57 #if HAVE_INLINE_ASM && HAVE_MIPSFPU 58 static void vector_fmul_mips(
float *
dst,
const float *src0,
const float *src1,
64 for (i = 0; i <
len; i++)
65 dst[i] = src0[i] * src1[i];
67 float *
d = (
float *)dst;
68 float *d_end = d +
len;
69 float *
s0 = (
float *)src0;
70 float *
s1 = (
float *)src1;
72 float src0_0, src0_1, src0_2, src0_3;
73 float src1_0, src1_1, src1_2, src1_3;
77 "lwc1 %[src0_0], 0(%[s0]) \n\t" 78 "lwc1 %[src1_0], 0(%[s1]) \n\t" 79 "lwc1 %[src0_1], 4(%[s0]) \n\t" 80 "lwc1 %[src1_1], 4(%[s1]) \n\t" 81 "lwc1 %[src0_2], 8(%[s0]) \n\t" 82 "lwc1 %[src1_2], 8(%[s1]) \n\t" 83 "lwc1 %[src0_3], 12(%[s0]) \n\t" 84 "lwc1 %[src1_3], 12(%[s1]) \n\t" 85 "mul.s %[src0_0], %[src0_0], %[src1_0] \n\t" 86 "mul.s %[src0_1], %[src0_1], %[src1_1] \n\t" 87 "mul.s %[src0_2], %[src0_2], %[src1_2] \n\t" 88 "mul.s %[src0_3], %[src0_3], %[src1_3] \n\t" 89 "swc1 %[src0_0], 0(%[d]) \n\t" 90 "swc1 %[src0_1], 4(%[d]) \n\t" 91 "swc1 %[src0_2], 8(%[d]) \n\t" 92 "swc1 %[src0_3], 12(%[d]) \n\t" 93 "addiu %[s0], %[s0], 16 \n\t" 94 "addiu %[s1], %[s1], 16 \n\t" 95 "addiu %[d], %[d], 16 \n\t" 96 "bne %[d], %[d_end], 1b \n\t" 98 : [src0_0]
"=&f"(src0_0), [src0_1]
"=&f"(src0_1),
99 [src0_2]
"=&f"(src0_2), [src0_3]
"=&f"(src0_3),
100 [src1_0]
"=&f"(src1_0), [src1_1]
"=&f"(src1_1),
101 [src1_2]
"=&f"(src1_2), [src1_3]
"=&f"(src1_3),
102 [
d]
"+r"(
d), [s0]
"+r"(s0), [
s1]
"+r"(
s1)
109 static void vector_fmul_scalar_mips(
float *dst,
const float *
src,
float mul,
112 float temp0, temp1, temp2, temp3;
113 float *local_src = (
float*)src;
114 float *
end = local_src +
len;
119 ".set noreorder \n\t" 121 "lwc1 %[temp0], 0(%[src]) \n\t" 122 "lwc1 %[temp1], 4(%[src]) \n\t" 123 "lwc1 %[temp2], 8(%[src]) \n\t" 124 "lwc1 %[temp3], 12(%[src]) \n\t" 125 "addiu %[dst], %[dst], 16 \n\t" 126 "mul.s %[temp0], %[temp0], %[mul] \n\t" 127 "mul.s %[temp1], %[temp1], %[mul] \n\t" 128 "mul.s %[temp2], %[temp2], %[mul] \n\t" 129 "mul.s %[temp3], %[temp3], %[mul] \n\t" 130 "addiu %[src], %[src], 16 \n\t" 131 "swc1 %[temp0], -16(%[dst]) \n\t" 132 "swc1 %[temp1], -12(%[dst]) \n\t" 133 "swc1 %[temp2], -8(%[dst]) \n\t" 134 "bne %[src], %[end], 1b \n\t" 135 " swc1 %[temp3], -4(%[dst]) \n\t" 138 : [temp0]
"=&f"(temp0), [temp1]
"=&f"(temp1),
139 [temp2]
"=&f"(temp2), [temp3]
"=&f"(temp3),
140 [
dst]
"+r"(
dst), [src]
"+r"(local_src)
141 : [
end]
"r"(
end), [mul]
"f"(mul)
146 static void vector_fmul_window_mips(
float *dst,
const float *src0,
147 const float *src1,
const float *win,
int len)
153 float * dst_i, * dst_j, * dst_i2, * dst_j2;
154 float temp, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
160 for (i = -len, j = len - 1; i < 0; i += 8, j -= 8) {
165 dst_i2 = dst + i + 4;
