temp2], 8(%[p_x]) \n\t"
289 "lwc1 %[temp3], 12(%[p_x]) \n\t"
290 "1: \n\t"
291 PTR_ADDIU "%[p_x], %[p_x], 16 \n\t"
292 "madd.s %[sum0], %[sum0], %[temp0], %[temp0] \n\t"
293 "lwc1 %[temp0], 0(%[p_x]) \n\t"
294 "madd.s %[sum1], %[sum1], %[temp1], %[temp1] \n\t"
295 "lwc1 %[temp1], 4(%[p_x]) \n\t"
296 "madd.s %[sum0], %[sum0], %[temp2], %[temp2] \n\t"
297 "lwc1 %[temp2], 8(%[p_x]) \n\t"
298 "madd.s %[sum1], %[sum1], %[temp3], %[temp3] \n\t"
299 "bne %[p_x], %[loop_end], 1b \n\t"
300 " lwc1 %[temp3], 12(%[p_x]) \n\t"
301 "madd.s %[sum0], %[sum0], %[temp0], %[temp0] \n\t"
302 "madd.s %[sum1], %[sum1], %[temp1], %[temp1] \n\t"
303 "madd.s %[sum0], %[sum0], %[temp2], %[temp2] \n\t"
304 "madd.s %[sum1], %[sum1], %[temp3], %[temp3] \n\t"
305 ".set pop \n\t"
306
307 : [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
308 [temp3]"=&f"(temp3), [sum0]"+f"(sum0), [sum1]"+f"(sum1),
309 [p_x]"+r"(p_x)
310 : [loop_end]"r"(loop_end)
311 : "memory"
312 );
313 return sum0 + sum1;
314 }
315
sbr_qmf_deint_bfly_mips(float * v,const float * src0,const float * src1)316 static void sbr_qmf_deint_bfly_mips(float *v, const float *src0, const float *src1)
317 {
318 int i;
319 float temp0, temp1, temp2, temp3, temp4, temp5;
320 float temp6, temp7, temp8, temp9, temp10, temp11;
321 float *v0 = v;
322 float *v1 = &v[127];
323 float *psrc0 = (float*)src0;
324 float *psrc1 = (float*)&src1[63];
325
326 for (i = 0; i < 4; i++) {
327
328 /* loop unrolled 16 times */
329 __asm__ volatile(
330 "lwc1 %[temp0], 0(%[src0]) \n\t"
331 "lwc1 %[temp1], 0(%[src1]) \n\t"
332 "lwc1 %[temp3], 4(%[src0]) \n\t"
333 "lwc1 %[temp4], -4(%[src1]) \n\t"
334 "lwc1 %[temp6], 8(%[src0]) \n\t"
335 "lwc1 %[temp7], -8(%[src1]) \n\t"
336 "lwc1 %[temp9], 12(%[src0]) \n\t"
337 "lwc1 %[temp10], -12(%[src1]) \n\t"
338 "add.s %[temp2], %[temp0], %[temp1] \n\t"
339 "add.s %[temp5], %[temp3], %[temp4] \n\t"
340 "add.s %[temp8], %[temp6], %[temp7] \n\t"
341 "add.s %[temp11], %[temp9], %[temp10] \n\t"
342 "sub.s %[temp0], %[temp0], %[temp1] \n\t"
343 "sub.s %[temp3], %[temp3], %[temp4] \n\t"
344 "sub.s %[temp6], %[temp6], %[temp7] \n\t"
345 "sub.s %[temp9], %[temp9], %[temp10] \n\t"
346 "swc1 %[temp2], 0(%[v1]) \n\t"
347 "swc1 %[temp0], 0(%[v0]) \n\t"
348 "swc1 %[temp5], -4(%[v1]) \n\t"
349 "swc1 %[temp3], 4(%[v0]) \n\t"
350 "swc1 %[temp8], -8(%[v1]) \n\t"
351 "swc1 %[temp6], 8(%[v0]) \n\t"
352 "swc1 %[temp11], -12(%[v1]) \n\t"
353 "swc1 %[temp9], 12(%[v0]) \n\t"
354 "lwc1 %[temp0], 16(%[src0]) \n\t"
355 "lwc1 %[temp1], -16(%[src1]) \n\t"
356 "lwc1 %[temp3], 20(%[src0]) \n\t"
357 "lwc1 %[temp4], -20(%[src1]) \n\t"
358 "lwc1 %[temp6], 24(%[src0]) \n\t"
359 "lwc1 %[temp7], -24(%[src1]) \n\t"
360 "lwc1 %[temp9], 28(%[src0]) \n\t"
361 "lwc1 %[temp10], -28(%[src1]) \n\t"
362 "add.s %[temp2], %[temp0], %[temp1] \n\t"
363 "add.s %[temp5], %[temp3], %[temp4] \n\t"
364 "add.s %[temp8], %[temp6], %[temp7] \n\t"
365 "add.s %[temp11], %[temp9], %[temp10] \n\t"
366 "sub.s %[temp0], %[temp0], %[temp1] \n\t"
367 "sub.s %[temp3], %[temp3], %[temp4] \n\t"
368 "sub.s %[temp6], %[temp6], %[temp7] \n\t"
369 "sub.s %[temp9], %[temp9], %[temp10] \n\t"
