Mercurial > hg > dmlib

comparison src/dmvecmat.c @ 1102:e06abfde6c39

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Cosmetics pass: Remove excess whitespace.
author Matti Hamalainen <ccr@tnsp.org>
date Tue, 03 Mar 2015 23:22:36 +0200
parents 27949209238b
children 848a88ce7a57
comparison
equal deleted inserted replaced
1101:0577f10dc3de 1102:e06abfde6c39
99 "movups 32(%%edx), %%xmm6\n" 99 "movups 32(%%edx), %%xmm6\n"
100 "movups 48(%%edx), %%xmm7\n" 100 "movups 48(%%edx), %%xmm7\n"
101 101
102 // vector -> xmm0 102 // vector -> xmm0
103 "movups %2, %%xmm0\n" 103 "movups %2, %%xmm0\n"
104 104
105 // zero final result in xmm2 105 // zero final result in xmm2
106 "xorps %%xmm2, %%xmm2\n" 106 "xorps %%xmm2, %%xmm2\n"
107 107
108 // perform shuffle and multiply and add whole "column" "X" 108 // perform shuffle and multiply and add whole "column" "X"
109 "movups %%xmm0, %%xmm1\n" 109 "movups %%xmm0, %%xmm1\n"
167 #ifdef DM_USE_SIMD 167 #ifdef DM_USE_SIMD
168 asm volatile 168 asm volatile
169 ( 169 (
170 // list[i] -> xmm0 170 // list[i] -> xmm0
171 "movups %1, %%xmm0\n" 171 "movups %1, %%xmm0\n"
172 172
173 // zero final result in xmm2 173 // zero final result in xmm2
174 "xorps %%xmm2, %%xmm2\n" 174 "xorps %%xmm2, %%xmm2\n"
175 175
176 // perform shuffle and multiply and add whole "column" "X" 176 // perform shuffle and multiply and add whole "column" "X"
177 "movups %%xmm0, %%xmm1\n" 177 "movups %%xmm0, %%xmm1\n"
253 "movups (%%ebx), %%xmm0\n" // mat1[0] 253 "movups (%%ebx), %%xmm0\n" // mat1[0]
254 "movups (%%edx), %%xmm1\n" // mat2[0] 254 "movups (%%edx), %%xmm1\n" // mat2[0]
255 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0] 255 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0]
256 "mulps %%xmm0, %%xmm1\n" 256 "mulps %%xmm0, %%xmm1\n"
257 "movups %%xmm1, %%xmm3\n" 257 "movups %%xmm1, %%xmm3\n"
258 258
259 // 1 259 // 1
260 "movups 16(%%ebx), %%xmm0\n" // mat1[0] 260 "movups 16(%%ebx), %%xmm0\n" // mat1[0]
261 "movups (%%edx), %%xmm1\n" // mat2[0] 261 "movups (%%edx), %%xmm1\n" // mat2[0]
262 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1] 262 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1]
263 "mulps %%xmm0, %%xmm1\n" 263 "mulps %%xmm0, %%xmm1\n"
288 "movups (%%ebx), %%xmm0\n" // mat1[0] 288 "movups (%%ebx), %%xmm0\n" // mat1[0]
289 "movups 16(%%edx), %%xmm1\n" // mat2[1] 289 "movups 16(%%edx), %%xmm1\n" // mat2[1]
290 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0] 290 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0]
291 "mulps %%xmm0, %%xmm1\n" 291 "mulps %%xmm0, %%xmm1\n"
292 "movups %%xmm1, %%xmm3\n" 292 "movups %%xmm1, %%xmm3\n"
293 293
294 // 1 294 // 1
295 "movups 16(%%ebx), %%xmm0\n" // mat1[0] 295 "movups 16(%%ebx), %%xmm0\n" // mat1[0]
296 "movups 16(%%edx), %%xmm1\n" // mat2[1] 296 "movups 16(%%edx), %%xmm1\n" // mat2[1]
297 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1] 297 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1]
298 "mulps %%xmm0, %%xmm1\n" 298 "mulps %%xmm0, %%xmm1\n"
323 "movups (%%ebx), %%xmm0\n" // mat1[0] 323 "movups (%%ebx), %%xmm0\n" // mat1[0]
324 "movups 32(%%edx), %%xmm1\n" // mat2[1] 324 "movups 32(%%edx), %%xmm1\n" // mat2[1]
325 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0] 325 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0]
326 "mulps %%xmm0, %%xmm1\n" 326 "mulps %%xmm0, %%xmm1\n"
327 "movups %%xmm1, %%xmm3\n" 327 "movups %%xmm1, %%xmm3\n"
328 328
329 // 1 329 // 1
330 "movups 16(%%ebx), %%xmm0\n" // mat1[0] 330 "movups 16(%%ebx), %%xmm0\n" // mat1[0]
331 "movups 32(%%edx), %%xmm1\n" // mat2[1] 331 "movups 32(%%edx), %%xmm1\n" // mat2[1]
332 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1] 332 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1]
333 "mulps %%xmm0, %%xmm1\n" 333 "mulps %%xmm0, %%xmm1\n"
358 "movups (%%ebx), %%xmm0\n" // mat1[0] 358 "movups (%%ebx), %%xmm0\n" // mat1[0]
359 "movups 48(%%edx), %%xmm1\n" // mat2[1] 359 "movups 48(%%edx), %%xmm1\n" // mat2[1]
360 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0] 360 "shufps $0x00, %%xmm1, %%xmm1\n" // mat2[0][0]
361 "mulps %%xmm0, %%xmm1\n" 361 "mulps %%xmm0, %%xmm1\n"
362 "movups %%xmm1, %%xmm3\n" 362 "movups %%xmm1, %%xmm3\n"
363 363
364 // 1 364 // 1
365 "movups 16(%%ebx), %%xmm0\n" // mat1[0] 365 "movups 16(%%ebx), %%xmm0\n" // mat1[0]
366 "movups 48(%%edx), %%xmm1\n" // mat2[1] 366 "movups 48(%%edx), %%xmm1\n" // mat2[1]
367 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1] 367 "shufps $0x55, %%xmm1, %%xmm1\n" // mat2[0][1]
368 "mulps %%xmm0, %%xmm1\n" 368 "mulps %%xmm0, %%xmm1\n"
429 q = cx * sz, 429 q = cx * sz,
430 l = cx * cz, 430 l = cx * cz,
431 i = sx * sz, 431 i = sx * sz,
432 j = sx * cz; 432 j = sx * cz;
433 433
434 mat->m[0][3] = 434 mat->m[0][3] =
435 mat->m[1][3] = 435 mat->m[1][3] =
436 mat->m[2][3] = 436 mat->m[2][3] =
437 mat->m[3][0] = 437 mat->m[3][0] =
438 mat->m[3][1] = 438 mat->m[3][1] =
439 mat->m[3][2] = 0; 439 mat->m[3][2] = 0;
440 440
441 mat->m[3][3] = 1.0f; 441 mat->m[3][3] = 1.0f;
442 442
443 mat->m[0][0] = cy * cz; 443 mat->m[0][0] = cy * cz;