1 /*
2 * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved.
3 *
4 * Use of this source code is governed by a BSD-style license
5 * that can be found in the LICENSE file in the root of the source
6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree.
9 */
10
11 #include "common_audio/third_party/ooura/fft_size_128/ooura_fft.h"
12 #include "common_audio/third_party/ooura/fft_size_128/ooura_fft_tables_common.h"
13
14 namespace webrtc {
15
16 #if defined(MIPS_FPU_LE)
bitrv2_128_mips(float * a)17 void bitrv2_128_mips(float* a) {
18 // n is 128
19 float xr, xi, yr, yi;
20
21 xr = a[8];
22 xi = a[9];
23 yr = a[16];
24 yi = a[17];
25 a[8] = yr;
26 a[9] = yi;
27 a[16] = xr;
28 a[17] = xi;
29
30 xr = a[64];
31 xi = a[65];
32 yr = a[2];
33 yi = a[3];
34 a[64] = yr;
35 a[65] = yi;
36 a[2] = xr;
37 a[3] = xi;
38
39 xr = a[72];
40 xi = a[73];
41 yr = a[18];
42 yi = a[19];
43 a[72] = yr;
44 a[73] = yi;
45 a[18] = xr;
46 a[19] = xi;
47
48 xr = a[80];
49 xi = a[81];
50 yr = a[10];
51 yi = a[11];
52 a[80] = yr;
53 a[81] = yi;
54 a[10] = xr;
55 a[11] = xi;
56
57 xr = a[88];
58 xi = a[89];
59 yr = a[26];
60 yi = a[27];
61 a[88] = yr;
62 a[89] = yi;
63 a[26] = xr;
64 a[27] = xi;
65
66 xr = a[74];
67 xi = a[75];
68 yr = a[82];
69 yi = a[83];
70 a[74] = yr;
71 a[75] = yi;
72 a[82] = xr;
73 a[83] = xi;
74
75 xr = a[32];
76 xi = a[33];
77 yr = a[4];
78 yi = a[5];
79 a[32] = yr;
80 a[33] = yi;
81 a[4] = xr;
82 a[5] = xi;
83
84 xr = a[40];
85 xi = a[41];
86 yr = a[20];
87 yi = a[21];
88 a[40] = yr;
89 a[41] = yi;
90 a[20] = xr;
91 a[21] = xi;
92
93 xr = a[48];
94 xi = a[49];
95 yr = a[12];
96 yi = a[13];
97 a[48] = yr;
98 a[49] = yi;
99 a[12] = xr;
100 a[13] = xi;
101
102 xr = a[56];
103 xi = a[57];
104 yr = a[28];
105 yi = a[29];
106 a[56] = yr;
107 a[57] = yi;
108 a[28] = xr;
109 a[29] = xi;
110
111 xr = a[34];
112 xi = a[35];
113 yr = a[68];
114 yi = a[69];
115 a[34] = yr;
116 a[35] = yi;
117 a[68] = xr;
118 a[69] = xi;
119
120 xr = a[42];
121 xi = a[43];
122 yr = a[84];
123 yi = a[85];
124 a[42] = yr;
125 a[43] = yi;
126 a[84] = xr;
127 a[85] = xi;
128
129 xr = a[50];
130 xi = a[51];
131 yr = a[76];
132 yi = a[77];
133 a[50] = yr;
134 a[51] = yi;
135 a[76] = xr;
136 a[77] = xi;
137
138 xr = a[58];
139 xi = a[59];
140 yr = a[92];
141 yi = a[93];
142 a[58] = yr;
143 a[59] = yi;
144 a[92] = xr;
145 a[93] = xi;
146
147 xr = a[44];
148 xi = a[45];
149 yr = a[52];
150 yi = a[53];
151 a[44] = yr;
152 a[45] = yi;
153 a[52] = xr;
154 a[53] = xi;
155
156 xr = a[96];
157 xi = a[97];
158 yr = a[6];
159 yi = a[7];
160 a[96] = yr;
161 a[97] = yi;
162 a[6] = xr;
163 a[7] = xi;
164
165 xr = a[104];
166 xi = a[105];
167 yr = a[22];
168 yi = a[23];
169 a[104] = yr;
170 a[105] = yi;
171 a[22] = xr;
172 a[23] = xi;
173
174 xr = a[112];
175 xi = a[113];
176 yr = a[14];
177 yi = a[15];
178 a[112] = yr;
179 a[113] = yi;
180 a[14] = xr;
181 a[15] = xi;
182
183 xr = a[120];
184 xi = a[121];
185 yr = a[30];
186 yi = a[31];
187 a[120] = yr;
188 a[121] = yi;
189 a[30] = xr;
190 a[31] = xi;
191
192 xr = a[98];
193 xi = a[99];
194 yr = a[70];
195 yi = a[71];
196 a[98] = yr;
197 a[99] = yi;
198 a[70] = xr;
199 a[71] = xi;
200
201 xr = a[106];
202 xi = a[107];
203 yr = a[86];
204 yi = a[87];
205 a[106] = yr;
206 a[107] = yi;
207 a[86] = xr;
208 a[87] = xi;
209
210 xr = a[114];
211 xi = a[115];
212 yr = a[78];
213 yi = a[79];
214 a[114] = yr;
215 a[115] = yi;
216 a[78] = xr;
217 a[79] = xi;
218
219 xr = a[122];
220 xi = a[123];
221 yr = a[94];
222 yi = a[95];
223 a[122] = yr;
224 a[123] = yi;
225 a[94] = xr;
226 a[95] = xi;
227
228 xr = a[100];
229 xi = a[101];
230 yr = a[38];
231 yi = a[39];
232 a[100] = yr;
233 a[101] = yi;
234 a[38] = xr;
235 a[39] = xi;
236
237 xr = a[108];
238 xi = a[109];
239 yr = a[54];
240 yi = a[55];
241 a[108] = yr;
242 a[109] = yi;
243 a[54] = xr;
244 a[55] = xi;
245
246 xr = a[116];
247 xi = a[117];
248 yr = a[46];
249 yi = a[47];
250 a[116] = yr;
251 a[117] = yi;
252 a[46] = xr;
253 a[47] = xi;
254
255 xr = a[124];
256 xi = a[125];
257 yr = a[62];
258 yi = a[63];
259 a[124] = yr;
260 a[125] = yi;
261 a[62] = xr;
262 a[63] = xi;
263
264 xr = a[110];
265 xi = a[111];
266 yr = a[118];
267 yi = a[119];
268 a[110] = yr;
269 a[111] = yi;
270 a[118] = xr;
271 a[119] = xi;
272 }
273
cft1st_128_mips(float * a)274 void cft1st_128_mips(float* a) {
275 float f0, f1, f2, f3, f4, f5, f6, f7, f8, f9, f10, f11, f12, f13, f14;
276 int a_ptr, p1_rdft, p2_rdft, count;
277 const float* first = rdft_wk3ri_first;
278 const float* second = rdft_wk3ri_second;
279
280 __asm __volatile(
281 ".set push \n\t"
282 ".set noreorder \n\t"
283 // first 8
284 "lwc1 %[f0], 0(%[a]) \n\t"
