« back to all changes in this revision
Viewing changes to blitz/asm_scale.S
- browse files at revision 1
- compare with another revision
- download diff
- download tarball
- view history from revision 1
- Committer: Bazaar Package Importer
- Author(s): Jonathan Riddell
- Date: 2007-08-31 09:50:41 UTC
- Revision ID: james.westby@ubuntu.com-20070831095041-rlguoifpqerdrs2o
Tags: upstream-0.0.706674
Import upstream version 0.0.706674
- files added:
- blitz
- blitz/private
- test
added
removed
blitz/asm_scale.S
1
2
#if defined(__EMX__)
3
/* Due to strange behaviour of as.exe we use this macros */
4
/* For all OS/2 coders - please use PGCC to compile this code */
5
#define PR_(foo) ___##foo
6
#define PT_(foo,func) ___##foo,##func
7
#define SIZE(sym) \
8
.___end_##sym:; \
9
.size ___##sym,.___end_##sym-___##sym; \
10
.align 8;
11
#elif defined(__APPLE__)
12
#define PR_(foo) ___##foo
13
#define PT_(foo,func) ___##foo,##func
14
#define SIZE(sym) \
15
.__end_##sym:; \
16
.align 8;
17
#else
18
#define PR_(foo) __##foo
19
#define PT_(foo,func) __##foo,##func
20
#define SIZE(sym) \
21
.__end_##sym:; \
22
.size __##sym,.__end_##sym-__##sym; \
23
.align 8;
24
#endif
25
26
#if defined(__i386__) && ( defined(__GNUC__) || defined(__INTEL_COMPILER) )
27
28
29
/*\
30
|*| MMX assembly scaling routine for Imlib2
31
|*| Written by Willem Monsuwe <willem@stack.nl>
32
\*/
33
34
.text
35
.align 8
36
.globl PR_(qimageScale_mmx_AARGBA)
37
/* .type PT_(qimageScale_mmx_AARGBA,@function) */
38
39
40
/*\ Prototype: __qimageScale_mmx_AARGBA(ImlibScaleInfo *isi, DATA32 *dest,
41
|*| int dxx, int dyy, int dx, int dy, int dw, int dh, int dow, int sow)
42
\*/
43
44
#define isi 8(%ebp)
45
#define dest 12(%ebp)
46
#define dxx 16(%ebp)
47
#define dyy 20(%ebp)
48
#define dx 24(%ebp)
49
#define dy 28(%ebp)
50
#define dw 32(%ebp)
51
#define dh 36(%ebp)
52
#define dow 40(%ebp)
53
#define sow 44(%ebp)
54
55
/*\ Local variables that didn't fit in registers \*/
56
#define y -4(%ebp)
57
#define yp -8(%ebp)
58
#define yap -12(%ebp)
59
#define xp -16(%ebp)
60
#define xap -20(%ebp)
61
#define Cx -24(%ebp)
62
#define Mx -28(%ebp)
63
#define Cy -32(%ebp)
64
#define My -36(%ebp)
65
#define sow_4 -40(%ebp)
66
67
/*\ When %edx points to ImlibScaleInfo, these are the members \*/
68
#define xpoints (%edx)
69
#define ypoints 4(%edx)
70
#define xapoints 8(%edx)
71
#define yapoints 12(%edx)
72
#define xup_yup 16(%edx)
73
74
PR_(qimageScale_mmx_AARGBA):
75
pushl %ebp
76
movl %esp, %ebp
77
subl $40, %esp
78
pushl %ebx
79
pushl %ecx
80
pushl %edx
81
pushl %edi
82
pushl %esi
83
movl isi, %edx
84
85
/*\ Check (dw > 0) && (dh > 0) \*/
86
cmpl $0, dw
87
jle .scale_leave
88
cmpl $0, dh
89
jle .scale_leave
90
91
/*\ X-based array pointers point to the end; we're looping up to 0 \*/
