1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
|
/*********************************************************************/
/* Copyright 2009, 2010 The University of Texas at Austin. */
/* All rights reserved. */
/* */
/* Redistribution and use in source and binary forms, with or */
/* without modification, are permitted provided that the following */
/* conditions are met: */
/* */
/* 1. Redistributions of source code must retain the above */
/* copyright notice, this list of conditions and the following */
/* disclaimer. */
/* */
/* 2. Redistributions in binary form must reproduce the above */
/* copyright notice, this list of conditions and the following */
/* disclaimer in the documentation and/or other materials */
/* provided with the distribution. */
/* */
/* THIS SOFTWARE IS PROVIDED BY THE UNIVERSITY OF TEXAS AT */
/* AUSTIN ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, */
/* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF */
/* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE */
/* DISCLAIMED. IN NO EVENT SHALL THE UNIVERSITY OF TEXAS AT */
/* AUSTIN OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, */
/* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES */
/* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE */
/* GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR */
/* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF */
/* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT */
/* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT */
/* OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE */
/* POSSIBILITY OF SUCH DAMAGE. */
/* */
/* The views and conclusions contained in the software and */
/* documentation are those of the authors and should not be */
/* interpreted as representing official policies, either expressed */
/* or implied, of The University of Texas at Austin. */
/*********************************************************************/
#define ASSEMBLER
#include "common.h"
#ifdef movsd
#undef movsd
#endif
#ifdef PENTIUM3
#ifdef HAVE_SSE
#define PREFETCH prefetcht0
#define PREFETCHW prefetcht0
#define PREFETCHSIZE (16 * 2)
#endif
#define movsd movlps
#endif
#ifdef PENTIUM4
#define PREFETCH prefetcht0
#define PREFETCHW prefetcht0
#define PREFETCHSIZE (16 * 4)
#endif
#if defined(CORE2) || defined(PENRYN) || defined(DUNNINGTON) || defined(NEHALEM)
#define PREFETCH prefetcht0
#define PREFETCHW prefetcht0
#define PREFETCHSIZE (16 * 7)
#endif
#ifdef OPTERON
#define PREFETCH prefetchnta
#define PREFETCHW prefetchw
#define PREFETCHSIZE (16 * 3)
#define movsd movlps
#endif
#ifdef BARCELONA
#define PREFETCH prefetchnta
#define PREFETCHW prefetchw
#define PREFETCHSIZE (16 * 5)
#endif
#ifdef ATOM
#define PREFETCH prefetchnta
#define PREFETCHW prefetcht0
#define PREFETCHSIZE (16 * 6)
#endif
#ifdef NANO
#define PREFETCH prefetcht0
#define PREFETCHSIZE (16 * 4)
#endif
#define STACKSIZE 16
#define ARGS 20
#define M 4 + STACKSIZE+ARGS(%esp)
#define N 8 + STACKSIZE+ARGS(%esp)
#define ALPHA 16 + STACKSIZE+ARGS(%esp)
#define A 20 + STACKSIZE+ARGS(%esp)
#define STACK_LDA 24 + STACKSIZE+ARGS(%esp)
#define STACK_X 28 + STACKSIZE+ARGS(%esp)
#define STACK_INCX 32 + STACKSIZE+ARGS(%esp)
#define Y 36 + STACKSIZE+ARGS(%esp)
#define STACK_INCY 40 + STACKSIZE+ARGS(%esp)
#define BUFFER 44 + STACKSIZE+ARGS(%esp)
#define MMM 0+STACKSIZE(%esp)
#define NN 4+STACKSIZE(%esp)
#define AA 8+STACKSIZE(%esp)
#define LDAX 12+STACKSIZE(%esp)
#define XX 16+STACKSIZE(%esp)
#define I %eax
#define J %ebx
#define INCX J
#define INCY %ecx
#define A1 %esi
#define X %edx
#define Y1 %edi
#define LDA %ebp
PROLOGUE
subl $ARGS,%esp
pushl %ebp
pushl %edi
pushl %esi
pushl %ebx
PROFCODE
movl STACK_LDA, LDA
movl LDA,LDAX # backup LDA
movl STACK_X, X
movl X,XX
movl N,J
movl J,NN # backup N
movl A,J
movl J,AA # backup A
movl M,J
movl J,MMM # mov M to MMM
.L0t:
xorl J,J
addl $1,J
sall $22,J # J=2^24*sizeof(float)=buffer size(16MB)
subl $8, J # Don't use last 8 float in the buffer.