370 "swc1 %[temp2], -16(%[v1]) \n\t"
371 "swc1 %[temp0], 16(%[v0]) \n\t"
372 "swc1 %[temp5], -20(%[v1]) \n\t"
373 "swc1 %[temp3], 20(%[v0]) \n\t"
374 "swc1 %[temp8], -24(%[v1]) \n\t"
375 "swc1 %[temp6], 24(%[v0]) \n\t"
376 "swc1 %[temp11], -28(%[v1]) \n\t"
377 "swc1 %[temp9], 28(%[v0]) \n\t"
378 "lwc1 %[temp0], 32(%[src0]) \n\t"
379 "lwc1 %[temp1], -32(%[src1]) \n\t"
380 "lwc1 %[temp3], 36(%[src0]) \n\t"
381 "lwc1 %[temp4], -36(%[src1]) \n\t"
382 "lwc1 %[temp6], 40(%[src0]) \n\t"
383 "lwc1 %[temp7], -40(%[src1]) \n\t"
384 "lwc1 %[temp9], 44(%[src0]) \n\t"
385 "lwc1 %[temp10], -44(%[src1]) \n\t"
386 "add.s %[temp2], %[temp0], %[temp1] \n\t"
387 "add.s %[temp5], %[temp3], %[temp4] \n\t"
388 "add.s %[temp8], %[temp6], %[temp7] \n\t"
389 "add.s %[temp11], %[temp9], %[temp10] \n\t"
390 "sub.s %[temp0], %[temp0], %[temp1] \n\t"
391 "sub.s %[temp3], %[temp3], %[temp4] \n\t"
392 "sub.s %[temp6], %[temp6], %[temp7] \n\t"
393 "sub.s %[temp9], %[temp9], %[temp10] \n\t"
394 "swc1 %[temp2], -32(%[v1]) \n\t"
395 "swc1 %[temp0], 32(%[v0]) \n\t"
396 "swc1 %[temp5], -36(%[v1]) \n\t"
397 "swc1 %[temp3], 36(%[v0]) \n\t"
398 "swc1 %[temp8], -40(%[v1]) \n\t"
399 "swc1 %[temp6], 40(%[v0]) \n\t"
400 "swc1 %[temp11], -44(%[v1]) \n\t"
401 "swc1 %[temp9], 44(%[v0]) \n\t"
402 "lwc1 %[temp0], 48(%[src0]) \n\t"
403 "lwc1 %[temp1], -48(%[src1]) \n\t"
404 "lwc1 %[temp3], 52(%[src0]) \n\t"
405 "lwc1 %[temp4], -52(%[src1]) \n\t"
406 "lwc1 %[temp6], 56(%[src0]) \n\t"
407 "lwc1 %[temp7], -56(%[src1]) \n\t"
408 "lwc1 %[temp9], 60(%[src0]) \n\t"
409 "lwc1 %[temp10], -60(%[src1]) \n\t"
410 "add.s %[temp2], %[temp0], %[temp1] \n\t"
411 "add.s %[temp5], %[temp3], %[temp4] \n\t"
412 "add.s %[temp8], %[temp6], %[temp7] \n\t"
413 "add.s %[temp11], %[temp9], %[temp10] \n\t"
414 "sub.s %[temp0], %[temp0], %[temp1] \n\t"
415 "sub.s %[temp3], %[temp3], %[temp4] \n\t"
416 "sub.s %[temp6], %[temp6], %[temp7] \n\t"
417 "sub.s %[temp9], %[temp9], %[temp10] \n\t"
418 "swc1 %[temp2], -48(%[v1]) \n\t"
419 "swc1 %[temp0], 48(%[v0]) \n\t"
420 "swc1 %[temp5], -52(%[v1]) \n\t"
421 "swc1 %[temp3], 52(%[v0]) \n\t"
422 "swc1 %[temp8], -56(%[v1]) \n\t"
423 "swc1 %[temp6], 56(%[v0]) \n\t"
424 "swc1 %[temp11], -60(%[v1]) \n\t"
425 "swc1 %[temp9], 60(%[v0]) \n\t"
426 PTR_ADDIU " %[src0], %[src0], 64 \n\t"
427 PTR_ADDIU " %[src1], %[src1], -64 \n\t"
428 PTR_ADDIU " %[v0], %[v0], 64 \n\t"
429 PTR_ADDIU " %[v1], %[v1], -64 \n\t"
430
431 : [v0]"+r"(v0), [v1]"+r"(v1), [src0]"+r"(psrc0), [src1]"+r"(psrc1),
432 [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
433 [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
434 [temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp8]"=&f"(temp8),
435 [temp9]"=&f"(temp9), [temp10]"=&f"(temp10), [temp11]"=&f"(temp11)
436 :
437 :"memory"
438 );
439 }
440 }
441
sbr_autocorrelate_mips(const float x[40][2],float phi[3][2][2])442 static void sbr_autocorrelate_mips(const float x[40][2], float phi[3][2][2])
443 {
444 int i;
445 float real_sum_0 = 0.0f;
446 float real_sum_1 = 0.0f;
447 float real_sum_2 = 0.0f;
448 float imag_sum_1 = 0.0f;
449 float imag_sum_2 = 0.0f;
