58 #define ENVELOPE_ADJUSTMENT_OFFSET 2
62 float X_low[32][40][2],
const float W[2][32][32][2],
66 int temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
67 float *p_x_low = &X_low[0][8][0];
68 float *p_w = (
float*)&W[buf_idx][0][0][0];
69 float *p_x1_low = &X_low[0][0][0];
70 float *p_w1 = (
float*)&W[1-buf_idx][24][0][0];
72 float *loop_end=p_x1_low + 2560;
77 "sw $0, 0(%[p_x1_low]) \n\t"
78 "sw $0, 4(%[p_x1_low]) \n\t"
79 "sw $0, 8(%[p_x1_low]) \n\t"
80 "sw $0, 12(%[p_x1_low]) \n\t"
81 "sw $0, 16(%[p_x1_low]) \n\t"
82 "sw $0, 20(%[p_x1_low]) \n\t"
83 "sw $0, 24(%[p_x1_low]) \n\t"
84 "sw $0, 28(%[p_x1_low]) \n\t"
85 PTR_ADDIU "%[p_x1_low],%[p_x1_low], 32 \n\t"
86 "bne %[p_x1_low], %[loop_end], 1b \n\t"
87 PTR_ADDIU "%[p_x1_low],%[p_x1_low], -10240 \n\t"
89 : [p_x1_low]
"+r"(p_x1_low)
90 : [loop_end]
"r"(loop_end)
94 for (k = 0; k < sbr->
kx[1]; k++) {
95 for (i = 0; i < 32; i+=4) {
98 "lw %[temp0], 0(%[p_w]) \n\t"
99 "lw %[temp1], 4(%[p_w]) \n\t"
100 "lw %[temp2], 256(%[p_w]) \n\t"
101 "lw %[temp3], 260(%[p_w]) \n\t"
102 "lw %[temp4], 512(%[p_w]) \n\t"
103 "lw %[temp5], 516(%[p_w]) \n\t"
104 "lw %[temp6], 768(%[p_w]) \n\t"
105 "lw %[temp7], 772(%[p_w]) \n\t"
106 "sw %[temp0], 0(%[p_x_low]) \n\t"
107 "sw %[temp1], 4(%[p_x_low]) \n\t"
108 "sw %[temp2], 8(%[p_x_low]) \n\t"
109 "sw %[temp3], 12(%[p_x_low]) \n\t"
110 "sw %[temp4], 16(%[p_x_low]) \n\t"
111 "sw %[temp5], 20(%[p_x_low]) \n\t"
112 "sw %[temp6], 24(%[p_x_low]) \n\t"
113 "sw %[temp7], 28(%[p_x_low]) \n\t"
114 PTR_ADDIU "%[p_x_low], %[p_x_low], 32 \n\t"
117 : [temp0]
"=&r"(temp0), [temp1]
"=&r"(temp1),
118 [temp2]
"=&r"(temp2), [temp3]
"=&r"(temp3),
119 [temp4]
"=&r"(temp4), [temp5]
"=&r"(temp5),
120 [temp6]
"=&r"(temp6), [temp7]
"=&r"(temp7),
121 [p_w]
"+r"(p_w), [p_x_low]
"+r"(p_x_low)
130 for (k = 0; k < sbr->
kx[0]; k++) {
131 for (i = 0; i < 2; i++) {
135 "lw %[temp0], 0(%[p_w1]) \n\t"
136 "lw %[temp1], 4(%[p_w1]) \n\t"
137 "lw %[temp2], 256(%[p_w1]) \n\t"
138 "lw %[temp3], 260(%[p_w1]) \n\t"
139 "lw %[temp4], 512(%[p_w1]) \n\t"
140 "lw %[temp5], 516(%[p_w1]) \n\t"
141 "lw %[temp6], 768(%[p_w1]) \n\t"
142 "lw %[temp7], 772(%[p_w1]) \n\t"
143 "sw %[temp0], 0(%[p_x1_low]) \n\t"
144 "sw %[temp1], 4(%[p_x1_low]) \n\t"
145 "sw %[temp2], 8(%[p_x1_low]) \n\t"
146 "sw %[temp3], 12(%[p_x1_low]) \n\t"
147 "sw %[temp4], 16(%[p_x1_low]) \n\t"
148 "sw %[temp5], 20(%[p_x1_low]) \n\t"
149 "sw %[temp6], 24(%[p_x1_low]) \n\t"
150 "sw %[temp7], 28(%[p_x1_low]) \n\t"
151 PTR_ADDIU "%[p_x1_low], %[p_x1_low], 32 \n\t"
154 : [temp0]
"=&r"(temp0), [temp1]