285 "lwc1 %[f1], 4(%[a]) \n\t"
286 "lwc1 %[f2], 8(%[a]) \n\t"
287 "lwc1 %[f3], 12(%[a]) \n\t"
288 "lwc1 %[f4], 16(%[a]) \n\t"
289 "lwc1 %[f5], 20(%[a]) \n\t"
290 "lwc1 %[f6], 24(%[a]) \n\t"
291 "lwc1 %[f7], 28(%[a]) \n\t"
292 "add.s %[f8], %[f0], %[f2] \n\t"
293 "sub.s %[f0], %[f0], %[f2] \n\t"
294 "add.s %[f2], %[f4], %[f6] \n\t"
295 "sub.s %[f4], %[f4], %[f6] \n\t"
296 "add.s %[f6], %[f1], %[f3] \n\t"
297 "sub.s %[f1], %[f1], %[f3] \n\t"
298 "add.s %[f3], %[f5], %[f7] \n\t"
299 "sub.s %[f5], %[f5], %[f7] \n\t"
300 "add.s %[f7], %[f8], %[f2] \n\t"
301 "sub.s %[f8], %[f8], %[f2] \n\t"
302 "sub.s %[f2], %[f1], %[f4] \n\t"
303 "add.s %[f1], %[f1], %[f4] \n\t"
304 "add.s %[f4], %[f6], %[f3] \n\t"
305 "sub.s %[f6], %[f6], %[f3] \n\t"
306 "sub.s %[f3], %[f0], %[f5] \n\t"
307 "add.s %[f0], %[f0], %[f5] \n\t"
308 "swc1 %[f7], 0(%[a]) \n\t"
309 "swc1 %[f8], 16(%[a]) \n\t"
310 "swc1 %[f2], 28(%[a]) \n\t"
311 "swc1 %[f1], 12(%[a]) \n\t"
312 "swc1 %[f4], 4(%[a]) \n\t"
313 "swc1 %[f6], 20(%[a]) \n\t"
314 "swc1 %[f3], 8(%[a]) \n\t"
315 "swc1 %[f0], 24(%[a]) \n\t"
316 // second 8
317 "lwc1 %[f0], 32(%[a]) \n\t"
318 "lwc1 %[f1], 36(%[a]) \n\t"
319 "lwc1 %[f2], 40(%[a]) \n\t"
320 "lwc1 %[f3], 44(%[a]) \n\t"
321 "lwc1 %[f4], 48(%[a]) \n\t"
322 "lwc1 %[f5], 52(%[a]) \n\t"
323 "lwc1 %[f6], 56(%[a]) \n\t"
324 "lwc1 %[f7], 60(%[a]) \n\t"
325 "add.s %[f8], %[f4], %[f6] \n\t"
326 "sub.s %[f4], %[f4], %[f6] \n\t"
327 "add.s %[f6], %[f1], %[f3] \n\t"
328 "sub.s %[f1], %[f1], %[f3] \n\t"
329 "add.s %[f3], %[f0], %[f2] \n\t"
330 "sub.s %[f0], %[f0], %[f2] \n\t"
331 "add.s %[f2], %[f5], %[f7] \n\t"
332 "sub.s %[f5], %[f5], %[f7] \n\t"
333 "add.s %[f7], %[f4], %[f1] \n\t"
334 "sub.s %[f4], %[f4], %[f1] \n\t"
335 "add.s %[f1], %[f3], %[f8] \n\t"
336 "sub.s %[f3], %[f3], %[f8] \n\t"
337 "sub.s %[f8], %[f0], %[f5] \n\t"
338 "add.s %[f0], %[f0], %[f5] \n\t"
339 "add.s %[f5], %[f6], %[f2] \n\t"
340 "sub.s %[f6], %[f2], %[f6] \n\t"
341 "lwc1 %[f9], 8(%[rdft_w]) \n\t"
342 "sub.s %[f2], %[f8], %[f7] \n\t"
343 "add.s %[f8], %[f8], %[f7] \n\t"
344 "sub.s %[f7], %[f4], %[f0] \n\t"
345 "add.s %[f4], %[f4], %[f0] \n\t"
346 // prepare for loop
347 "addiu %[a_ptr], %[a], 64 \n\t"
348 "addiu %[p1_rdft], %[rdft_w], 8 \n\t"
349 "addiu %[p2_rdft], %[rdft_w], 16 \n\t"
350 "addiu %[count], $zero, 7 \n\t"
351 // finish second 8
352 "mul.s %[f2], %[f9], %[f2] \n\t"
353 "mul.s %[f8], %[f9], %[f8] \n\t"
354 "mul.s %[f7], %[f9], %[f7] \n\t"
355 "mul.s %[f4], %[f9], %[f4] \n\t"
356 "swc1 %[f1], 32(%[a]) \n\t"
357 "swc1 %[f3], 52(%[a]) \n\t"
358 "swc1 %[f5], 36(%[a]) \n\t"
359 "swc1 %[f6], 48(%[a]) \n\t"
360 "swc1 %[f2], 40(%[a]) \n\t"
361 "swc1 %[f8], 44(%[a]) \n\t"
362 "swc1 %[f7], 56(%[a]) \n\t"
363 "swc1 %[f4], 60(%[a]) \n\t"
364 // loop
365 "1: \n\t"
366 "lwc1 %[f0], 0(%[a_ptr]) \n\t"
367 "lwc1 %[f1], 4(%[a_ptr]) \n\t"
368 "lwc1 %[f2], 8(%[a_ptr]) \n\t"
369 "lwc1 %[f3], 12(%[a_ptr]) \n\t"
370 "lwc1 %[f4], 16(%[a_ptr]) \n\t"
371 "lwc1 %[f5], 20(%[a_ptr]) \n\t"
372 "lwc1 %[f6], 24(%[a_ptr]) \n\t"
373 "lwc1 %[f7], 28(%[a_ptr]) \n\t"
374 "add.s %[f8], %[f0], %[f2] \n\t"
375 "sub.s %[f0], %[f0], %[f2] \n\t"
376 "add.s %[f2], %[f4], %[f6] \n\t"
377 "sub.s %[f4], %[f4], %[f6] \n\t"
378 "add.s %[f6], %[f1], %[f3] \n\t"
379 "sub.s %[f1], %[f1], %[f3] \n\t"
380 "add.s %[f3], %[f5], %[f7] \n\t"
381 "sub.s %[f5], %[f5], %[f7] \n\t"
382 "lwc1 %[f10], 4(%[p1_rdft]) \n\t"
383 "lwc1 %[f11], 0(%[p2_rdft]) \n\t"
384 "lwc1 %[f12], 4(%[p2_rdft]) \n\t"
385 "lwc1 %[f13], 8(%[first]) \n\t"
386 "lwc1 %[f14], 12(%[first]) \n\t"
387 "add.s %[f7], %[f8], %[f2] \n\t"
388 "sub.s %[f8], %[f8], %[f2] \n\t"
389 "add.s %[f2], %[f6], %[f3] \n\t"
390 "sub.s %[f6], %[f6], %[f3] \n\t"
391 "add.s %[f3], %[f0], %[f5] \n\t"
392 "sub.s %[f0], %[f0], %[f5] \n\t"
393 "add.s %[f5], %[f1], %[f4] \n\t"
394 "sub.s %[f1], %[f1], %[f4] \n\t"
395 "swc1 %[f7], 0(%[a_ptr]) \n\t"
396 "swc1 %[f2], 4(%[a_ptr]) \n\t"
397 "mul.s %[f4], %[f9], %[f8] \n\t"
398 #if defined(MIPS32_R2_LE)
399 "mul.s %[f8], %[f10], %[f8] \n\t"
400 "mul.s %[f7], %[f11], %[f0] \n\t"
401 "mul.s %[f0], %[f12], %[f0] \n\t"
402 "mul.s %[f2], %[f13], %[f3] \n\t"
403 "mul.s %[f3], %[f14], %[f3] \n\t"
404 "nmsub.s %[f4], %[f4], %[f10], %[f6] \n\t"
405 "madd.s %[f8], %[f8], %[f9], %[f6] \n\t"
406 "nmsub.s %[f7], %[f7], %[f12], %[f5] \n\t"
407 "madd.s %[f0], %[f0], %[f11], %[f5] \n\t"
408 "nmsub.s %[f2], %[f2], %[f14], %[f1] \n\t"
409 "madd.s %[f3], %[f3], %[f13], %[f1] \n\t"
410 #else
411 "mul.s %[f7], %[f10], %[f6] \n\t"
412 "mul.s %[f6], %[f9], %[f6] \n\t"
413 "mul.s %[f8], %[f10], %[f8] \n\t"
414 "mul.s %[f2], %[f11], %[f0] \n\t"