450 float *p_x, *p_phi;
451 float temp0, temp1, temp2, temp3, temp4, temp5, temp6;
452 float temp7, temp_r, temp_r1, temp_r2, temp_r3, temp_r4;
453 p_x = (float*)&x[0][0];
454 p_phi = &phi[0][0][0];
455
456 __asm__ volatile (
457 "lwc1 %[temp0], 8(%[p_x]) \n\t"
458 "lwc1 %[temp1], 12(%[p_x]) \n\t"
459 "lwc1 %[temp2], 16(%[p_x]) \n\t"
460 "lwc1 %[temp3], 20(%[p_x]) \n\t"
461 "lwc1 %[temp4], 24(%[p_x]) \n\t"
462 "lwc1 %[temp5], 28(%[p_x]) \n\t"
463 "mul.s %[temp_r], %[temp1], %[temp1] \n\t"
464 "mul.s %[temp_r1], %[temp1], %[temp3] \n\t"
465 "mul.s %[temp_r2], %[temp1], %[temp2] \n\t"
466 "mul.s %[temp_r3], %[temp1], %[temp5] \n\t"
467 "mul.s %[temp_r4], %[temp1], %[temp4] \n\t"
468 "madd.s %[temp_r], %[temp_r], %[temp0], %[temp0] \n\t"
469 "madd.s %[temp_r1], %[temp_r1], %[temp0], %[temp2] \n\t"
470 "msub.s %[temp_r2], %[temp_r2], %[temp0], %[temp3] \n\t"
471 "madd.s %[temp_r3], %[temp_r3], %[temp0], %[temp4] \n\t"
472 "msub.s %[temp_r4], %[temp_r4], %[temp0], %[temp5] \n\t"
473 "add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
474 "add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
475 "add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
476 "add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
477 "add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
478 PTR_ADDIU "%[p_x], %[p_x], 8 \n\t"
479
480 : [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
481 [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
482 [real_sum_0]"+f"(real_sum_0), [real_sum_1]"+f"(real_sum_1),
483 [imag_sum_1]"+f"(imag_sum_1), [real_sum_2]"+f"(real_sum_2),
484 [temp_r]"=&f"(temp_r), [temp_r1]"=&f"(temp_r1), [temp_r2]"=&f"(temp_r2),
485 [temp_r3]"=&f"(temp_r3), [temp_r4]"=&f"(temp_r4),
486 [p_x]"+r"(p_x), [imag_sum_2]"+f"(imag_sum_2)
487 :
488 : "memory"
489 );
490
491 for (i = 0; i < 12; i++) {
492 __asm__ volatile (
493 "lwc1 %[temp0], 8(%[p_x]) \n\t"
494 "lwc1 %[temp1], 12(%[p_x]) \n\t"
495 "lwc1 %[temp2], 16(%[p_x]) \n\t"
496 "lwc1 %[temp3], 20(%[p_x]) \n\t"
497 "lwc1 %[temp4], 24(%[p_x]) \n\t"
498 "lwc1 %[temp5], 28(%[p_x]) \n\t"
499 "mul.s %[temp_r], %[temp1], %[temp1] \n\t"
500 "mul.s %[temp_r1], %[temp1], %[temp3] \n\t"
501 "mul.s %[temp_r2], %[temp1], %[temp2] \n\t"
502 "mul.s %[temp_r3], %[temp1], %[temp5] \n\t"
503 "mul.s %[temp_r4], %[temp1], %[temp4] \n\t"
504 "madd.s %[temp_r], %[temp_r], %[temp0], %[temp0] \n\t"
505 "madd.s %[temp_r1], %[temp_r1], %[temp0], %[temp2] \n\t"
506 "msub.s %[temp_r2], %[temp_r2], %[temp0], %[temp3] \n\t"
507 "madd.s %[temp_r3], %[temp_r3], %[temp0], %[temp4] \n\t"
508 "msub.s %[temp_r4], %[temp_r4], %[temp0], %[temp5] \n\t"
509 "add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
510 "add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
511 "add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
512 "add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
513 "add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
514 "lwc1 %[temp0], 32(%[p_x]) \n\t"
515 "lwc1 %[temp1], 36(%[p_x]) \n\t"
516 "mul.s %[temp_r], %[temp3], %[temp3] \n\t"
517 "mul.s %[temp_r1], %[temp3], %[temp5] \n\t"