92
/*\ %edi = dest + dow * dy + dx + dw \*/
93
movl dow, %eax
94
imull dy, %eax
95
addl dx, %eax
96
addl dw, %eax
97
movl dest, %edi
98
leal (%edi, %eax, 4), %edi
99
/*\ xp = xpoints + dxx + dw \*/
100
movl dxx, %ebx
101
addl dw, %ebx
102
movl xpoints, %eax
103
leal (%eax, %ebx, 4), %eax
104
movl %eax, xp
105
/*\ xap = xapoints + dxx + dw \*/
106
movl xapoints, %eax
107
leal (%eax, %ebx, 4), %eax
108
movl %eax, xap
109
/*\ y = dh \*/
110
movl dh, %eax
111
movl %eax, y
112
/*\ yp = ypoints + dyy \*/
113
movl dyy, %ebx
114
movl ypoints, %eax
115
leal (%eax, %ebx, 4), %eax
116
movl %eax, yp
117
/*\ yap = yapoints + dyy \*/
118
movl yapoints, %eax
119
leal (%eax, %ebx, 4), %eax
120
movl %eax, yap
121
122
pxor %mm7, %mm7
123
124
/*\ Test xup bit \*/
125
movl xup_yup, %eax
126
sarl $1, %eax
127
jnc .scale_x_down
128
129
.scale_x_up:
130
/*\ Test yup bit \*/
131
sarl $1, %eax
132
jnc .scale_x_up_y_down
133
134
135
/*\ Scaling up both ways \*/
136
137
.scale_x_up_y_up:
138
movl sow, %ebx
139
140
.up_up_loop_y:
141
142
/*\ x = -dw \*/
143
movl dw, %ecx
144
negl %ecx
145
146
/*\ %eax = *yap << 4 \*/
147
movl yap, %eax
148
movl (%eax), %eax
149
sall $4, %eax
150
jz .up_up_yap_0
151
movd %eax, %mm1
152
punpcklwd %mm1, %mm1
153
punpckldq %mm1, %mm1
154
155
.up_up_loop1_x:
156
/*\ %esi = *yp + xp[x] \*/
157
movl yp, %eax
158
movl (%eax), %esi
159
movl xp, %eax
160
movl (%eax, %ecx, 4), %eax
161
leal (%esi, %eax, 4), %esi
162
163
/*\ %eax = xap[x] << 4 \*/
164
movl xap, %eax
165
movl (%eax, %ecx, 4), %eax
166
sall $4, %eax
167
jz .up_up_xap_0
168
169
/*\ %mm0 = xap[x] << 4 \*/
170
movd %eax, %mm0
171
punpcklwd %mm0, %mm0
172
punpckldq %mm0, %mm0
173
174
/*\ Load and unpack four pixels in parralel
175
|*| %mm2 = ptr[0], %mm3 = ptr[1]
176
|*| %mm4 = ptr[sow], %mm5 = ptr[sow + 1]
177
\*/
178
movq (%esi), %mm2
179
movq (%esi, %ebx, 4), %mm4
180
movq %mm2, %mm3
181
movq %mm4, %mm5
182
punpcklbw %mm7, %mm2
183
punpcklbw %mm7, %mm4
184
punpckhbw %mm7, %mm3
185
punpckhbw %mm7, %mm5
186
187
/*\ X interpolation: r = l + (r - l) * xap \*/
188
psubw %mm2, %mm3
189
psubw %mm4, %mm5
190
psllw $4, %mm3
191
psllw $4, %mm5
192
pmulhw %mm0, %mm3
193
pmulhw %mm0, %mm5
194
paddw %mm2, %mm3
195
paddw %mm4, %mm5
196
/*\ Now %mm3 = I(ptr[0], ptr[1]), %mm5 = I(ptr[sow], ptr[sow + 1]) \*/
197
jmp .up_up_common
198
.up_up_xap_0:
199
/*\ Load and unpack two pixels
200
|*| %mm3 = ptr[0], %mm5 = ptr[sow]
201
\*/
202
movd (%esi), %mm3
203
movd (%esi, %ebx, 4), %mm5
204
punpcklbw %mm7, %mm3
205
punpcklbw %mm7, %mm5
206
.up_up_common:
207
/*\ Y interpolation: d = u + (d - u) * yap \*/
208
psubw %mm3, %mm5
209
psllw $4, %mm5
210
pmulhw %mm1, %mm5
211
paddw %mm3, %mm5
212
packuswb %mm5, %mm5