415 "mul.s %[f11], %[f11], %[f5] \n\t"
416 "mul.s %[f5], %[f12], %[f5] \n\t"
417 "mul.s %[f0], %[f12], %[f0] \n\t"
418 "mul.s %[f12], %[f13], %[f3] \n\t"
419 "mul.s %[f13], %[f13], %[f1] \n\t"
420 "mul.s %[f1], %[f14], %[f1] \n\t"
421 "mul.s %[f3], %[f14], %[f3] \n\t"
422 "sub.s %[f4], %[f4], %[f7] \n\t"
423 "add.s %[f8], %[f6], %[f8] \n\t"
424 "sub.s %[f7], %[f2], %[f5] \n\t"
425 "add.s %[f0], %[f11], %[f0] \n\t"
426 "sub.s %[f2], %[f12], %[f1] \n\t"
427 "add.s %[f3], %[f13], %[f3] \n\t"
428 #endif
429 "swc1 %[f4], 16(%[a_ptr]) \n\t"
430 "swc1 %[f8], 20(%[a_ptr]) \n\t"
431 "swc1 %[f7], 8(%[a_ptr]) \n\t"
432 "swc1 %[f0], 12(%[a_ptr]) \n\t"
433 "swc1 %[f2], 24(%[a_ptr]) \n\t"
434 "swc1 %[f3], 28(%[a_ptr]) \n\t"
435 "lwc1 %[f0], 32(%[a_ptr]) \n\t"
436 "lwc1 %[f1], 36(%[a_ptr]) \n\t"
437 "lwc1 %[f2], 40(%[a_ptr]) \n\t"
438 "lwc1 %[f3], 44(%[a_ptr]) \n\t"
439 "lwc1 %[f4], 48(%[a_ptr]) \n\t"
440 "lwc1 %[f5], 52(%[a_ptr]) \n\t"
441 "lwc1 %[f6], 56(%[a_ptr]) \n\t"
442 "lwc1 %[f7], 60(%[a_ptr]) \n\t"
443 "add.s %[f8], %[f0], %[f2] \n\t"
444 "sub.s %[f0], %[f0], %[f2] \n\t"
445 "add.s %[f2], %[f4], %[f6] \n\t"
446 "sub.s %[f4], %[f4], %[f6] \n\t"
447 "add.s %[f6], %[f1], %[f3] \n\t"
448 "sub.s %[f1], %[f1], %[f3] \n\t"
449 "add.s %[f3], %[f5], %[f7] \n\t"
450 "sub.s %[f5], %[f5], %[f7] \n\t"
451 "lwc1 %[f11], 8(%[p2_rdft]) \n\t"
452 "lwc1 %[f12], 12(%[p2_rdft]) \n\t"
453 "lwc1 %[f13], 8(%[second]) \n\t"
454 "lwc1 %[f14], 12(%[second]) \n\t"
455 "add.s %[f7], %[f8], %[f2] \n\t"
456 "sub.s %[f8], %[f2], %[f8] \n\t"
457 "add.s %[f2], %[f6], %[f3] \n\t"
458 "sub.s %[f6], %[f3], %[f6] \n\t"
459 "add.s %[f3], %[f0], %[f5] \n\t"
460 "sub.s %[f0], %[f0], %[f5] \n\t"
461 "add.s %[f5], %[f1], %[f4] \n\t"
462 "sub.s %[f1], %[f1], %[f4] \n\t"
463 "swc1 %[f7], 32(%[a_ptr]) \n\t"
464 "swc1 %[f2], 36(%[a_ptr]) \n\t"
465 "mul.s %[f4], %[f10], %[f8] \n\t"
466 #if defined(MIPS32_R2_LE)
467 "mul.s %[f10], %[f10], %[f6] \n\t"
468 "mul.s %[f7], %[f11], %[f0] \n\t"
469 "mul.s %[f11], %[f11], %[f5] \n\t"
470 "mul.s %[f2], %[f13], %[f3] \n\t"
471 "mul.s %[f13], %[f13], %[f1] \n\t"
472 "madd.s %[f4], %[f4], %[f9], %[f6] \n\t"
473 "nmsub.s %[f10], %[f10], %[f9], %[f8] \n\t"
474 "nmsub.s %[f7], %[f7], %[f12], %[f5] \n\t"
475 "madd.s %[f11], %[f11], %[f12], %[f0] \n\t"
476 "nmsub.s %[f2], %[f2], %[f14], %[f1] \n\t"
477 "madd.s %[f13], %[f13], %[f14], %[f3] \n\t"
478 #else
479 "mul.s %[f2], %[f9], %[f6] \n\t"
480 "mul.s %[f10], %[f10], %[f6] \n\t"
481 "mul.s %[f9], %[f9], %[f8] \n\t"
482 "mul.s %[f7], %[f11], %[f0] \n\t"
483 "mul.s %[f8], %[f12], %[f5] \n\t"
484 "mul.s %[f11], %[f11], %[f5] \n\t"
485 "mul.s %[f12], %[f12], %[f0] \n\t"
486 "mul.s %[f5], %[f13], %[f3] \n\t"
487 "mul.s %[f0], %[f14], %[f1] \n\t"
488 "mul.s %[f13], %[f13], %[f1] \n\t"
489 "mul.s %[f14], %[f14], %[f3] \n\t"
490 "add.s %[f4], %[f4], %[f2] \n\t"
491 "sub.s %[f10], %[f10], %[f9] \n\t"
492 "sub.s %[f7], %[f7], %[f8] \n\t"
493 "add.s %[f11], %[f11], %[f12] \n\t"
494 "sub.s %[f2], %[f5], %[f0] \n\t"
495 "add.s %[f13], %[f13], %[f14] \n\t"
496 #endif
497 "swc1 %[f4], 48(%[a_ptr]) \n\t"
498 "swc1 %[f10], 52(%[a_ptr]) \n\t"
499 "swc1 %[f7], 40(%[a_ptr]) \n\t"
500 "swc1 %[f11], 44(%[a_ptr]) \n\t"
501 "swc1 %[f2], 56(%[a_ptr]) \n\t"
502 "swc1 %[f13], 60(%[a_ptr]) \n\t"
503 "addiu %[count], %[count], -1 \n\t"
504 "lwc1 %[f9], 8(%[p1_rdft]) \n\t"
505 "addiu %[a_ptr], %[a_ptr], 64 \n\t"
506 "addiu %[p1_rdft], %[p1_rdft], 8 \n\t"
507 "addiu %[p2_rdft], %[p2_rdft], 16 \n\t"
508 "addiu %[first], %[first], 8 \n\t"
509 "bgtz %[count], 1b \n\t"
510 " addiu %[second], %[second], 8 \n\t"
511 ".set pop \n\t"
512 : [f0] "=&f"(f0), [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3),
513 [f4] "=&f"(f4), [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7),
514 [f8] "=&f"(f8), [f9] "=&f"(f9), [f10] "=&f"(f10), [f11] "=&f"(f11),
515 [f12] "=&f"(f12), [f13] "=&f"(f13), [f14] "=&f"(f14),
516 [a_ptr] "=&r"(a_ptr), [p1_rdft] "=&r"(p1_rdft), [first] "+r"(first),
517 [p2_rdft] "=&r"(p2_rdft), [count] "=&r"(count), [second] "+r"(second)
518 : [a] "r"(a), [rdft_w] "r"(rdft_w)
519 : "memory");
520 }
521
cftmdl_128_mips(float * a)522 void cftmdl_128_mips(float* a) {
523 float f0, f1, f2, f3, f4, f5, f6, f7, f8, f9, f10, f11, f12, f13, f14;
524 int tmp_a, count;
525 __asm __volatile(
526 ".set push \n\t"
527 ".set noreorder \n\t"
528 "addiu %[tmp_a], %[a], 0 \n\t"
529 "addiu %[count], $zero, 4 \n\t"
530 "1: \n\t"
531 "addiu %[count], %[count], -1 \n\t"
532 "lwc1 %[f0], 0(%[tmp_a]) \n\t"
533 "lwc1 %[f2], 32(%[tmp_a]) \n\t"
534 "lwc1 %[f4], 64(%[tmp_a]) \n\t"
535 "lwc1 %[f6], 96(%[tmp_a]) \n\t"
536 "lwc1 %[f1], 4(%[tmp_a]) \n\t"