"=&r"(temp1),
155 [temp2]
"=&r"(temp2), [temp3]
"=&r"(temp3),
156 [temp4]
"=&r"(temp4), [temp5]
"=&r"(temp5),
157 [temp6]
"=&r"(temp6), [temp7]
"=&r"(temp7),
158 [p_w1]
"+r"(p_w1), [p_x1_low]
"+r"(p_x1_low)
170 const float Y0[38][64][2],
const float Y1[38][64][2],
171 const float X_low[32][40][2],
int ch)
175 int temp0, temp1, temp2, temp3;
176 const float *X_low1, *Y01, *Y11;
177 float *x1=&X[0][0][0];
184 "sw $0, 0(%[x1]) \n\t"
185 "sw $0, 4(%[x1]) \n\t"
186 "sw $0, 8(%[x1]) \n\t"
187 "sw $0, 12(%[x1]) \n\t"
188 "sw $0, 16(%[x1]) \n\t"
189 "sw $0, 20(%[x1]) \n\t"
190 "sw $0, 24(%[x1]) \n\t"
191 "sw $0, 28(%[x1]) \n\t"
193 "bne %[x1], %[j], 1b \n\t"
203 X_low1=&X_low[0][2][0];
205 for (k = 0; k < sbr->
kx[0]; k++) {
208 "move %[i], $zero \n\t"
210 "lw %[temp0], 0(%[X_low1]) \n\t"
211 "lw %[temp1], 4(%[X_low1]) \n\t"
212 "sw %[temp0], 0(%[x1]) \n\t"
213 "sw %[temp1], 9728(%[x1]) \n\t"
216 "addiu %[i], %[i], 1 \n\t"
217 "bne %[i], %[i_Temp], 2b \n\t"
219 : [x1]
"+r"(x1), [X_low1]
"+r"(X_low1), [i]
"=&r"(i),
220 [temp0]
"=&r"(temp0), [temp1]
"=&r"(temp1)
221 : [i_Temp]
"r"(i_Temp)
225 X_low1-=(i_Temp<<1)-80;
229 Y01=(
float*)&Y0[32][k][0];
231 for (; k < sbr->
kx[0] + sbr->
m[0]; k++) {
233 "move %[i], $zero \n\t"
235 "lw %[temp0], 0(%[Y01]) \n\t"
236 "lw %[temp1], 4(%[Y01]) \n\t"
237 "sw %[temp0], 0(%[x1]) \n\t"
238 "sw %[temp1], 9728(%[x1]) \n\t"
241 "addiu %[i], %[i], 1 \n\t"
242 "bne %[i], %[i_Temp], 3b \n\t"
244 : [x1]
"+r"(x1), [Y01]
"+r"(Y01), [i]
"=&r"(i),
245 [temp0]
"=&r"(temp0), [temp1]
"=&r"(temp1)
246 : [i_Temp]
"r"(i_Temp)
255 X_low1=&X_low[0][i_Temp+2][0];
258 for (k = 0; k < sbr->
kx[1]; k++) {
261 "move %[i], %[i_Temp] \n\t"
263 "lw %[temp0], 0(%[X_low1]) \n\t"
264 "lw %[temp1], 4(%[X_low1]) \n\t"
265 "sw %[temp0], 0(%[x1]) \n\t"
266 "sw %[temp1], 9728(%[x1]) \n\t"
269 "addiu %[i], %[i], 1 \n\t"
270 "bne %[i], %[temp3], 4b \n\t"
272 : [x1]
"+r"(x1), [X_low1]
"+r"(X_low1), [i]
"=&r"(i),
273 [temp0]
"=&r"(temp0), [temp1]
"=&r"(temp1),
275 : [i_Temp]
"r"(i_Temp), [temp3]
"r"(temp3)
278 x1 -= ((38-i_Temp)<<6)-1;
279 X_low1 -= ((38-i_Temp)<<1)- 80;
283 Y11=&Y1[i_Temp][k][0];
286 for (; k < sbr->
kx[1] + sbr->
m[1]; k++) {
289 "move %[i], %[i_Temp] \n\t"
291 "lw %[temp0], 0(%[Y11]) \n\t"
292 "lw %[temp1], 4(%[Y11]) \n\t"
293 "sw %[temp0], 0(%[x1]) \n\t"
294 "sw %[temp1], 9728(%[x1]) \n\t"
297 "addiu %[i], %[i], 1 \n\t"
298 "bne %[i], %[temp2], 5b \n\t"
300 : [x1]
"+r"(x1), [Y11]
"+r"(Y11), [i]
"=&r"(i),
301 [temp0]
"=&r"(temp0), [temp1]
"=&r"(temp1)
302 : [i_Temp]
"r"(i_Temp), [temp3]
"r"(temp3),