166 dst_j2 = dst + j - 4;
169 "mul.s %[temp], %[s1], %[wi] \n\t" 170 "mul.s %[temp1], %[s1], %[wj] \n\t" 171 "mul.s %[temp2], %[s11], %[wi1] \n\t" 172 "mul.s %[temp3], %[s11], %[wj1] \n\t" 174 "msub.s %[temp], %[temp], %[s0], %[wj] \n\t" 175 "madd.s %[temp1], %[temp1], %[s0], %[wi] \n\t" 176 "msub.s %[temp2], %[temp2], %[s01], %[wj1] \n\t" 177 "madd.s %[temp3], %[temp3], %[s01], %[wi1] \n\t" 179 "swc1 %[temp], 0(%[dst_i]) \n\t" 180 "swc1 %[temp1], 0(%[dst_j]) \n\t" 181 "swc1 %[temp2], 4(%[dst_i]) \n\t" 182 "swc1 %[temp3], -4(%[dst_j]) \n\t" 184 "mul.s %[temp4], %[s12], %[wi2] \n\t" 185 "mul.s %[temp5], %[s12], %[wj2] \n\t" 186 "mul.s %[temp6], %[s13], %[wi3] \n\t" 187 "mul.s %[temp7], %[s13], %[wj3] \n\t" 189 "msub.s %[temp4], %[temp4], %[s02], %[wj2] \n\t" 190 "madd.s %[temp5], %[temp5], %[s02], %[wi2] \n\t" 191 "msub.s %[temp6], %[temp6], %[s03], %[wj3] \n\t" 192 "madd.s %[temp7], %[temp7], %[s03], %[wi3] \n\t" 194 "swc1 %[temp4], 8(%[dst_i]) \n\t" 195 "swc1 %[temp5], -8(%[dst_j]) \n\t" 196 "swc1 %[temp6], 12(%[dst_i]) \n\t" 197 "swc1 %[temp7], -12(%[dst_j]) \n\t" 198 : [
temp]
"=&f"(
temp), [temp1]
"=&f"(temp1), [temp2]
"=&f"(temp2),
199 [temp3]
"=&f"(temp3), [temp4]
"=&f"(temp4), [temp5]
"=&f"(temp5),
200 [temp6]
"=&f"(temp6), [temp7]
"=&f"(temp7)
201 : [dst_j]
"r"(dst_j), [dst_i]
"r" (dst_i),
202 [
s0]
"f"(src0[
i]), [wj]
"f"(win[j]), [
s1]
"f"(src1[j]),
203 [wi]
"f"(win[i]), [s01]
"f"(src0[i + 1]),[wj1]
"f"(win[j - 1]),
204 [s11]
"f"(src1[j - 1]), [wi1]
"f"(win[i + 1]), [s02]
"f"(src0[i + 2]),
205 [wj2]
"f"(win[j - 2]), [s12]
"f"(src1[j - 2]),[wi2]
"f"(win[i + 2]),
206 [s03]
"f"(src0[i + 3]), [wj3]
"f"(win[j - 3]), [s13]
"f"(src1[j - 3]),
212 "mul.s %[temp], %[s1], %[wi] \n\t" 213 "mul.s %[temp1], %[s1], %[wj] \n\t" 214 "mul.s %[temp2], %[s11], %[wi1] \n\t" 215 "mul.s %[temp3], %[s11], %[wj1] \n\t" 217 "msub.s %[temp], %[temp], %[s0], %[wj] \n\t" 218 "madd.s %[temp1], %[temp1], %[s0], %[wi] \n\t" 219 "msub.s %[temp2], %[temp2], %[s01], %[wj1] \n\t" 220 "madd.s %[temp3], %[temp3], %[s01], %[wi1] \n\t" 222 "swc1 %[temp], 0(%[dst_i2]) \n\t" 223 "swc1 %[temp1], 0(%[dst_j2]) \n\t" 224 "swc1 %[temp2], 4(%[dst_i2]) \n\t" 225 "swc1 %[temp3], -4(%[dst_j2]) \n\t" 227 "mul.s %[temp4], %[s12], %[wi2] \n\t" 228 "mul.s %[temp5], %[s12], %[wj2] \n\t" 229 "mul.s %[temp6], %[s13], %[wi3] \n\t" 230 "mul.s %[temp7], %[s13], %[wj3] \n\t" 232 "msub.s %[temp4], %[temp4], %[s02], %[wj2] \n\t" 233 "madd.s %[temp5], %[temp5], %[s02], %[wi2] \n\t" 234 "msub.s %[temp6], %[temp6], %[s03], %[wj3] \n\t" 235 "madd.s %[temp7], %[temp7], %[s03], %[wi3] \n\t" 237 "swc1 %[temp4], 8(%[dst_i2]) \n\t" 238 "swc1 %[temp5], -8(%[dst_j2]) \n\t" 239 "swc1 %[temp6], 12(%[dst_i2]) \n\t" 240 "swc1 %[temp7], -12(%[dst_j2]) \n\t" 242 [temp1]