537 "lwc1 %[f3], 36(%[tmp_a]) \n\t"
538 "lwc1 %[f5], 68(%[tmp_a]) \n\t"
539 "lwc1 %[f7], 100(%[tmp_a]) \n\t"
540 "add.s %[f8], %[f0], %[f2] \n\t"
541 "sub.s %[f0], %[f0], %[f2] \n\t"
542 "add.s %[f2], %[f4], %[f6] \n\t"
543 "sub.s %[f4], %[f4], %[f6] \n\t"
544 "add.s %[f6], %[f1], %[f3] \n\t"
545 "sub.s %[f1], %[f1], %[f3] \n\t"
546 "add.s %[f3], %[f5], %[f7] \n\t"
547 "sub.s %[f5], %[f5], %[f7] \n\t"
548 "add.s %[f7], %[f8], %[f2] \n\t"
549 "sub.s %[f8], %[f8], %[f2] \n\t"
550 "add.s %[f2], %[f1], %[f4] \n\t"
551 "sub.s %[f1], %[f1], %[f4] \n\t"
552 "add.s %[f4], %[f6], %[f3] \n\t"
553 "sub.s %[f6], %[f6], %[f3] \n\t"
554 "sub.s %[f3], %[f0], %[f5] \n\t"
555 "add.s %[f0], %[f0], %[f5] \n\t"
556 "swc1 %[f7], 0(%[tmp_a]) \n\t"
557 "swc1 %[f8], 64(%[tmp_a]) \n\t"
558 "swc1 %[f2], 36(%[tmp_a]) \n\t"
559 "swc1 %[f1], 100(%[tmp_a]) \n\t"
560 "swc1 %[f4], 4(%[tmp_a]) \n\t"
561 "swc1 %[f6], 68(%[tmp_a]) \n\t"
562 "swc1 %[f3], 32(%[tmp_a]) \n\t"
563 "swc1 %[f0], 96(%[tmp_a]) \n\t"
564 "bgtz %[count], 1b \n\t"
565 " addiu %[tmp_a], %[tmp_a], 8 \n\t"
566 ".set pop \n\t"
567 : [f0] "=&f"(f0), [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3),
568 [f4] "=&f"(f4), [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7),
569 [f8] "=&f"(f8), [tmp_a] "=&r"(tmp_a), [count] "=&r"(count)
570 : [a] "r"(a)
571 : "memory");
572 f9 = rdft_w[2];
573 __asm __volatile(
574 ".set push \n\t"
575 ".set noreorder \n\t"
576 "addiu %[tmp_a], %[a], 128 \n\t"
577 "addiu %[count], $zero, 4 \n\t"
578 "1: \n\t"
579 "addiu %[count], %[count], -1 \n\t"
580 "lwc1 %[f0], 0(%[tmp_a]) \n\t"
581 "lwc1 %[f2], 32(%[tmp_a]) \n\t"
582 "lwc1 %[f5], 68(%[tmp_a]) \n\t"
583 "lwc1 %[f7], 100(%[tmp_a]) \n\t"
584 "lwc1 %[f1], 4(%[tmp_a]) \n\t"
585 "lwc1 %[f3], 36(%[tmp_a]) \n\t"
586 "lwc1 %[f4], 64(%[tmp_a]) \n\t"
587 "lwc1 %[f6], 96(%[tmp_a]) \n\t"
588 "sub.s %[f8], %[f0], %[f2] \n\t"
589 "add.s %[f0], %[f0], %[f2] \n\t"
590 "sub.s %[f2], %[f5], %[f7] \n\t"
591 "add.s %[f5], %[f5], %[f7] \n\t"
592 "sub.s %[f7], %[f1], %[f3] \n\t"
593 "add.s %[f1], %[f1], %[f3] \n\t"
594 "sub.s %[f3], %[f4], %[f6] \n\t"
595 "add.s %[f4], %[f4], %[f6] \n\t"
596 "sub.s %[f6], %[f8], %[f2] \n\t"
597 "add.s %[f8], %[f8], %[f2] \n\t"
598 "add.s %[f2], %[f5], %[f1] \n\t"
599 "sub.s %[f5], %[f5], %[f1] \n\t"
600 "add.s %[f1], %[f3], %[f7] \n\t"
601 "sub.s %[f3], %[f3], %[f7] \n\t"
602 "add.s %[f7], %[f0], %[f4] \n\t"
603 "sub.s %[f0], %[f0], %[f4] \n\t"
604 "sub.s %[f4], %[f6], %[f1] \n\t"
605 "add.s %[f6], %[f6], %[f1] \n\t"
606 "sub.s %[f1], %[f3], %[f8] \n\t"
607 "add.s %[f3], %[f3], %[f8] \n\t"
608 "mul.s %[f4], %[f4], %[f9] \n\t"
609 "mul.s %[f6], %[f6], %[f9] \n\t"
610 "mul.s %[f1], %[f1], %[f9] \n\t"
611 "mul.s %[f3], %[f3], %[f9] \n\t"
612 "swc1 %[f7], 0(%[tmp_a]) \n\t"
613 "swc1 %[f2], 4(%[tmp_a]) \n\t"
614 "swc1 %[f5], 64(%[tmp_a]) \n\t"
615 "swc1 %[f0], 68(%[tmp_a]) \n\t"
616 "swc1 %[f4], 32(%[tmp_a]) \n\t"
617 "swc1 %[f6], 36(%[tmp_a]) \n\t"
618 "swc1 %[f1], 96(%[tmp_a]) \n\t"
619 "swc1 %[f3], 100(%[tmp_a]) \n\t"
620 "bgtz %[count], 1b \n\t"
621 " addiu %[tmp_a], %[tmp_a], 8 \n\t"
622 ".set pop \n\t"
623 : [f0] "=&f"(f0), [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3),
624 [f4] "=&f"(f4), [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7),
625 [f8] "=&f"(f8), [tmp_a] "=&r"(tmp_a), [count] "=&r"(count)
626 : [a] "r"(a), [f9] "f"(f9)
627 : "memory");
628 f10 = rdft_w[3];
629 f11 = rdft_w[4];
630 f12 = rdft_w[5];
631 f13 = rdft_wk3ri_first[2];
632 f14 = rdft_wk3ri_first[3];
633
634 __asm __volatile(
635 ".set push \n\t"
636 ".set noreorder \n\t"
637 "addiu %[tmp_a], %[a], 256 \n\t"
638 "addiu %[count], $zero, 4 \n\t"
639 "1: \n\t"
640 "addiu %[count], %[count], -1 \n\t"
641 "lwc1 %[f0], 0(%[tmp_a]) \n\t"
642 "lwc1 %[f2], 32(%[tmp_a]) \n\t"
643 "lwc1 %[f4], 64(%[tmp_a]) \n\t"
644 "lwc1 %[f6], 96(%[tmp_a]) \n\t"
645 "lwc1 %[f1], 4(%[tmp_a]) \n\t"
646 "lwc1 %[f3], 36(%[tmp_a]) \n\t"
647 "lwc1 %[f5], 68(%[tmp_a]) \n\t"
648 "lwc1 %[f7], 100(%[tmp_a]) \n\t"
649 "add.s %[f8], %[f0], %[f2] \n\t"
650 "sub.s %[f0], %[f0], %[f2] \n\t"
651 "add.s %[f2], %[f4], %[f6] \n\t"
652 "sub.s %[f4], %[f4], %[f6] \n\t"
653 "add.s %[f6], %[f1], %[f3] \n\t"
654 "sub.s %[f1], %[f1], %[f3] \n\t"
655 "add.s %[f3], %[f5], %[f7] \n\t"
656 "sub.s %[f5], %[f5], %[f7] \n\t"
657 "sub.s %[f7], %[f8], %[f2] \n\t"
658 "add.s %[f8], %[f8], %[f2] \n\t"
659 "add.s %[f2], %[f1], %[f4] \n\t"
660 "sub.s %[f1], %[f1], %[f4] \n\t"
661 "sub.s %[f4], %[f6], %[f3] \n\t"
662 "add.s %[f6], %[f6], %[f3] \n\t"
663 "sub.s %[f3], %[f0], %[f5] \n\t"
664 "add.s %[f0], %[f0], %[f5] \n\t"
665 "swc1 %[f8], 0(%[tmp_a]) \n\t"