307 x1 -= ((32-i_Temp)<<6)-1;
308 Y11 -= ((32-i_Temp)<<7)-2;
314 static void sbr_hf_assemble_mips(
float Y1[38][64][2],
315 const float X_high[64][40][2],
321 const int kx = sbr->
kx[1];
322 const int m_max = sbr->
m[1];
323 static const float h_smooth[5] = {
331 float (*g_temp)[48] = ch_data->
g_temp, (*q_temp)[48] = ch_data->
q_temp;
334 float *g_temp1, *q_temp1, *pok, *pok1;
335 float temp1, temp2, temp3, temp4;
339 for (i = 0; i < h_SL; i++) {
340 memcpy(g_temp[i + 2*ch_data->
t_env[0]], sbr->
gain[0], m_max *
sizeof(sbr->
gain[0][0]));
341 memcpy(q_temp[i + 2*ch_data->
t_env[0]], sbr->
q_m[0], m_max *
sizeof(sbr->
q_m[0][0]));
349 for (i = 2 * ch_data->
t_env[e]; i < 2 * ch_data->t_env[e + 1]; i++) {
350 g_temp1 = g_temp[h_SL + i];
352 q_temp1 = q_temp[h_SL + i];
356 for (j=0; j<(size>>2); j++) {
358 "lw %[temp1], 0(%[pok]) \n\t"
359 "lw %[temp2], 4(%[pok]) \n\t"
360 "lw %[temp3], 8(%[pok]) \n\t"
361 "lw %[temp4], 12(%[pok]) \n\t"
362 "sw %[temp1], 0(%[g_temp1]) \n\t"
363 "sw %[temp2], 4(%[g_temp1]) \n\t"
364 "sw %[temp3], 8(%[g_temp1]) \n\t"
365 "sw %[temp4], 12(%[g_temp1]) \n\t"
366 "lw %[temp1], 0(%[pok1]) \n\t"
367 "lw %[temp2], 4(%[pok1]) \n\t"
368 "lw %[temp3], 8(%[pok1]) \n\t"
369 "lw %[temp4], 12(%[pok1]) \n\t"
370 "sw %[temp1], 0(%[q_temp1]) \n\t"
371 "sw %[temp2], 4(%[q_temp1]) \n\t"
372 "sw %[temp3], 8(%[q_temp1]) \n\t"
373 "sw %[temp4], 12(%[q_temp1]) \n\t"
375 PTR_ADDIU "%[g_temp1], %[g_temp1], 16 \n\t"
377 PTR_ADDIU "%[q_temp1], %[q_temp1], 16 \n\t"
379 : [temp1]
"=&r"(temp1), [temp2]
"=&r"(temp2),
380 [temp3]
"=&r"(temp3), [temp4]
"=&r"(temp4),
381 [pok]
"+r"(pok), [g_temp1]
"+r"(g_temp1),
382 [pok1]
"+r"(pok1), [q_temp1]
"+r"(q_temp1)
388 for (j=0; j<(size&3); j++) {
390 "lw %[temp1], 0(%[pok]) \n\t"
391 "lw %[temp2], 0(%[pok1]) \n\t"
392 "sw %[temp1], 0(%[g_temp1]) \n\t"
393 "sw %[temp2], 0(%[q_temp1]) \n\t"
395 PTR_ADDIU "%[g_temp1], %[g_temp1], 4 \n\t"
397 PTR_ADDIU "%[q_temp1], %[q_temp1], 4 \n\t"
399 : [temp1]
"=&r"(temp1), [temp2]
"=&r"(temp2),
400 [temp3]
"=&r"(temp3), [temp4]
"=&r"(temp4),
401 [pok]
"+r"(pok), [g_temp1]
"+r"(g_temp1),
402 [pok1]
"+r"(pok1), [q_temp1]
"+r"(q_temp1)
411 for (i = 2 * ch_data->
t_env[e]; i < 2 * ch_data->t_env[e + 1]; i++) {
414 float *g_filt, *q_filt;
416 if (h_SL && e != e_a[0] && e != e_a[1]) {
420 for (m = 0; m < m_max; m++) {
421 const int idx1 = i + h_SL;
425 for (j = 0; j <= h_SL; j++) {
426 g_filt[
m] += g_temp[idx1 - j][
m] * h_smooth[j];
427 q_filt[
m] += q_temp[idx1 - j][
m] * h_smooth[j];
431 g_filt = g_temp[i + h_SL];
435 sbr->
dsp.