# Now, split M by block J
subl J,MMM # MMM=MMM-J
movl J,M
jge .L00t
ALIGN_4
movl MMM,%eax
addl J,%eax
jle .L999x
movl %eax,M
.L00t:
movl AA,%eax
movl %eax,A # mov AA to A
movl NN,%eax
movl %eax,N # reset N
movl LDAX, LDA # reset LDA
movl XX,X
movl STACK_INCX, INCX
movl STACK_INCY, INCY
leal (,INCX, SIZE), INCX
leal (,INCY, SIZE), INCY
leal (,LDA, SIZE), LDA
subl $-32 * SIZE, A
cmpl $0, N
jle .L999
cmpl $0, M
jle .L999
movl BUFFER, Y1
movl M, I
sarl $3, I
jle .L05
ALIGN_4
.L02:
movss (X), %xmm0
addl INCX, X
movss (X), %xmm1
addl INCX, X
unpcklps %xmm1, %xmm0
movss (X), %xmm2
addl INCX, X
movss (X), %xmm3
addl INCX, X
unpcklps %xmm3, %xmm2
movss (X), %xmm4
addl INCX, X
movss (X), %xmm5
addl INCX, X
unpcklps %xmm5, %xmm4
movss (X), %xmm6
addl INCX, X
movss (X), %xmm7
addl INCX, X
unpcklps %xmm7, %xmm6
movlps %xmm0, 0 * SIZE(Y1)
movlps %xmm2, 2 * SIZE(Y1)
movlps %xmm4, 4 * SIZE(Y1)
movlps %xmm6, 6 * SIZE(Y1)
addl $8 * SIZE, Y1
decl I
jg .L02
ALIGN_4
.L05:
movl M, I
andl $7, I
jle .L10
ALIGN_2
.L06:
movss (X), %xmm0
addl INCX, X
movss %xmm0, 0 * SIZE(Y1)
addl $SIZE, Y1
decl I
jg .L06
ALIGN_4
//Padding zero to prevent loading the dirty number from buffer.
movl M, I
movl $8, J
andl $7, I
xorps %xmm0, %xmm0
subl I, J
ALIGN_2
.L07:
movss %xmm0, 0 * SIZE(Y1)
addl $SIZE, Y1
decl J
jg .L07
ALIGN_4
.L10:
movl Y, Y1
movl N, J
sarl $1, J
jle .L20
ALIGN_3
.L11:
movl BUFFER, X
addl $32 * SIZE, X
movl A, A1
leal (A1, LDA, 2), %eax
movl %eax, A
xorps %xmm0, %xmm0
xorps %xmm1, %xmm1
movaps -32 * SIZE(X), %xmm2
movaps -28 * SIZE(X), %xmm3
movl M, I
sarl $4, I
jle .L15
movsd -32 * SIZE(A1), %xmm4
movhps -30 * SIZE(A1), %xmm4
movsd -32 * SIZE(A1, LDA), %xmm5
movhps -30 * SIZE(A1, LDA), %xmm5
movsd -28 * SIZE(A1), %xmm6
movhps -26 * SIZE(A1), %xmm6
movsd -28 * SIZE(A1, LDA), %xmm7
movhps -26 * SIZE(A1, LDA), %xmm7
decl I
jle .L13
ALIGN_4
.L12:
#ifdef PREFETCH
PREFETCH PREFETCHSIZE * SIZE(A1)
#endif
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
movsd -24 * SIZE(A1), %xmm4
movhps -22 * SIZE(A1), %xmm4
mulps %xmm2, %xmm5
movaps -24 * SIZE(X), %xmm2
addps %xmm5, %xmm1
movsd -24 * SIZE(A1, LDA), %xmm5
movhps -22 * SIZE(A1, LDA), %xmm5
mulps %xmm3, %xmm6
addps %xmm6, %xmm0
movsd -20 * SIZE(A1), %xmm6
movhps -18 * SIZE(A1), %xmm6
mulps %xmm3, %xmm7
movaps -20 * SIZE(X), %xmm3
addps %xmm7, %xmm1
movsd -20 * SIZE(A1, LDA), %xmm7
movhps -18 * SIZE(A1, LDA), %xmm7
#ifdef PREFETCH
PREFETCH PREFETCHSIZE * SIZE(A1, LDA)
#endif
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
movsd -16 * SIZE(A1), %xmm4
movhps -14 * SIZE(A1), %xmm4
mulps %xmm2, %xmm5
movaps -16 * SIZE(X), %xmm2
addps %xmm5, %xmm1
movsd -16 * SIZE(A1, LDA), %xmm5
movhps -14 * SIZE(A1, LDA), %xmm5
mulps %xmm3, %xmm6
addps %xmm6, %xmm0