666 "swc1 %[f6], 4(%[tmp_a]) \n\t"
667 "mul.s %[f5], %[f9], %[f7] \n\t"
668 #if defined(MIPS32_R2_LE)
669 "mul.s %[f7], %[f10], %[f7] \n\t"
670 "mul.s %[f8], %[f11], %[f3] \n\t"
671 "mul.s %[f3], %[f12], %[f3] \n\t"
672 "mul.s %[f6], %[f13], %[f0] \n\t"
673 "mul.s %[f0], %[f14], %[f0] \n\t"
674 "nmsub.s %[f5], %[f5], %[f10], %[f4] \n\t"
675 "madd.s %[f7], %[f7], %[f9], %[f4] \n\t"
676 "nmsub.s %[f8], %[f8], %[f12], %[f2] \n\t"
677 "madd.s %[f3], %[f3], %[f11], %[f2] \n\t"
678 "nmsub.s %[f6], %[f6], %[f14], %[f1] \n\t"
679 "madd.s %[f0], %[f0], %[f13], %[f1] \n\t"
680 "swc1 %[f5], 64(%[tmp_a]) \n\t"
681 "swc1 %[f7], 68(%[tmp_a]) \n\t"
682 #else
683 "mul.s %[f8], %[f10], %[f4] \n\t"
684 "mul.s %[f4], %[f9], %[f4] \n\t"
685 "mul.s %[f7], %[f10], %[f7] \n\t"
686 "mul.s %[f6], %[f11], %[f3] \n\t"
687 "mul.s %[f3], %[f12], %[f3] \n\t"
688 "sub.s %[f5], %[f5], %[f8] \n\t"
689 "mul.s %[f8], %[f12], %[f2] \n\t"
690 "mul.s %[f2], %[f11], %[f2] \n\t"
691 "add.s %[f7], %[f4], %[f7] \n\t"
692 "mul.s %[f4], %[f13], %[f0] \n\t"
693 "mul.s %[f0], %[f14], %[f0] \n\t"
694 "sub.s %[f8], %[f6], %[f8] \n\t"
695 "mul.s %[f6], %[f14], %[f1] \n\t"
696 "mul.s %[f1], %[f13], %[f1] \n\t"
697 "add.s %[f3], %[f2], %[f3] \n\t"
698 "swc1 %[f5], 64(%[tmp_a]) \n\t"
699 "swc1 %[f7], 68(%[tmp_a]) \n\t"
700 "sub.s %[f6], %[f4], %[f6] \n\t"
701 "add.s %[f0], %[f1], %[f0] \n\t"
702 #endif
703 "swc1 %[f8], 32(%[tmp_a]) \n\t"
704 "swc1 %[f3], 36(%[tmp_a]) \n\t"
705 "swc1 %[f6], 96(%[tmp_a]) \n\t"
706 "swc1 %[f0], 100(%[tmp_a]) \n\t"
707 "bgtz %[count], 1b \n\t"
708 " addiu %[tmp_a], %[tmp_a], 8 \n\t"
709 ".set pop \n\t"
710 : [f0] "=&f"(f0), [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3),
711 [f4] "=&f"(f4), [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7),
712 [f8] "=&f"(f8), [tmp_a] "=&r"(tmp_a), [count] "=&r"(count)
713 : [a] "r"(a), [f9] "f"(f9), [f10] "f"(f10), [f11] "f"(f11),
714 [f12] "f"(f12), [f13] "f"(f13), [f14] "f"(f14)
715 : "memory");
716 f11 = rdft_w[6];
717 f12 = rdft_w[7];
718 f13 = rdft_wk3ri_second[2];
719 f14 = rdft_wk3ri_second[3];
720 __asm __volatile(
721 ".set push "
722 "\n\t"
723 ".set noreorder "
724 "\n\t"
725 "addiu %[tmp_a], %[a], 384 "
726 "\n\t"
727 "addiu %[count], $zero, 4 "
728 "\n\t"
729 "1: "
730 "\n\t"
731 "addiu %[count], %[count], -1 "
732 "\n\t"
733 "lwc1 %[f0], 0(%[tmp_a]) "
734 "\n\t"
735 "lwc1 %[f1], 4(%[tmp_a]) "
736 "\n\t"
737 "lwc1 %[f2], 32(%[tmp_a]) "
738 "\n\t"
739 "lwc1 %[f3], 36(%[tmp_a]) "
740 "\n\t"
741 "lwc1 %[f4], 64(%[tmp_a]) "
742 "\n\t"
743 "lwc1 %[f5], 68(%[tmp_a]) "
744 "\n\t"
745 "lwc1 %[f6], 96(%[tmp_a]) "
746 "\n\t"
747 "lwc1 %[f7], 100(%[tmp_a]) "
748 "\n\t"
749 "add.s %[f8], %[f0], %[f2] "
750 "\n\t"
751 "sub.s %[f0], %[f0], %[f2] "
752 "\n\t"
753 "add.s %[f2], %[f4], %[f6] "
754 "\n\t"
755 "sub.s %[f4], %[f4], %[f6] "
756 "\n\t"
757 "add.s %[f6], %[f1], %[f3] "
758 "\n\t"
759 "sub.s %[f1], %[f1], %[f3] "
760 "\n\t"
761 "add.s %[f3], %[f5], %[f7] "
762 "\n\t"
763 "sub.s %[f5], %[f5], %[f7] "
764 "\n\t"
765 "sub.s %[f7], %[f2], %[f8] "
766 "\n\t"
767 "add.s %[f2], %[f2], %[f8] "
768 "\n\t"
769 "add.s %[f8], %[f1], %[f4] "
770 "\n\t"
771 "sub.s %[f1], %[f1], %[f4] "
772 "\n\t"
773 "sub.s %[f4], %[f3], %[f6] "
774 "\n\t"
775 "add.s %[f3], %[f3], %[f6] "
776 "\n\t"
777 "sub.s %[f6], %[f0], %[f5] "
778 "\n\t"
779 "add.s %[f0], %[f0], %[f5] "
780 "\n\t"
781 "swc1 %[f2], 0(%[tmp_a]) "
782 "\n\t"
783 "swc1 %[f3], 4(%[tmp_a]) "
784 "\n\t"
785 "mul.s %[f5], %[f10], %[f7] "
786 "\n\t"
787 #if defined(MIPS32_R2_LE)
788 "mul.s %[f7], %[f9], %[f7] "
789 "\n\t"
790 "mul.s %[f2], %[f12], %[f8] "
791 "\n\t"
792 "mul.s %[f8], %[f11], %[f8] "
793 "\n\t"
794 "mul.s %[f3], %[f14], %[f1] "
795 "\n\t"
796 "mul.s %[f1], %[f13], %[f1] "
797 "\n\t"
798 "madd.s %[f5], %[f5], %[f9], %[f4] "
799 "\n\t"
800 "msub.s %[f7], %[f7], %[f10], %[f4] "
801 "\n\t"
802 "msub.s %[f2], %[f2], %[f11], %[f6] "
803 "\n\t"
804 "madd.s %[f8], %[f8], %[f12], %[f6] "
805 "\n\t"
806 "msub.s %[f3], %[f3], %[f13], %[f0] "
807 "\n\t"
808 "madd.s %[f1], %[f1], %[f14], %[f0] "
809 "\n\t"
810 "swc1 %[f5], 64(%[tmp_a]) "
811 "\n\t"
812 "swc1 %[f7], 68(%[tmp_a]) "
813 "\n\t"
814 #else
815 "mul.s %[f2], %[f9], %[f4] "
816 "\n\t"
817 "mul.s %[f4], %[f10], %[f4] "
818 "\n\t"
819 "mul.s %[f7], %[f9], %[f7] "
820 "\n\t"
821 "mul.s %[f3], %[f11], %[f6] "
822 "\n\t"
823 "mul.s %[f6], %[f12], %[f6] "
824 "\n\t"
825 "add.s %[f5], %[f5], %[f2] "
826 "\n\t"
827 "sub.s %[f7], %[f4], %[f7] "
828 "\n\t"
829 "mul.s %[f2], %[f12], %[f8] "
830 "\n\t"
831 "mul.s %[f8], %[f11], %[f8] "
832 "\n\t"
833 "mul.s %[f4], %[f14], %[f1] "
834 "\n\t"
835 "mul.s %[f1], %[f13], %[f1] "