hf_g_filt(Y1[i] + kx, X_high + kx, g_filt, m_max,
438 if (e != e_a[0] && e != e_a[1]) {
443 int idx = indexsine&1;
444 int A = (1-((indexsine+(kx & 1))&2));
445 int B = (A^(-idx)) + idx;
446 float *
out = &Y1[i][kx][idx];
447 float *
in = sbr->
s_m[e];
448 float temp0, temp1, temp2, temp3, temp4, temp5;
449 float A_f = (float)A;
450 float B_f = (float)B;
452 for (m = 0; m+1 < m_max; m+=2) {
458 "lwc1 %[temp0], 0(%[in]) \n\t"
459 "lwc1 %[temp1], 4(%[in]) \n\t"
460 "madd.s %[temp4], %[temp2], %[temp0], %[A_f] \n\t"
461 "madd.s %[temp5], %[temp3], %[temp1], %[B_f] \n\t"
462 "swc1 %[temp4], 0(%[out]) \n\t"
463 "swc1 %[temp5], 8(%[out]) \n\t"
467 : [temp0]
"=&f" (temp0), [temp1]
"=&f"(temp1),
468 [temp4]
"=&f" (temp4), [temp5]
"=&f"(temp5),
469 [
in]
"+r"(
in), [out]
"+r"(out)
470 : [A_f]
"f"(A_f), [B_f]
"f"(B_f), [temp2]
"f"(temp2),
476 out[2*
m ] += in[
m ] *
A;
478 indexnoise = (indexnoise + m_max) & 0x1ff;
479 indexsine = (indexsine + 1) & 3;
487 float (*alpha0)[2],
float (*alpha1)[2],
488 const float X_low[32][40][2],
int k0)
491 float temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7,
c;
492 float *phi1, *alpha_1, *alpha_0, res1, res2, temp_real, temp_im;
496 for (k = 0; k < k0; k++) {
499 phi1 = &phi[0][0][0];
500 alpha_1 = &alpha1[k][0];
501 alpha_0 = &alpha0[k][0];
505 "lwc1 %[temp0], 40(%[phi1]) \n\t"
506 "lwc1 %[temp1], 16(%[phi1]) \n\t"
507 "lwc1 %[temp2], 24(%[phi1]) \n\t"
508 "lwc1 %[temp3], 28(%[phi1]) \n\t"
509 "mul.s %[dk], %[temp0], %[temp1] \n\t"
510 "lwc1 %[temp4], 0(%[phi1]) \n\t"
511 "mul.s %[res2], %[temp2], %[temp2] \n\t"
512 "lwc1 %[temp5], 4(%[phi1]) \n\t"
513 "madd.s %[res2], %[res2], %[temp3], %[temp3] \n\t"
514 "lwc1 %[temp6], 8(%[phi1]) \n\t"
515 "div.s %[res2], %[res2], %[c] \n\t"
516 "lwc1 %[temp0], 12(%[phi1]) \n\t"
517 "sub.s %[dk], %[dk], %[res2] \n\t"
519 : [temp0]
"=&f"(temp0), [temp1]
"=&f"(temp1), [temp2]
"=&f"(temp2),
520 [temp3]
"=&f"(temp3), [temp4]
"=&f"(temp4), [temp5]
"=&f"(temp5),
521 [temp6]
"=&f"(temp6), [res2]
"=&f"(res2), [dk]
"=&f"(dk)
522 : [phi1]
"r"(phi1), [
c]
"f"(
c)
531 "mul.s %[temp_real], %[temp4], %[temp2] \n\t"
532 "nmsub.s %[temp_real], %[temp_real], %[temp5], %[temp3] \n\t"
533 "nmsub.s %[temp_real], %[temp_real], %[temp6], %[temp1] \n\t"