518 "mul.s %[temp_r2], %[temp3], %[temp4] \n\t"
519 "mul.s %[temp_r3], %[temp3], %[temp1] \n\t"
520 "mul.s %[temp_r4], %[temp3], %[temp0] \n\t"
521 "madd.s %[temp_r], %[temp_r], %[temp2], %[temp2] \n\t"
522 "madd.s %[temp_r1], %[temp_r1], %[temp2], %[temp4] \n\t"
523 "msub.s %[temp_r2], %[temp_r2], %[temp2], %[temp5] \n\t"
524 "madd.s %[temp_r3], %[temp_r3], %[temp2], %[temp0] \n\t"
525 "msub.s %[temp_r4], %[temp_r4], %[temp2], %[temp1] \n\t"
526 "add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
527 "add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
528 "add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
529 "add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
530 "add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
531 "lwc1 %[temp2], 40(%[p_x]) \n\t"
532 "lwc1 %[temp3], 44(%[p_x]) \n\t"
533 "mul.s %[temp_r], %[temp5], %[temp5] \n\t"
534 "mul.s %[temp_r1], %[temp5], %[temp1] \n\t"
535 "mul.s %[temp_r2], %[temp5], %[temp0] \n\t"
536 "mul.s %[temp_r3], %[temp5], %[temp3] \n\t"
537 "mul.s %[temp_r4], %[temp5], %[temp2] \n\t"
538 "madd.s %[temp_r], %[temp_r], %[temp4], %[temp4] \n\t"
539 "madd.s %[temp_r1], %[temp_r1], %[temp4], %[temp0] \n\t"
540 "msub.s %[temp_r2], %[temp_r2], %[temp4], %[temp1] \n\t"
541 "madd.s %[temp_r3], %[temp_r3], %[temp4], %[temp2] \n\t"
542 "msub.s %[temp_r4], %[temp_r4], %[temp4], %[temp3] \n\t"
543 "add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
544 "add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
545 "add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
546 "add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
547 "add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
548 PTR_ADDIU "%[p_x], %[p_x], 24 \n\t"
549
550 : [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
551 [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
552 [real_sum_0]"+f"(real_sum_0), [real_sum_1]"+f"(real_sum_1),
553 [imag_sum_1]"+f"(imag_sum_1), [real_sum_2]"+f"(real_sum_2),
554 [temp_r]"=&f"(temp_r), [temp_r1]"=&f"(temp_r1),
555 [temp_r2]"=&f"(temp_r2), [temp_r3]"=&f"(temp_r3),
556 [temp_r4]"=&f"(temp_r4), [p_x]"+r"(p_x),
557 [imag_sum_2]"+f"(imag_sum_2)
558 :
559 : "memory"
560 );
561 }
562 __asm__ volatile (
563 "lwc1 %[temp0], -296(%[p_x]) \n\t"
564 "lwc1 %[temp1], -292(%[p_x]) \n\t"
565 "lwc1 %[temp2], 8(%[p_x]) \n\t"
566 "lwc1 %[temp3], 12(%[p_x]) \n\t"
567 "lwc1 %[temp4], -288(%[p_x]) \n\t"
568 "lwc1 %[temp5], -284(%[p_x]) \n\t"
569 "lwc1 %[temp6], -280(%[p_x]) \n\t"
570 "lwc1 %[temp7], -276(%[p_x]) \n\t"
571 "madd.s %[temp_r], %[real_sum_0], %[temp0], %[temp0] \n\t"
572 "madd.s %[temp_r1], %[real_sum_0], %[temp2], %[temp2] \n\t"
573 "madd.s %[temp_r2], %[real_sum_1], %[temp0], %[temp4] \n\t"
574 "madd.s %[temp_r3], %[imag_sum_1], %[temp0], %[temp5] \n\t"
575 "madd.s %[temp_r], %[temp_r], %[temp1], %[temp1] \n\t"
576 "madd.s %[temp_r1], %[temp_r1], %[temp3], %[temp3] \n\t"
577 "madd.s %[temp_r2], %[temp_r2], %[temp1], %[temp5] \n\t"
578 "nmsub.s %[temp_r3], %[temp_r3], %[temp1], %[temp4] \n\t"
579 "lwc1 %[temp4], 16(%[p_x]) \n\t"
580 "lwc1 %[temp5], 20(%[p_x]) \n\t"