movsd -12 * SIZE(A1), %xmm6
movhps -10 * SIZE(A1), %xmm6
mulps %xmm3, %xmm7
movaps -12 * SIZE(X), %xmm3
addps %xmm7, %xmm1
movsd -12 * SIZE(A1, LDA), %xmm7
movhps -10 * SIZE(A1, LDA), %xmm7
addl $16 * SIZE, A1
addl $16 * SIZE, X
decl I
jg .L12
ALIGN_4
.L13:
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
movsd -24 * SIZE(A1), %xmm4
movhps -22 * SIZE(A1), %xmm4
mulps %xmm2, %xmm5
movaps -24 * SIZE(X), %xmm2
addps %xmm5, %xmm1
movsd -24 * SIZE(A1, LDA), %xmm5
movhps -22 * SIZE(A1, LDA), %xmm5
mulps %xmm3, %xmm6
addps %xmm6, %xmm0
movsd -20 * SIZE(A1), %xmm6
movhps -18 * SIZE(A1), %xmm6
mulps %xmm3, %xmm7
movaps -20 * SIZE(X), %xmm3
addps %xmm7, %xmm1
movsd -20 * SIZE(A1, LDA), %xmm7
movhps -18 * SIZE(A1, LDA), %xmm7
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
mulps %xmm2, %xmm5
movaps -16 * SIZE(X), %xmm2
addps %xmm5, %xmm1
mulps %xmm3, %xmm6
addps %xmm6, %xmm0
mulps %xmm3, %xmm7
movaps -12 * SIZE(X), %xmm3
addps %xmm7, %xmm1
addl $16 * SIZE, A1
addl $16 * SIZE, X
ALIGN_4
.L15:
testl $8, M
jle .L16
movsd -32 * SIZE(A1), %xmm4
movhps -30 * SIZE(A1), %xmm4
movsd -32 * SIZE(A1, LDA), %xmm5
movhps -30 * SIZE(A1, LDA), %xmm5
movsd -28 * SIZE(A1), %xmm6
movhps -26 * SIZE(A1), %xmm6
movsd -28 * SIZE(A1, LDA), %xmm7
movhps -26 * SIZE(A1, LDA), %xmm7
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
mulps %xmm2, %xmm5
movaps -24 * SIZE(X), %xmm2
addps %xmm5, %xmm1
mulps %xmm3, %xmm6
addps %xmm6, %xmm0
mulps %xmm3, %xmm7
movaps -20 * SIZE(X), %xmm3
addps %xmm7, %xmm1
addl $8 * SIZE, A1
addl $8 * SIZE, X
ALIGN_4
.L16:
testl $4, M
jle .L17
movsd -32 * SIZE(A1), %xmm4
movhps -30 * SIZE(A1), %xmm4
movsd -32 * SIZE(A1, LDA), %xmm5
movhps -30 * SIZE(A1, LDA), %xmm5
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
mulps %xmm2, %xmm5
addps %xmm5, %xmm1
movaps %xmm3, %xmm2
addl $4 * SIZE, A1
ALIGN_4
.L17:
testl $2, M
jle .L18
#ifdef movsd
xorps %xmm4, %xmm4
#endif
movsd -32 * SIZE(A1), %xmm4
#ifdef movsd
xorps %xmm5, %xmm5
#endif
movsd -32 * SIZE(A1, LDA), %xmm5
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
mulps %xmm2, %xmm5
addps %xmm5, %xmm1
movhlps %xmm2, %xmm2
addl $2 * SIZE, A1
ALIGN_4
.L18:
testl $1, M
jle .L19
movss -32 * SIZE(A1), %xmm4
mulss %xmm2, %xmm4
addss %xmm4, %xmm0
movss -32 * SIZE(A1, LDA), %xmm5
mulss %xmm2, %xmm5
addss %xmm5, %xmm1
ALIGN_4
.L19:
#ifdef HAVE_SSE3
haddps %xmm0, %xmm0
haddps %xmm1, %xmm1
haddps %xmm0, %xmm0
haddps %xmm1, %xmm1
#else
movhlps %xmm0, %xmm2
movhlps %xmm1, %xmm3
addps %xmm2, %xmm0
addps %xmm3, %xmm1
movaps %xmm0, %xmm2
shufps $1, %xmm0, %xmm0
movaps %xmm1, %xmm3
shufps $1, %xmm1, %xmm1
addss %xmm2, %xmm0
addss %xmm3, %xmm1
#endif
movss ALPHA, %xmm7
mulss %xmm7, %xmm0
mulss %xmm7, %xmm1
addss (Y1), %xmm0