"=&f"(temp1), [temp2]
"=&f"(temp2), [temp3]
"=&f"(temp3),
243 [temp4]
"=&f"(temp4), [temp5]
"=&f"(temp5), [temp6]
"=&f"(temp6),
244 [temp7]
"=&f" (temp7)
245 : [dst_j2]
"r"(dst_j2), [dst_i2]
"r"(dst_i2),
246 [
s0]
"f"(src0[i + 4]), [wj]
"f"(win[j - 4]), [
s1]
"f"(src1[j - 4]),
247 [wi]
"f"(win[i + 4]), [s01]
"f"(src0[i + 5]),[wj1]
"f"(win[j - 5]),
248 [s11]
"f"(src1[j - 5]), [wi1]
"f"(win[i + 5]), [s02]
"f"(src0[i + 6]),
249 [wj2]
"f"(win[j - 6]), [s12]
"f"(src1[j - 6]),[wi2]
"f"(win[i + 6]),
250 [s03]
"f"(src0[i + 7]), [wj3]
"f"(win[j - 7]), [s13]
"f"(src1[j - 7]),
260 float temp0, temp1, temp2, temp3, temp4;
261 float temp5, temp6, temp7, temp8, temp9;
262 float temp10, temp11, temp12, temp13, temp14, temp15;
268 "lwc1 %[temp0], 0(%[v1]) \n\t" 269 "lwc1 %[temp1], 4(%[v1]) \n\t" 270 "lwc1 %[temp2], 8(%[v1]) \n\t" 271 "lwc1 %[temp3], 12(%[v1]) \n\t" 272 "lwc1 %[temp4], 0(%[v2]) \n\t" 273 "lwc1 %[temp5], 4(%[v2]) \n\t" 274 "lwc1 %[temp6], 8(%[v2]) \n\t" 275 "lwc1 %[temp7], 12(%[v2]) \n\t" 276 "beq %[pom], $zero, 2f \n\t" 278 "sub.s %[temp8], %[temp0], %[temp4] \n\t" 279 "add.s %[temp9], %[temp0], %[temp4] \n\t" 280 "sub.s %[temp10], %[temp1], %[temp5] \n\t" 281 "add.s %[temp11], %[temp1], %[temp5] \n\t" 282 "sub.s %[temp12], %[temp2], %[temp6] \n\t" 283 "add.s %[temp13], %[temp2], %[temp6] \n\t" 284 "sub.s %[temp14], %[temp3], %[temp7] \n\t" 285 "add.s %[temp15], %[temp3], %[temp7] \n\t" 286 "addiu %[v1], %[v1], 16 \n\t" 287 "addiu %[v2], %[v2], 16 \n\t" 288 "addiu %[pom], %[pom], -1 \n\t" 289 "lwc1 %[temp0], 0(%[v1]) \n\t" 290 "lwc1 %[temp1], 4(%[v1]) \n\t" 291 "lwc1 %[temp2], 8(%[v1]) \n\t" 292 "lwc1 %[temp3], 12(%[v1]) \n\t" 293 "lwc1 %[temp4], 0(%[v2]) \n\t" 294 "lwc1 %[temp5], 4(%[v2]) \n\t" 295 "lwc1 %[temp6], 8(%[v2]) \n\t" 296 "lwc1 %[temp7], 12(%[v2]) \n\t" 297 "swc1 %[temp9], -16(%[v1]) \n\t" 298 "swc1 %[temp8], -16(%[v2]) \n\t" 299 "swc1 %[temp11], -12(%[v1]) \n\t" 300 "swc1 %[temp10], -12(%[v2]) \n\t" 301 "swc1 %[temp13], -8(%[v1]) \n\t" 302 "swc1 %[temp12], -8(%[v2]) \n\t" 303 "swc1 %[temp15], -4(%[v1]) \n\t" 304 "swc1 %[temp14], -4(%[v2]) \n\t" 305 "bgtz %[pom], 1b \n\t" 307 "sub.s %[temp8], %[temp0], %[temp4] \n\t" 308 "add.s %[temp9], %[temp0], %[temp4] \n\t" 309 "sub.s %[temp10], %[temp1], %[temp5] \n\t" 310 "add.s %[temp11], %[temp1], %[temp5] \n\t" 311 "sub.s %[temp12], %[temp2], %[temp6] \n\t" 312 "add.s %[temp13], %[temp2], %[temp6] \n\t" 313 "sub.s %[temp14], %[temp3], %[temp7] \n\t" 314 "add.s %[temp15], %[temp3], %[temp7] \n\t" 315 "swc1 %[temp9], 0(%[v1]) \n\t" 316 "swc1 %[temp8], 0(%[v2]) \n\t" 317 "swc1 %[temp11], 4(%[v1]) \n\t" 318 "swc1 %[temp10], 4(%[v2]) \n\t" 319 "swc1 %[temp13], 8(%[v1]) \n\t" 320 "swc1 %[temp12], 8(%[v2]) \n\t" 321 "swc1 %[temp15], 12(%[v1]) \n\t" 322 "swc1 %[temp14], 12(%[v2]) \n\t" 324 : [v1]