581 "swc1 %[temp_r], 40(%[p_phi]) \n\t"
582 "swc1 %[temp_r1], 16(%[p_phi]) \n\t"
583 "swc1 %[temp_r2], 24(%[p_phi]) \n\t"
584 "swc1 %[temp_r3], 28(%[p_phi]) \n\t"
585 "madd.s %[temp_r], %[real_sum_1], %[temp2], %[temp4] \n\t"
586 "madd.s %[temp_r1], %[imag_sum_1], %[temp2], %[temp5] \n\t"
587 "madd.s %[temp_r2], %[real_sum_2], %[temp0], %[temp6] \n\t"
588 "madd.s %[temp_r3], %[imag_sum_2], %[temp0], %[temp7] \n\t"
589 "madd.s %[temp_r], %[temp_r], %[temp3], %[temp5] \n\t"
590 "nmsub.s %[temp_r1], %[temp_r1], %[temp3], %[temp4] \n\t"
591 "madd.s %[temp_r2], %[temp_r2], %[temp1], %[temp7] \n\t"
592 "nmsub.s %[temp_r3], %[temp_r3], %[temp1], %[temp6] \n\t"
593 "swc1 %[temp_r], 0(%[p_phi]) \n\t"
594 "swc1 %[temp_r1], 4(%[p_phi]) \n\t"
595 "swc1 %[temp_r2], 8(%[p_phi]) \n\t"
596 "swc1 %[temp_r3], 12(%[p_phi]) \n\t"
597
598 : [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
599 [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
600 [temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp_r]"=&f"(temp_r),
601 [real_sum_0]"+f"(real_sum_0), [real_sum_1]"+f"(real_sum_1),
602 [real_sum_2]"+f"(real_sum_2), [imag_sum_1]"+f"(imag_sum_1),
603 [temp_r2]"=&f"(temp_r2), [temp_r3]"=&f"(temp_r3),
604 [temp_r1]"=&f"(temp_r1), [p_phi]"+r"(p_phi),
605 [imag_sum_2]"+f"(imag_sum_2)
606 : [p_x]"r"(p_x)
607 : "memory"
608 );
609 }
610
sbr_hf_gen_mips(float (* X_high)[2],const float (* X_low)[2],const float alpha0[2],const float alpha1[2],float bw,int start,int end)611 static void sbr_hf_gen_mips(float (*X_high)[2], const float (*X_low)[2],
612 const float alpha0[2], const float alpha1[2],
613 float bw, int start, int end)
614 {
615 float alpha[4];
616 int i;
617 float *p_x_low = (float*)&X_low[0][0] + 2*start;
618 float *p_x_high = &X_high[0][0] + 2*start;
619 float temp0, temp1, temp2, temp3, temp4, temp5, temp6;
620 float temp7, temp8, temp9, temp10, temp11, temp12;
621
622 alpha[0] = alpha1[0] * bw * bw;
623 alpha[1] = alpha1[1] * bw * bw;
624 alpha[2] = alpha0[0] * bw;
625 alpha[3] = alpha0[1] * bw;
626
627 for (i = start; i < end; i++) {
628 __asm__ volatile (
629 "lwc1 %[temp0], -16(%[p_x_low]) \n\t"
630 "lwc1 %[temp1], -12(%[p_x_low]) \n\t"
631 "lwc1 %[temp2], -8(%[p_x_low]) \n\t"
632 "lwc1 %[temp3], -4(%[p_x_low]) \n\t"
633 "lwc1 %[temp5], 0(%[p_x_low]) \n\t"
634 "lwc1 %[temp6], 4(%[p_x_low]) \n\t"
635 "lwc1 %[temp7], 0(%[alpha]) \n\t"
636 "lwc1 %[temp8], 4(%[alpha]) \n\t"
637 "lwc1 %[temp9], 8(%[alpha]) \n\t"
638 "lwc1 %[temp10], 12(%[alpha]) \n\t"
639 PTR_ADDIU "%[p_x_high], %[p_x_high], 8 \n\t"
640 PTR_ADDIU "%[p_x_low], %[p_x_low], 8 \n\t"
641 "mul.s %[temp11], %[temp1], %[temp8] \n\t"
642 "msub.s %[temp11], %[temp11], %[temp0], %[temp7] \n\t"
643 "madd.s %[temp11], %[temp11], %[temp2], %[temp9] \n\t"
644 "nmsub.s %[temp11], %[temp11], %[temp3], %[temp10] \n\t"
645 "add.s %[temp11], %[temp11], %[temp5] \n\t"
646 "swc1 %[temp11], -8(%[p_x_high]) \n\t"
647 "mul.s %[temp12], %[temp1], %[temp7] \n\t"
648 "madd.s %[temp12], %[temp12], %[temp0], %[temp8] \n\t"
649 "madd.s %[temp12], %[temp12], %[temp3], %[temp9] \n\t"
650 "madd.s %[temp12], %[temp12], %[temp2], %[temp10] \n\t"