addss (Y1, INCY), %xmm1
movss %xmm0, (Y1)
movss %xmm1, (Y1, INCY)
leal (Y1, INCY, 2), Y1
decl J
jg .L11
ALIGN_4
.L20:
testl $1, N
jle .L999
movl BUFFER, X
addl $32 * SIZE, X
movl A, A1
xorps %xmm0, %xmm0
xorps %xmm1, %xmm1
movaps -32 * SIZE(X), %xmm2
movaps -28 * SIZE(X), %xmm3
movl M, I
sarl $4, I
jle .L25
movsd -32 * SIZE(A1), %xmm4
movhps -30 * SIZE(A1), %xmm4
movsd -28 * SIZE(A1), %xmm6
movhps -26 * SIZE(A1), %xmm6
decl I
jle .L23
ALIGN_4
.L22:
#ifdef PREFETCH
PREFETCH PREFETCHSIZE * SIZE(A1)
#endif
mulps %xmm2, %xmm4
movaps -24 * SIZE(X), %xmm2
addps %xmm4, %xmm0
movsd -24 * SIZE(A1), %xmm4
movhps -22 * SIZE(A1), %xmm4
mulps %xmm3, %xmm6
movaps -20 * SIZE(X), %xmm3
addps %xmm6, %xmm0
movsd -20 * SIZE(A1), %xmm6
movhps -18 * SIZE(A1), %xmm6
mulps %xmm2, %xmm4
movaps -16 * SIZE(X), %xmm2
addps %xmm4, %xmm0
movsd -16 * SIZE(A1), %xmm4
movhps -14 * SIZE(A1), %xmm4
mulps %xmm3, %xmm6
movaps -12 * SIZE(X), %xmm3
addps %xmm6, %xmm0
movsd -12 * SIZE(A1), %xmm6
movhps -10 * SIZE(A1), %xmm6
addl $16 * SIZE, A1
addl $16 * SIZE, X
decl I
jg .L22
ALIGN_4
.L23:
mulps %xmm2, %xmm4
movaps -24 * SIZE(X), %xmm2
addps %xmm4, %xmm0
movsd -24 * SIZE(A1), %xmm4
movhps -22 * SIZE(A1), %xmm4
mulps %xmm3, %xmm6
movaps -20 * SIZE(X), %xmm3
addps %xmm6, %xmm0
movsd -20 * SIZE(A1), %xmm6
movhps -18 * SIZE(A1), %xmm6
mulps %xmm2, %xmm4
movaps -16 * SIZE(X), %xmm2
addps %xmm4, %xmm0
mulps %xmm3, %xmm6
movaps -12 * SIZE(X), %xmm3
addps %xmm6, %xmm0
addl $16 * SIZE, A1
addl $16 * SIZE, X
ALIGN_4
.L25:
testl $8, M
jle .L26
movsd -32 * SIZE(A1), %xmm4
movhps -30 * SIZE(A1), %xmm4
movsd -28 * SIZE(A1), %xmm6
movhps -26 * SIZE(A1), %xmm6
mulps %xmm2, %xmm4
movaps -24 * SIZE(X), %xmm2
addps %xmm4, %xmm0
mulps %xmm3, %xmm6
movaps -20 * SIZE(X), %xmm3
addps %xmm6, %xmm0
addl $8 * SIZE, A1
addl $8 * SIZE, X
ALIGN_4
.L26:
testl $4, M
jle .L27
movsd -32 * SIZE(A1), %xmm4
movhps -30 * SIZE(A1), %xmm4
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
movaps %xmm3, %xmm2
addl $4 * SIZE, A1
ALIGN_4
.L27:
testl $2, M
jle .L28
#ifdef movsd
xorps %xmm4, %xmm4
#endif
movsd -32 * SIZE(A1), %xmm4
mulps %xmm2, %xmm4
addps %xmm4, %xmm0
movhlps %xmm2, %xmm2
addl $2 * SIZE, A1
ALIGN_4
.L28:
testl $1, M
jle .L29
movss -32 * SIZE(A1), %xmm4
mulss %xmm2, %xmm4
addss %xmm4, %xmm0
ALIGN_4
.L29:
#ifdef HAVE_SSE3
haddps %xmm0, %xmm0
haddps %xmm0, %xmm0
#else
movhlps %xmm0, %xmm2
addps %xmm2, %xmm0
movaps %xmm0, %xmm2
shufps $1, %xmm0, %xmm0
addss %xmm2, %xmm0
#endif
movss ALPHA, %xmm7
mulss %xmm7, %xmm0
addss (Y1), %xmm0
movss %xmm0, (Y1)
ALIGN_4
.L999:
movl M,J
leal (,J,SIZE),%eax
addl %eax,AA
movl XX,J
addl %eax,J
movl J,XX
jmp .L0t
ALIGN_4
.L999x:
popl %ebx
popl %esi
popl %edi
popl %ebp
addl $ARGS,%esp
ret
EPILOGUE
|