213
movd %mm5, (%edi, %ecx, 4)
214
215
/*\ while (++x) \*/
216
incl %ecx
217
jnz .up_up_loop1_x
218
jmp .up_up_yap_end
219
.up_up_yap_0:
220
221
.up_up_loop2_x:
222
/*\ %esi = *yp + xp[x] \*/
223
movl yp, %eax
224
movl (%eax), %esi
225
movl xp, %eax
226
movl (%eax, %ecx, 4), %eax
227
leal (%esi, %eax, 4), %esi
228
229
/*\ %eax = xap[x] << 4 \*/
230
movl xap, %eax
231
movl (%eax, %ecx, 4), %eax
232
sall $4, %eax
233
jz .up_up_0
234
235
/*\ %mm0 = xap[x] << 4 \*/
236
movd %eax, %mm0
237
punpcklwd %mm0, %mm0
238
punpckldq %mm0, %mm0
239
240
/*\ Load and unpack two pixels in parralel
241
|*| %mm2 = ptr[0], %mm3 = ptr[1]
242
\*/
243
movq (%esi), %mm2
244
movq %mm2, %mm3
245
punpcklbw %mm7, %mm2
246
punpckhbw %mm7, %mm3
247
248
/*\ X interpolation: r = l + (r - l) * xap \*/
249
psubw %mm2, %mm3
250
psllw $4, %mm3
251
pmulhw %mm0, %mm3
252
paddw %mm2, %mm3
253
packuswb %mm3, %mm3
254
movd %mm3, (%edi, %ecx, 4)
255
jmp .up_up_1
256
.up_up_0:
257
/*\ dptr[x] = *sptr \*/
258
movl (%esi), %eax
259
movl %eax, (%edi, %ecx, 4)
260
.up_up_1:
261
incl %ecx
262
jnz .up_up_loop2_x
263
264
.up_up_yap_end:
265
/*\ dptr += dow \*/
266
movl dow, %eax
267
leal (%edi, %eax, 4), %edi
268
/*\ yap++; yp++ \*/
269
addl $4, yap
270
addl $4, yp
271
/*\ while (y--) \*/
272
decl y
273
jnz .up_up_loop_y
274
275
jmp .scale_leave
276
277
278
/*\ Scaling down vertically \*/
279
280
.scale_x_up_y_down:
281
/*\ sow_4 = sow * 4 \*/
282
movl sow, %eax
283
sall $2, %eax
284
movl %eax, sow_4
285
286
.up_down_loop_y:
287
288
/*\ Setup My and Cy \*/
289
movl yap, %eax
290
movzwl (%eax), %ebx
291
movl %ebx, My
292
movzwl 2(%eax), %eax
293
movl %eax, Cy
294
295
/*\ mm4 = Cy \*/
296
movd %eax, %mm4
297
punpcklwd %mm4, %mm4
298
punpckldq %mm4, %mm4
299
/*\ mm5 = My \*/
300
movd %ebx, %mm5
301
punpcklwd %mm5, %mm5
302
punpckldq %mm5, %mm5
303
304
/*\ x = -dw \*/
305
movl dw, %ecx
306
negl %ecx
307
.up_down_loop_x:
308
/*\ %esi = *yp + xp[x] \*/
309
movl yp, %eax
310
movl (%eax), %esi
311
movl xp, %eax
312
movl (%eax, %ecx, 4), %eax
313
leal (%esi, %eax, 4), %esi
314
315
movl %esi, %eax
316
/*\ v = (*p * My) >> 10 \*/
317
movd (%eax), %mm0
318
punpcklbw %mm7, %mm0
319
psllw $6, %mm0
320
pmulhw %mm5, %mm0
321
322
/*\ i = 0x4000 - My \*/
323
movl $0x4000, %ebx
324
subl My, %ebx
325
jbe 5f
326
jmp 2f
327
1:
328
/*\ p += sow; v += (*p * Cy) >> 10 \*/
329
addl sow_4, %eax
330
movd (%eax), %mm1
331
punpcklbw %mm7, %mm1
332
psllw $6, %mm1
333
pmulhw %mm4, %mm1
334
paddw %mm1, %mm0
335
336
/*\ i -= Cy; while (i > Cy) \*/
337
subl Cy, %ebx
338
2:
339
cmpl Cy, %ebx
340
jg 1b
341
342
/*\ mm6 = i \*/
343
movd %ebx, %mm6
344
punpcklwd %mm6, %mm6
345
punpckldq %mm6, %mm6
346
347
/*\ p += sow; v += (*p * i) >> 10 \*/
348
addl sow_4, %eax
349