836 "\n\t"
837 "sub.s %[f2], %[f3], %[f2] "
838 "\n\t"
839 "mul.s %[f3], %[f13], %[f0] "
840 "\n\t"
841 "mul.s %[f0], %[f14], %[f0] "
842 "\n\t"
843 "add.s %[f8], %[f8], %[f6] "
844 "\n\t"
845 "swc1 %[f5], 64(%[tmp_a]) "
846 "\n\t"
847 "swc1 %[f7], 68(%[tmp_a]) "
848 "\n\t"
849 "sub.s %[f3], %[f3], %[f4] "
850 "\n\t"
851 "add.s %[f1], %[f1], %[f0] "
852 "\n\t"
853 #endif
854 "swc1 %[f2], 32(%[tmp_a]) "
855 "\n\t"
856 "swc1 %[f8], 36(%[tmp_a]) "
857 "\n\t"
858 "swc1 %[f3], 96(%[tmp_a]) "
859 "\n\t"
860 "swc1 %[f1], 100(%[tmp_a]) "
861 "\n\t"
862 "bgtz %[count], 1b "
863 "\n\t"
864 " addiu %[tmp_a], %[tmp_a], 8 "
865 "\n\t"
866 ".set pop "
867 "\n\t"
868 : [f0] "=&f"(f0), [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3),
869 [f4] "=&f"(f4), [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7),
870 [f8] "=&f"(f8), [tmp_a] "=&r"(tmp_a), [count] "=&r"(count)
871 : [a] "r"(a), [f9] "f"(f9), [f10] "f"(f10), [f11] "f"(f11),
872 [f12] "f"(f12), [f13] "f"(f13), [f14] "f"(f14)
873 : "memory");
874 }
875
cftfsub_128_mips(float * a)876 void cftfsub_128_mips(float* a) {
877 float f0, f1, f2, f3, f4, f5, f6, f7, f8;
878 int tmp_a, count;
879
880 cft1st_128_mips(a);
881 cftmdl_128_mips(a);
882
883 __asm __volatile(
884 ".set push \n\t"
885 ".set noreorder \n\t"
886 "addiu %[tmp_a], %[a], 0 \n\t"
887 "addiu %[count], $zero, 16 \n\t"
888 "1: \n\t"
889 "addiu %[count], %[count], -1 \n\t"
890 "lwc1 %[f0], 0(%[tmp_a]) \n\t"
891 "lwc1 %[f2], 128(%[tmp_a]) \n\t"
892 "lwc1 %[f4], 256(%[tmp_a]) \n\t"
893 "lwc1 %[f6], 384(%[tmp_a]) \n\t"
894 "lwc1 %[f1], 4(%[tmp_a]) \n\t"
895 "lwc1 %[f3], 132(%[tmp_a]) \n\t"
896 "lwc1 %[f5], 260(%[tmp_a]) \n\t"
897 "lwc1 %[f7], 388(%[tmp_a]) \n\t"
898 "add.s %[f8], %[f0], %[f2] \n\t"
899 "sub.s %[f0], %[f0], %[f2] \n\t"
900 "add.s %[f2], %[f4], %[f6] \n\t"
901 "sub.s %[f4], %[f4], %[f6] \n\t"
902 "add.s %[f6], %[f1], %[f3] \n\t"
903 "sub.s %[f1], %[f1], %[f3] \n\t"
904 "add.s %[f3], %[f5], %[f7] \n\t"
905 "sub.s %[f5], %[f5], %[f7] \n\t"
906 "add.s %[f7], %[f8], %[f2] \n\t"
907 "sub.s %[f8], %[f8], %[f2] \n\t"
908 "add.s %[f2], %[f1], %[f4] \n\t"
909 "sub.s %[f1], %[f1], %[f4] \n\t"
910 "add.s %[f4], %[f6], %[f3] \n\t"
911 "sub.s %[f6], %[f6], %[f3] \n\t"
912 "sub.s %[f3], %[f0], %[f5] \n\t"
913 "add.s %[f0], %[f0], %[f5] \n\t"
914 "swc1 %[f7], 0(%[tmp_a]) \n\t"
915 "swc1 %[f8], 256(%[tmp_a]) \n\t"
916 "swc1 %[f2], 132(%[tmp_a]) \n\t"
917 "swc1 %[f1], 388(%[tmp_a]) \n\t"
918 "swc1 %[f4], 4(%[tmp_a]) \n\t"
919 "swc1 %[f6], 260(%[tmp_a]) \n\t"
920 "swc1 %[f3], 128(%[tmp_a]) \n\t"
921 "swc1 %[f0], 384(%[tmp_a]) \n\t"
922 "bgtz %[count], 1b \n\t"
923 " addiu %[tmp_a], %[tmp_a], 8 \n\t"
924 ".set pop \n\t"
925 : [f0] "=&f"(f0), [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3),
926 [f4] "=&f"(f4), [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7),
927 [f8] "=&f"(f8), [tmp_a] "=&r"(tmp_a), [count] "=&r"(count)
928 : [a] "r"(a)
929 : "memory");
930 }
931
cftbsub_128_mips(float * a)932 void cftbsub_128_mips(float* a) {
933 float f0, f1, f2, f3, f4, f5, f6, f7, f8;
934 int tmp_a, count;
935
936 cft1st_128_mips(a);
937 cftmdl_128_mips(a);
938
939 __asm __volatile(
940 ".set push \n\t"
941 ".set noreorder \n\t"
942 "addiu %[tmp_a], %[a], 0 \n\t"
943 "addiu %[count], $zero, 16 \n\t"
944 "1: \n\t"
945 "addiu %[count], %[count], -1 \n\t"
946 "lwc1 %[f0], 0(%[tmp_a]) \n\t"
947 "lwc1 %[f2], 128(%[tmp_a]) \n\t"
948 "lwc1 %[f4], 256(%[tmp_a]) \n\t"
949 "lwc1 %[f6], 384(%[tmp_a]) \n\t"
950 "lwc1 %[f1], 4(%[tmp_a]) \n\t"
951 "lwc1 %[f3], 132(%[tmp_a]) \n\t"
952 "lwc1 %[f5], 260(%[tmp_a]) \n\t"
953 "lwc1 %[f7], 388(%[tmp_a]) \n\t"
954 "add.s %[f8], %[f0], %[f2] \n\t"
955 "sub.s %[f0], %[f0], %[f2] \n\t"
956 "add.s %[f2], %[f4], %[f6] \n\t"
957 "sub.s %[f4], %[f4], %[f6] \n\t"
958 "add.s %[f6], %[f1], %[f3] \n\t"
959 "sub.s %[f1], %[f3], %[f1] \n\t"
960 "add.s %[f3], %[f5], %[f7] \n\t"
961 "sub.s %[f5], %[f5], %[f7] \n\t"
962 "add.s %[f7], %[f8], %[f2] \n\t"
963 "sub.s %[f8], %[f8], %[f2] \n\t"
964 "sub.s %[f2], %[f1], %[f4] \n\t"
965 "add.s %[f1], %[f1], %[f4] \n\t"
966 "add.s %[f4], %[f3], %[f6] \n\t"
967 "sub.s %[f6], %[f3], %[f6] \n\t"
968 "sub.s %[f3], %[f0], %[f5] \n\t"
969 "add.s %[f0], %[f0], %[f5] \n\t"
970 "neg.s %[f4], %[f4] \n\t"
971 "swc1 %[f7], 0(%[tmp_a]) \n\t"
972 "swc1 %[f8], 256(%[tmp_a]) \n\t"
973 "swc1 %[f2], 132(%[tmp_a]) \n\t"
974 "swc1 %[f1], 388(%[tmp_a]) \n\t"
975 "swc1 %[f6], 260(%[tmp_a]) \n\t"
976 "swc1 %[f3], 128(%[tmp_a]) \n\t"
977 "swc1 %[f0], 384(%[tmp_a]) \n\t"
978 "swc1 %[f4], 4(%[tmp_a]) \n\t"
979 "bgtz %[count], 1b \n\t"
980 " addiu %[tmp_a], %[tmp_a], 8 \n\t"