651 "add.s %[temp12], %[temp12], %[temp6] \n\t"
652 "swc1 %[temp12], -4(%[p_x_high]) \n\t"
653
654 : [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
655 [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
656 [temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp8]"=&f"(temp8),
657 [temp9]"=&f"(temp9), [temp10]"=&f"(temp10), [temp11]"=&f"(temp11),
658 [temp12]"=&f"(temp12), [p_x_high]"+r"(p_x_high),
659 [p_x_low]"+r"(p_x_low)
660 : [alpha]"r"(alpha)
661 : "memory"
662 );
663 }
664 }
665
sbr_hf_g_filt_mips(float (* Y)[2],const float (* X_high)[40][2],const float * g_filt,int m_max,intptr_t ixh)666 static void sbr_hf_g_filt_mips(float (*Y)[2], const float (*X_high)[40][2],
667 const float *g_filt, int m_max, intptr_t ixh)
668 {
669 const float *p_x, *p_g, *loop_end;
670 float *p_y;
671 float temp0, temp1, temp2;
672
673 p_g = &g_filt[0];
674 p_y = &Y[0][0];
675 p_x = &X_high[0][ixh][0];
676 loop_end = p_g + m_max;
677
678 __asm__ volatile(
679 ".set push \n\t"
680 ".set noreorder \n\t"
681 "1: \n\t"
682 "lwc1 %[temp0], 0(%[p_g]) \n\t"
683 "lwc1 %[temp1], 0(%[p_x]) \n\t"
684 "lwc1 %[temp2], 4(%[p_x]) \n\t"
685 "mul.s %[temp1], %[temp1], %[temp0] \n\t"
686 "mul.s %[temp2], %[temp2], %[temp0] \n\t"
687 PTR_ADDIU "%[p_g], %[p_g], 4 \n\t"
688 PTR_ADDIU "%[p_x], %[p_x], 320 \n\t"
689 "swc1 %[temp1], 0(%[p_y]) \n\t"
690 "swc1 %[temp2], 4(%[p_y]) \n\t"
691 "bne %[p_g], %[loop_end], 1b \n\t"
692 PTR_ADDIU "%[p_y], %[p_y], 8 \n\t"
693 ".set pop \n\t"
694
695 : [temp0]"=&f"(temp0), [temp1]"=&f"(temp1),
696 [temp2]"=&f"(temp2), [p_x]"+r"(p_x),
697 [p_y]"+r"(p_y), [p_g]"+r"(p_g)
698 : [loop_end]"r"(loop_end)
699 : "memory"
700 );
701 }
702
sbr_hf_apply_noise_0_mips(float (* Y)[2],const float * s_m,const float * q_filt,int noise,int kx,int m_max)703 static void sbr_hf_apply_noise_0_mips(float (*Y)[2], const float *s_m,
704 const float *q_filt, int noise,
705 int kx, int m_max)
706 {
707 int m;
708
709 for (m = 0; m < m_max; m++){
710
711 float *Y1=&Y[m][0];
712 float *ff_table;
713 float y0,y1, temp1, temp2, temp4, temp5;
714 int temp0, temp3;
715 const float *s_m1=&s_m[m];
716 const float *q_filt1= &q_filt[m];
717
718 __asm__ volatile(
719 "lwc1 %[y0], 0(%[Y1]) \n\t"
720 "lwc1 %[temp1], 0(%[s_m1]) \n\t"
721 "addiu %[noise], %[noise], 1 \n\t"
722 "andi %[noise], %[noise], 0x1ff \n\t"
723 "sll %[temp0], %[noise], 3 \n\t"
724 PTR_ADDU "%[ff_table],%[ff_sbr_noise_table], %[temp0] \n\t"
725 "add.s %[y0], %[y0], %[temp1] \n\t"
726 "mfc1 %[temp3], %[temp1] \n\t"
727 "bne %[temp3], $0, 1f \n\t"
728 "lwc1 %[y1], 4(%[Y1]) \n\t"
729 "lwc1 %[temp2], 0(%[q_filt1]) \n\t"
730 "lwc1 %[temp4], 0(%[ff_table]) \n\t"
731 "lwc1 %[temp5], 4(%[ff_table]) \n\t"
732 "madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
733 "madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
734 "swc1 %[y1], 4(%[Y1]) \n\t"
735 "1: \n\t"
736 "swc1 %[y0], 0(%[Y1]) \n\t"
737
738 : [ff_table]"=&r"(ff_table), [y0]"=&f"(y0), [y1]"=&f"(y1),
739 [temp0]"=&r"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
740 [temp3]"=&r"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5)