"+r"(v1), [v2]
"+r"(v2), [pom]
"+r"(pom), [temp0]
"=&f" (temp0),
325 [temp1]
"=&f"(temp1), [temp2]
"=&f"(temp2), [temp3]
"=&f"(temp3),
326 [temp4]
"=&f"(temp4), [temp5]
"=&f"(temp5), [temp6]
"=&f"(temp6),
327 [temp7]
"=&f"(temp7), [temp8]
"=&f"(temp8), [temp9]
"=&f"(temp9),
328 [temp10]
"=&f"(temp10), [temp11]
"=&f"(temp11), [temp12]
"=&f"(temp12),
329 [temp13]
"=&f"(temp13), [temp14]
"=&f"(temp14), [temp15]
"=&f"(temp15)
335 static void vector_fmul_reverse_mips(
float *dst,
const float *src0,
const float *src1,
int len){
337 float temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
340 for(i=0; i<(len>>2); i++)
344 "lwc1 %[temp0], 0(%[src0]) \n\t" 345 "lwc1 %[temp1], 0(%[src1]) \n\t" 346 "lwc1 %[temp2], 4(%[src0]) \n\t" 347 "lwc1 %[temp3], -4(%[src1]) \n\t" 348 "lwc1 %[temp4], 8(%[src0]) \n\t" 349 "lwc1 %[temp5], -8(%[src1]) \n\t" 350 "lwc1 %[temp6], 12(%[src0]) \n\t" 351 "lwc1 %[temp7], -12(%[src1]) \n\t" 352 "mul.s %[temp0], %[temp1], %[temp0] \n\t" 353 "mul.s %[temp2], %[temp3], %[temp2] \n\t" 354 "mul.s %[temp4], %[temp5], %[temp4] \n\t" 355 "mul.s %[temp6], %[temp7], %[temp6] \n\t" 356 "addiu %[src0], %[src0], 16 \n\t" 357 "addiu %[src1], %[src1], -16 \n\t" 358 "addiu %[dst], %[dst], 16 \n\t" 359 "swc1 %[temp0], -16(%[dst]) \n\t" 360 "swc1 %[temp2], -12(%[dst]) \n\t" 361 "swc1 %[temp4], -8(%[dst]) \n\t" 362 "swc1 %[temp6], -4(%[dst]) \n\t" 364 : [
dst]
"+r"(
dst), [src0]
"+r"(src0), [src1]
"+r"(src1),
365 [temp0]
"=&f"(temp0), [temp1]
"=&f"(temp1),[temp2]
"=&f"(temp2),
366 [temp3]
"=&f"(temp3), [temp4]
"=&f"(temp4), [temp5]
"=&f"(temp5),
367 [temp6]
"=&f"(temp6), [temp7]
"=&f"(temp7)
376 #if HAVE_INLINE_ASM && HAVE_MIPSFPU
void(* vector_fmul)(float *dst, const float *src0, const float *src1, int len)
Calculate the product of two vectors of floats and store the result in a vector of floats...
void(* vector_fmul_window)(float *dst, const float *src0, const float *src1, const float *win, int len)
Overlap/add with window function.
void ff_float_dsp_init_mips(AVFloatDSPContext *fdsp)
void(* butterflies_float)(float *av_restrict v1, float *av_restrict v2, int len)
Calculate the sum and difference of two vectors of floats.
synthesis window for stochastic i
void(* vector_fmul_scalar)(float *dst, const float *src, float mul, int len)
Multiply a vector of floats by a scalar float.
else dst[i][x+y *dst_stride[i]]
void(* vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len)
Calculate the product of two vectors of floats, and store the result in a vector of floats...