981 ".set pop \n\t"
982 : [f0] "=&f"(f0), [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3),
983 [f4] "=&f"(f4), [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7),
984 [f8] "=&f"(f8), [tmp_a] "=&r"(tmp_a), [count] "=&r"(count)
985 : [a] "r"(a)
986 : "memory");
987 }
988
rftfsub_128_mips(float * a)989 void rftfsub_128_mips(float* a) {
990 const float* c = rdft_w + 32;
991 const float f0 = 0.5f;
992 float* a1 = &a[2];
993 float* a2 = &a[126];
994 const float* c1 = &c[1];
995 const float* c2 = &c[31];
996 float f1, f2, f3, f4, f5, f6, f7, f8, f9, f10, f11, f12, f13, f14, f15;
997 int count;
998
999 __asm __volatile(
1000 ".set push \n\t"
1001 ".set noreorder \n\t"
1002 "lwc1 %[f6], 0(%[c2]) \n\t"
1003 "lwc1 %[f1], 0(%[a1]) \n\t"
1004 "lwc1 %[f2], 0(%[a2]) \n\t"
1005 "lwc1 %[f3], 4(%[a1]) \n\t"
1006 "lwc1 %[f4], 4(%[a2]) \n\t"
1007 "lwc1 %[f5], 0(%[c1]) \n\t"
1008 "sub.s %[f6], %[f0], %[f6] \n\t"
1009 "sub.s %[f7], %[f1], %[f2] \n\t"
1010 "add.s %[f8], %[f3], %[f4] \n\t"
1011 "addiu %[count], $zero, 15 \n\t"
1012 "mul.s %[f9], %[f6], %[f7] \n\t"
1013 "mul.s %[f6], %[f6], %[f8] \n\t"
1014 #if !defined(MIPS32_R2_LE)
1015 "mul.s %[f8], %[f5], %[f8] \n\t"
1016 "mul.s %[f5], %[f5], %[f7] \n\t"
1017 "sub.s %[f9], %[f9], %[f8] \n\t"
1018 "add.s %[f6], %[f6], %[f5] \n\t"
1019 #else
1020 "nmsub.s %[f9], %[f9], %[f5], %[f8] \n\t"
1021 "madd.s %[f6], %[f6], %[f5], %[f7] \n\t"
1022 #endif
1023 "sub.s %[f1], %[f1], %[f9] \n\t"
1024 "add.s %[f2], %[f2], %[f9] \n\t"
1025 "sub.s %[f3], %[f3], %[f6] \n\t"
1026 "sub.s %[f4], %[f4], %[f6] \n\t"
1027 "swc1 %[f1], 0(%[a1]) \n\t"
1028 "swc1 %[f2], 0(%[a2]) \n\t"
1029 "swc1 %[f3], 4(%[a1]) \n\t"
1030 "swc1 %[f4], 4(%[a2]) \n\t"
1031 "addiu %[a1], %[a1], 8 \n\t"
1032 "addiu %[a2], %[a2], -8 \n\t"
1033 "addiu %[c1], %[c1], 4 \n\t"
1034 "addiu %[c2], %[c2], -4 \n\t"
1035 "1: \n\t"
1036 "lwc1 %[f6], 0(%[c2]) \n\t"
1037 "lwc1 %[f1], 0(%[a1]) \n\t"
1038 "lwc1 %[f2], 0(%[a2]) \n\t"
1039 "lwc1 %[f3], 4(%[a1]) \n\t"
1040 "lwc1 %[f4], 4(%[a2]) \n\t"
1041 "lwc1 %[f5], 0(%[c1]) \n\t"
1042 "sub.s %[f6], %[f0], %[f6] \n\t"
1043 "sub.s %[f7], %[f1], %[f2] \n\t"
1044 "add.s %[f8], %[f3], %[f4] \n\t"
1045 "lwc1 %[f10], -4(%[c2]) \n\t"
1046 "lwc1 %[f11], 8(%[a1]) \n\t"
1047 "lwc1 %[f12], -8(%[a2]) \n\t"
1048 "mul.s %[f9], %[f6], %[f7] \n\t"
1049 "mul.s %[f6], %[f6], %[f8] \n\t"
1050 #if !defined(MIPS32_R2_LE)
1051 "mul.s %[f8], %[f5], %[f8] \n\t"
1052 "mul.s %[f5], %[f5], %[f7] \n\t"
1053 "lwc1 %[f13], 12(%[a1]) \n\t"
1054 "lwc1 %[f14], -4(%[a2]) \n\t"
1055 "lwc1 %[f15], 4(%[c1]) \n\t"
1056 "sub.s %[f9], %[f9], %[f8] \n\t"
1057 "add.s %[f6], %[f6], %[f5] \n\t"
1058 #else
1059 "lwc1 %[f13], 12(%[a1]) \n\t"
1060 "lwc1 %[f14], -4(%[a2]) \n\t"
1061 "lwc1 %[f15], 4(%[c1]) \n\t"
1062 "nmsub.s %[f9], %[f9], %[f5], %[f8] \n\t"
1063 "madd.s %[f6], %[f6], %[f5], %[f7] \n\t"
1064 #endif
1065 "sub.s %[f10], %[f0], %[f10] \n\t"
1066 "sub.s %[f5], %[f11], %[f12] \n\t"
1067 "add.s %[f7], %[f13], %[f14] \n\t"
1068 "sub.s %[f1], %[f1], %[f9] \n\t"
1069 "add.s %[f2], %[f2], %[f9] \n\t"
1070 "sub.s %[f3], %[f3], %[f6] \n\t"
1071 "mul.s %[f8], %[f10], %[f5] \n\t"
1072 "mul.s %[f10], %[f10], %[f7] \n\t"
1073 #if !defined(MIPS32_R2_LE)
1074 "mul.s %[f9], %[f15], %[f7] \n\t"
1075 "mul.s %[f15], %[f15], %[f5] \n\t"
1076 "sub.s %[f4], %[f4], %[f6] \n\t"
1077 "swc1 %[f1], 0(%[a1]) \n\t"
1078 "swc1 %[f2], 0(%[a2]) \n\t"
1079 "sub.s %[f8], %[f8], %[f9] \n\t"
1080 "add.s %[f10], %[f10], %[f15] \n\t"
1081 #else
1082 "swc1 %[f1], 0(%[a1]) \n\t"
1083 "swc1 %[f2], 0(%[a2]) \n\t"
1084 "sub.s %[f4], %[f4], %[f6] \n\t"
1085 "nmsub.s %[f8], %[f8], %[f15], %[f7] \n\t"
1086 "madd.s %[f10], %[f10], %[f15], %[f5] \n\t"
1087 #endif
1088 "swc1 %[f3], 4(%[a1]) \n\t"
1089 "swc1 %[f4], 4(%[a2]) \n\t"
1090 "sub.s %[f11], %[f11], %[f8] \n\t"
1091 "add.s %[f12], %[f12], %[f8] \n\t"
1092 "sub.s %[f13], %[f13], %[f10] \n\t"
1093 "sub.s %[f14], %[f14], %[f10] \n\t"
1094 "addiu %[c2], %[c2], -8 \n\t"
1095 "addiu %[c1], %[c1], 8 \n\t"
1096 "swc1 %[f11], 8(%[a1]) \n\t"
1097 "swc1 %[f12], -8(%[a2]) \n\t"
1098 "swc1 %[f13], 12(%[a1]) \n\t"
1099 "swc1 %[f14], -4(%[a2]) \n\t"
1100 "addiu %[a1], %[a1], 16 \n\t"
1101 "addiu %[count], %[count], -1 \n\t"
1102 "bgtz %[count], 1b \n\t"
1103 " addiu %[a2], %[a2], -16 \n\t"
1104 ".set pop \n\t"
1105 : [a1] "+r"(a1), [a2] "+r"(a2), [c1] "+r"(c1), [c2] "+r"(c2),
1106 [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3), [f4] "=&f"(f4),
1107 [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7), [f8] "=&f"(f8),
1108 [f9] "=&f"(f9), [f10] "=&f"(f10), [f11] "=&f"(f11), [f12] "=&f"(f12),
1109 [f13] "=&f"(f13), [f14] "=&f"(f14), [f15] "=&f"(f15),