741 : [ff_sbr_noise_table]"r"(ff_sbr_noise_table), [noise]"r"(noise),
742 [Y1]"r"(Y1), [s_m1]"r"(s_m1), [q_filt1]"r"(q_filt1)
743 : "memory"
744 );
745 }
746 }
747
sbr_hf_apply_noise_1_mips(float (* Y)[2],const float * s_m,const float * q_filt,int noise,int kx,int m_max)748 static void sbr_hf_apply_noise_1_mips(float (*Y)[2], const float *s_m,
749 const float *q_filt, int noise,
750 int kx, int m_max)
751 {
752 float y0,y1,temp1, temp2, temp4, temp5;
753 int temp0, temp3, m;
754 float phi_sign = 1 - 2 * (kx & 1);
755
756 for (m = 0; m < m_max; m++) {
757
758 float *ff_table;
759 float *Y1=&Y[m][0];
760 const float *s_m1=&s_m[m];
761 const float *q_filt1= &q_filt[m];
762
763 __asm__ volatile(
764 "lwc1 %[y1], 4(%[Y1]) \n\t"
765 "lwc1 %[temp1], 0(%[s_m1]) \n\t"
766 "lw %[temp3], 0(%[s_m1]) \n\t"
767 "addiu %[noise], %[noise], 1 \n\t"
768 "andi %[noise], %[noise], 0x1ff \n\t"
769 "sll %[temp0], %[noise], 3 \n\t"
770 PTR_ADDU "%[ff_table],%[ff_sbr_noise_table],%[temp0] \n\t"
771 "madd.s %[y1], %[y1], %[temp1], %[phi_sign] \n\t"
772 "bne %[temp3], $0, 1f \n\t"
773 "lwc1 %[y0], 0(%[Y1]) \n\t"
774 "lwc1 %[temp2], 0(%[q_filt1]) \n\t"
775 "lwc1 %[temp4], 0(%[ff_table]) \n\t"
776 "lwc1 %[temp5], 4(%[ff_table]) \n\t"
777 "madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
778 "madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
779 "swc1 %[y0], 0(%[Y1]) \n\t"
780 "1: \n\t"
781 "swc1 %[y1], 4(%[Y1]) \n\t"
782
783 : [ff_table] "=&r" (ff_table), [y0] "=&f" (y0), [y1] "=&f" (y1),
784 [temp0] "=&r" (temp0), [temp1] "=&f" (temp1), [temp2] "=&f" (temp2),
785 [temp3] "=&r" (temp3), [temp4] "=&f" (temp4), [temp5] "=&f" (temp5)
786 : [ff_sbr_noise_table] "r" (ff_sbr_noise_table), [noise] "r" (noise),
787 [Y1] "r" (Y1), [s_m1] "r" (s_m1), [q_filt1] "r" (q_filt1),
788 [phi_sign] "f" (phi_sign)
789 : "memory"
790 );
791 phi_sign = -phi_sign;
792 }
793 }
794
sbr_hf_apply_noise_2_mips(float (* Y)[2],const float * s_m,const float * q_filt,int noise,int kx,int m_max)795 static void sbr_hf_apply_noise_2_mips(float (*Y)[2], const float *s_m,
796 const float *q_filt, int noise,
797 int kx, int m_max)
798 {
799 int m, temp0, temp1;
800 float *ff_table;
801 float y0, y1, temp2, temp3, temp4, temp5;
802
803 for (m = 0; m < m_max; m++) {
804
805 float *Y1=&Y[m][0];
806 const float *s_m1=&s_m[m];
807 const float *q_filt1= &q_filt[m];
808
809 __asm__ volatile(
810 "lwc1 %[y0], 0(%[Y1]) \n\t"
811 "lwc1 %[temp3], 0(%[s_m1]) \n\t"
812 "addiu %[noise], %[noise], 1 \n\t"
813 "andi %[noise], %[noise], 0x1ff \n\t"
814 "sll %[temp0], %[noise], 3 \n\t"
815 PTR_ADDU "%[ff_table],%[ff_sbr_noise_table],%[temp0] \n\t"
816 "sub.s %[y0], %[y0], %[temp3] \n\t"
817 "mfc1 %[temp1], %[temp3] \n\t"
818 "bne %[temp1], $0, 1f \n\t"
819 "lwc1 %[y1], 4(%[Y1]) \n\t"
820 "lwc1 %[temp2], 0(%[q_filt1]) \n\t"
821 "lwc1 %[temp4], 0(%[ff_table]) \n\t"
822 "lwc1 %[temp5], 4(%[ff_table]) \n\t"
823 "madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
824 "madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
825 "swc1 %[y1], 4(%[Y1]) \n\t"
826 "1: \n\t"
827 "swc1 %[y0], 0(%[Y1]) \n\t"