534 "mul.s %[temp_im], %[temp4], %[temp3] \n\t"
535 "madd.s %[temp_im], %[temp_im], %[temp5], %[temp2] \n\t"
536 "nmsub.s %[temp_im], %[temp_im], %[temp0], %[temp1] \n\t"
537 "div.s %[temp_real], %[temp_real], %[dk] \n\t"
538 "div.s %[temp_im], %[temp_im], %[dk] \n\t"
539 "swc1 %[temp_real], 0(%[alpha_1]) \n\t"
540 "swc1 %[temp_im], 4(%[alpha_1]) \n\t"
542 : [temp_real]
"=&f" (temp_real), [temp_im]
"=&f"(temp_im)
543 : [phi1]
"r"(phi1), [temp0]
"f"(temp0), [temp1]
"f"(temp1),
544 [temp2]
"f"(temp2), [temp3]
"f"(temp3), [temp4]
"f"(temp4),
545 [temp5]
"f"(temp5), [temp6]
"f"(temp6),
546 [alpha_1]
"r"(alpha_1), [dk]
"f"(dk)
556 "lwc1 %[temp6], 0(%[alpha_1]) \n\t"
557 "lwc1 %[temp7], 4(%[alpha_1]) \n\t"
558 "mul.s %[temp_real], %[temp6], %[temp2] \n\t"
559 "add.s %[temp_real], %[temp_real], %[temp4] \n\t"
560 "madd.s %[temp_real], %[temp_real], %[temp7], %[temp3] \n\t"
561 "mul.s %[temp_im], %[temp7], %[temp2] \n\t"
562 "add.s %[temp_im], %[temp_im], %[temp5] \n\t"
563 "nmsub.s %[temp_im], %[temp_im], %[temp6], %[temp3] \n\t"
564 "div.s %[temp_real], %[temp_real], %[temp1] \n\t"
565 "div.s %[temp_im], %[temp_im], %[temp1] \n\t"
566 "neg.s %[temp_real], %[temp_real] \n\t"
567 "neg.s %[temp_im], %[temp_im] \n\t"
568 "swc1 %[temp_real], 0(%[alpha_0]) \n\t"
569 "swc1 %[temp_im], 4(%[alpha_0]) \n\t"
571 : [temp_real]
"=&f"(temp_real), [temp_im]
"=&f"(temp_im),
572 [temp6]
"=&f"(temp6), [temp7]
"=&f"(temp7),
573 [res1]
"=&f"(res1), [res2]
"=&f"(res2)
574 : [alpha_1]
"r"(alpha_1), [alpha_0]
"r"(alpha_0),
575 [temp0]
"f"(temp0), [temp1]
"f"(temp1), [temp2]
"f"(temp2),
576 [temp3]
"f"(temp3), [temp4]
"f"(temp4), [temp5]
"f"(temp5)
582 "lwc1 %[temp1], 0(%[alpha_1]) \n\t"
583 "lwc1 %[temp2], 4(%[alpha_1]) \n\t"
584 "lwc1 %[temp_real], 0(%[alpha_0]) \n\t"
585 "lwc1 %[temp_im], 4(%[alpha_0]) \n\t"
586 "mul.s %[res1], %[temp1], %[temp1] \n\t"
587 "madd.s %[res1], %[res1], %[temp2], %[temp2] \n\t"
588 "mul.s %[res2], %[temp_real], %[temp_real] \n\t"
589 "madd.s %[res2], %[res2], %[temp_im], %[temp_im] \n\t"
591 : [temp_real]
"=&f"(temp_real), [temp_im]
"=&f"(temp_im),
592 [temp1]
"=&f"(temp1), [temp2]
"=&f"(temp2),
593 [res1]
"=&f"(res1), [res2]
"=&f"(res2)
594 : [alpha_1]
"r"(alpha_1), [alpha_0]
"r"(alpha_0)
598 if (res1 >= 16.0f || res2 >= 16.0f) {