1110 [count] "=&r"(count)
1111 : [f0] "f"(f0)
1112 : "memory");
1113 }
1114
rftbsub_128_mips(float * a)1115 void rftbsub_128_mips(float* a) {
1116 const float* c = rdft_w + 32;
1117 const float f0 = 0.5f;
1118 float* a1 = &a[2];
1119 float* a2 = &a[126];
1120 const float* c1 = &c[1];
1121 const float* c2 = &c[31];
1122 float f1, f2, f3, f4, f5, f6, f7, f8, f9, f10, f11, f12, f13, f14, f15;
1123 int count;
1124
1125 a[1] = -a[1];
1126 a[65] = -a[65];
1127
1128 __asm __volatile(
1129 ".set push \n\t"
1130 ".set noreorder \n\t"
1131 "lwc1 %[f6], 0(%[c2]) \n\t"
1132 "lwc1 %[f1], 0(%[a1]) \n\t"
1133 "lwc1 %[f2], 0(%[a2]) \n\t"
1134 "lwc1 %[f3], 4(%[a1]) \n\t"
1135 "lwc1 %[f4], 4(%[a2]) \n\t"
1136 "lwc1 %[f5], 0(%[c1]) \n\t"
1137 "sub.s %[f6], %[f0], %[f6] \n\t"
1138 "sub.s %[f7], %[f1], %[f2] \n\t"
1139 "add.s %[f8], %[f3], %[f4] \n\t"
1140 "addiu %[count], $zero, 15 \n\t"
1141 "mul.s %[f9], %[f6], %[f7] \n\t"
1142 "mul.s %[f6], %[f6], %[f8] \n\t"
1143 #if !defined(MIPS32_R2_LE)
1144 "mul.s %[f8], %[f5], %[f8] \n\t"
1145 "mul.s %[f5], %[f5], %[f7] \n\t"
1146 "add.s %[f9], %[f9], %[f8] \n\t"
1147 "sub.s %[f6], %[f6], %[f5] \n\t"
1148 #else
1149 "madd.s %[f9], %[f9], %[f5], %[f8] \n\t"
1150 "nmsub.s %[f6], %[f6], %[f5], %[f7] \n\t"
1151 #endif
1152 "sub.s %[f1], %[f1], %[f9] \n\t"
1153 "add.s %[f2], %[f2], %[f9] \n\t"
1154 "sub.s %[f3], %[f6], %[f3] \n\t"
1155 "sub.s %[f4], %[f6], %[f4] \n\t"
1156 "swc1 %[f1], 0(%[a1]) \n\t"
1157 "swc1 %[f2], 0(%[a2]) \n\t"
1158 "swc1 %[f3], 4(%[a1]) \n\t"
1159 "swc1 %[f4], 4(%[a2]) \n\t"
1160 "addiu %[a1], %[a1], 8 \n\t"
1161 "addiu %[a2], %[a2], -8 \n\t"
1162 "addiu %[c1], %[c1], 4 \n\t"
1163 "addiu %[c2], %[c2], -4 \n\t"
1164 "1: \n\t"
1165 "lwc1 %[f6], 0(%[c2]) \n\t"
1166 "lwc1 %[f1], 0(%[a1]) \n\t"
1167 "lwc1 %[f2], 0(%[a2]) \n\t"
1168 "lwc1 %[f3], 4(%[a1]) \n\t"
1169 "lwc1 %[f4], 4(%[a2]) \n\t"
1170 "lwc1 %[f5], 0(%[c1]) \n\t"
1171 "sub.s %[f6], %[f0], %[f6] \n\t"
1172 "sub.s %[f7], %[f1], %[f2] \n\t"
1173 "add.s %[f8], %[f3], %[f4] \n\t"
1174 "lwc1 %[f10], -4(%[c2]) \n\t"
1175 "lwc1 %[f11], 8(%[a1]) \n\t"
1176 "lwc1 %[f12], -8(%[a2]) \n\t"
1177 "mul.s %[f9], %[f6], %[f7] \n\t"
1178 "mul.s %[f6], %[f6], %[f8] \n\t"
1179 #if !defined(MIPS32_R2_LE)
1180 "mul.s %[f8], %[f5], %[f8] \n\t"
1181 "mul.s %[f5], %[f5], %[f7] \n\t"
1182 "lwc1 %[f13], 12(%[a1]) \n\t"
1183 "lwc1 %[f14], -4(%[a2]) \n\t"
1184 "lwc1 %[f15], 4(%[c1]) \n\t"
1185 "add.s %[f9], %[f9], %[f8] \n\t"
1186 "sub.s %[f6], %[f6], %[f5] \n\t"
1187 #else
1188 "lwc1 %[f13], 12(%[a1]) \n\t"
1189 "lwc1 %[f14], -4(%[a2]) \n\t"
1190 "lwc1 %[f15], 4(%[c1]) \n\t"
1191 "madd.s %[f9], %[f9], %[f5], %[f8] \n\t"
1192 "nmsub.s %[f6], %[f6], %[f5], %[f7] \n\t"
1193 #endif
1194 "sub.s %[f10], %[f0], %[f10] \n\t"
1195 "sub.s %[f5], %[f11], %[f12] \n\t"
1196 "add.s %[f7], %[f13], %[f14] \n\t"
1197 "sub.s %[f1], %[f1], %[f9] \n\t"
1198 "add.s %[f2], %[f2], %[f9] \n\t"
1199 "sub.s %[f3], %[f6], %[f3] \n\t"
1200 "mul.s %[f8], %[f10], %[f5] \n\t"
1201 "mul.s %[f10], %[f10], %[f7] \n\t"
1202 #if !defined(MIPS32_R2_LE)
1203 "mul.s %[f9], %[f15], %[f7] \n\t"
1204 "mul.s %[f15], %[f15], %[f5] \n\t"
1205 "sub.s %[f4], %[f6], %[f4] \n\t"
1206 "swc1 %[f1], 0(%[a1]) \n\t"
1207 "swc1 %[f2], 0(%[a2]) \n\t"
1208 "add.s %[f8], %[f8], %[f9] \n\t"
1209 "sub.s %[f10], %[f10], %[f15] \n\t"
1210 #else
1211 "swc1 %[f1], 0(%[a1]) \n\t"
1212 "swc1 %[f2], 0(%[a2]) \n\t"
1213 "sub.s %[f4], %[f6], %[f4] \n\t"
1214 "madd.s %[f8], %[f8], %[f15], %[f7] \n\t"
1215 "nmsub.s %[f10], %[f10], %[f15], %[f5] \n\t"
1216 #endif
1217 "swc1 %[f3], 4(%[a1]) \n\t"
1218 "swc1 %[f4], 4(%[a2]) \n\t"
1219 "sub.s %[f11], %[f11], %[f8] \n\t"
1220 "add.s %[f12], %[f12], %[f8] \n\t"
1221 "sub.s %[f13], %[f10], %[f13] \n\t"
1222 "sub.s %[f14], %[f10], %[f14] \n\t"
1223 "addiu %[c2], %[c2], -8 \n\t"
1224 "addiu %[c1], %[c1], 8 \n\t"
1225 "swc1 %[f11], 8(%[a1]) \n\t"
1226 "swc1 %[f12], -8(%[a2]) \n\t"
1227 "swc1 %[f13], 12(%[a1]) \n\t"
1228 "swc1 %[f14], -4(%[a2]) \n\t"
1229 "addiu %[a1], %[a1], 16 \n\t"
1230 "addiu %[count], %[count], -1 \n\t"
1231 "bgtz %[count], 1b \n\t"
1232 " addiu %[a2], %[a2], -16 \n\t"
1233 ".set pop \n\t"
1234 : [a1] "+r"(a1), [a2] "+r"(a2), [c1] "+r"(c1), [c2] "+r"(c2),
1235 [f1] "=&f"(f1), [f2] "=&f"(f2), [f3] "=&f"(f3), [f4] "=&f"(f4),
1236 [f5] "=&f"(f5), [f6] "=&f"(f6), [f7] "=&f"(f7), [f8] "=&f"(f8),
1237 [f9] "=&f"(f9), [f10] "=&f"(f10), [f11] "=&f"(f11), [f12] "=&f"(f12),
1238 [f13] "=&f"(f13), [f14] "=&f"(f14), [f15] "=&f"(f15),
1239 [count] "=&r"(count)
1240 : [f0] "f"(f0)
1241 : "memory");
1242 }
1243 #endif
1244
1245 } // namespace webrtc
1246