828
829 : [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [y0]"=&f"(y0),
830 [y1]"=&f"(y1), [ff_table]"=&r"(ff_table),
831 [temp2]"=&f"(temp2), [temp3]"=&f"(temp3),
832 [temp4]"=&f"(temp4), [temp5]"=&f"(temp5)
833 : [ff_sbr_noise_table]"r"(ff_sbr_noise_table), [noise]"r"(noise),
834 [Y1]"r"(Y1), [s_m1]"r"(s_m1), [q_filt1]"r"(q_filt1)
835 : "memory"
836 );
837 }
838 }
839
sbr_hf_apply_noise_3_mips(float (* Y)[2],const float * s_m,const float * q_filt,int noise,int kx,int m_max)840 static void sbr_hf_apply_noise_3_mips(float (*Y)[2], const float *s_m,
841 const float *q_filt, int noise,
842 int kx, int m_max)
843 {
844 float phi_sign = 1 - 2 * (kx & 1);
845 int m;
846
847 for (m = 0; m < m_max; m++) {
848
849 float *Y1=&Y[m][0];
850 float *ff_table;
851 float y0,y1, temp1, temp2, temp4, temp5;
852 int temp0, temp3;
853 const float *s_m1=&s_m[m];
854 const float *q_filt1= &q_filt[m];
855
856 __asm__ volatile(
857 "lwc1 %[y1], 4(%[Y1]) \n\t"
858 "lwc1 %[temp1], 0(%[s_m1]) \n\t"
859 "addiu %[noise], %[noise], 1 \n\t"
860 "andi %[noise], %[noise], 0x1ff \n\t"
861 "sll %[temp0], %[noise], 3 \n\t"
862 PTR_ADDU "%[ff_table],%[ff_sbr_noise_table], %[temp0] \n\t"
863 "nmsub.s %[y1], %[y1], %[temp1], %[phi_sign] \n\t"
864 "mfc1 %[temp3], %[temp1] \n\t"
865 "bne %[temp3], $0, 1f \n\t"
866 "lwc1 %[y0], 0(%[Y1]) \n\t"
867 "lwc1 %[temp2], 0(%[q_filt1]) \n\t"
868 "lwc1 %[temp4], 0(%[ff_table]) \n\t"
869 "lwc1 %[temp5], 4(%[ff_table]) \n\t"
870 "madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
871 "madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
872 "swc1 %[y0], 0(%[Y1]) \n\t"
873 "1: \n\t"
874 "swc1 %[y1], 4(%[Y1]) \n\t"
875
876 : [ff_table]"=&r"(ff_table), [y0]"=&f"(y0), [y1]"=&f"(y1),
877 [temp0]"=&r"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
878 [temp3]"=&r"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5)
879 : [ff_sbr_noise_table]"r"(ff_sbr_noise_table), [noise]"r"(noise),
880 [Y1]"r"(Y1), [s_m1]"r"(s_m1), [q_filt1]"r"(q_filt1),
881 [phi_sign]"f"(phi_sign)
882 : "memory"
883 );
884 phi_sign = -phi_sign;
885 }
886 }
887 #endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */
888 #endif /* HAVE_MIPSFPU */
889 #endif /* HAVE_INLINE_ASM */
890
ff_sbrdsp_init_mips(SBRDSPContext * s)891 void ff_sbrdsp_init_mips(SBRDSPContext *s)
892 {
893 #if HAVE_INLINE_ASM
894 #if HAVE_MIPSFPU
895 s->qmf_pre_shuffle = sbr_qmf_pre_shuffle_mips;
896 s->qmf_post_shuffle = sbr_qmf_post_shuffle_mips;
897 #if !HAVE_MIPS32R6 && !HAVE_MIPS64R6
898 s->sum64x5 = sbr_sum64x5_mips;
899 s->sum_square = sbr_sum_square_mips;
900 s->qmf_deint_bfly = sbr_qmf_deint_bfly_mips;
901 s->autocorrelate = sbr_autocorrelate_mips;
902 s->hf_gen = sbr_hf_gen_mips;
903 s->hf_g_filt = sbr_hf_g_filt_mips;
904
905 s->hf_apply_noise[0] = sbr_hf_apply_noise_0_mips;
906 s->hf_apply_noise[1] = sbr_hf_apply_noise_1_mips;
907 s->hf_apply_noise[2] = sbr_hf_apply_noise_2_mips;
908 s->hf_apply_noise[3] = sbr_hf_apply_noise_3_mips;
909 #endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */
910 #endif /* HAVE_MIPSFPU */
911 #endif /* HAVE_INLINE_ASM */
912 }
913