movd (%eax), %mm1
350
punpcklbw %mm7, %mm1
351
psllw $6, %mm1
352
pmulhw %mm6, %mm1
353
paddw %mm1, %mm0
354
5:
355
/*\ %eax = xap[x] << 5 \*/
356
movl xap, %eax
357
movl (%eax, %ecx, 4), %eax
358
sall $5, %eax
359
jz 6f
360
/*\ mm3 = xap[x] << 5 \*/
361
movd %eax, %mm3
362
punpcklwd %mm3, %mm3
363
punpckldq %mm3, %mm3
364
365
/*\ p + 1 \*/
366
movl %esi, %eax
367
addl $4, %eax
368
/*\ vv = (*p * My) >> 10 \*/
369
movd (%eax), %mm2
370
punpcklbw %mm7, %mm2
371
psllw $6, %mm2
372
pmulhw %mm5, %mm2
373
374
/*\ i = 0x4000 - My \*/
375
movl $0x4000, %ebx
376
subl My, %ebx
377
jbe 5f
378
jmp 2f
379
1:
380
/*\ p += sow; vv += (*p * Cy) >> 10 \*/
381
addl sow_4, %eax
382
movd (%eax), %mm1
383
punpcklbw %mm7, %mm1
384
psllw $6, %mm1
385
pmulhw %mm4, %mm1
386
paddw %mm1, %mm2
387
388
/*\ i -= Cy; while (i > Cy) \*/
389
subl Cy, %ebx
390
2:
391
cmpl Cy, %ebx
392
jg 1b
393
394
/*\ p += sow; v += (*p * i) >> 10 \*/
395
addl sow_4, %eax
396
movd (%eax), %mm1
397
punpcklbw %mm7, %mm1
398
psllw $6, %mm1
399
pmulhw %mm6, %mm1
400
paddw %mm1, %mm2
401
5:
402
/*\ v = v + (vv - v) * xap \*/
403
psubw %mm0, %mm2
404
psllw $3, %mm2
405
pmulhw %mm3, %mm2
406
paddw %mm2, %mm0
407
6:
408
/*\ dest[x] = v >> 4 \*/
409
psrlw $4, %mm0
410
packuswb %mm0, %mm0
411
movd %mm0, (%edi, %ecx, 4)
412
413
/*\ while (++x) \*/
414
incl %ecx
415
jnz .up_down_loop_x
416
417
/*\ dptr += dow \*/
418
movl dow, %eax
419
leal (%edi, %eax, 4), %edi
420
/*\ yap++; yp++ \*/
421
addl $4, yap
422
addl $4, yp
423
/*\ while (y--) \*/
424
decl y
425
jnz .up_down_loop_y
426
427
jmp .scale_leave
428
429
.scale_x_down:
430
/*\ Test yup bit \*/
431
sarl $1, %eax
432
jnc .scale_x_down_y_down
433
434
435
/*\ Scaling down horizontally \*/
436
437
.scale_x_down_y_up:
438
/*\ sow_4 = sow * 4 \*/
439
movl sow, %eax
440
sall $2, %eax
441
movl %eax, sow_4
442
443
.down_up_loop_y:
444
445
/*\ %eax = *yap << 5 \*/
446
movl yap, %eax
447
movl (%eax), %eax
448
sall $5, %eax
449
/*\ mm3 = *yap << 5 \*/
450
movd %eax, %mm3
451
punpcklwd %mm3, %mm3
452
punpckldq %mm3, %mm3
453
454
/*\ x = -dw \*/
455
movl dw, %ecx
456
negl %ecx
457
.down_up_loop_x:
458
/*\ %esi = *yp + xp[x] \*/
459
movl yp, %eax
460
movl (%eax), %esi
461
movl xp, %eax
462
movl (%eax, %ecx, 4), %eax
463
leal (%esi, %eax, 4), %esi
464
465
/*\ Setup Mx and Cx \*/
466
movl xap, %eax
467
movzwl (%eax, %ecx, 4), %ebx
468
movl %ebx, Mx
469
movzwl 2(%eax, %ecx, 4), %eax
470
movl %eax, Cx
471
472
/*\ mm4 = Cx \*/
473
movd %eax, %mm4
474
punpcklwd %mm4, %mm4
475
punpckldq %mm4, %mm4
476
/*\ mm5 = Mx \*/
477
movd %ebx, %mm5
478
punpcklwd %mm5, %mm5
479
punpckldq %mm5, %mm5
480
481
movl %esi, %eax
482
/*\ v = (*p * Mx) >> 10 \*/
483
movd (%eax), %mm0
484
punpcklbw %mm7, %mm0
485
psllw $6, %mm0
486
pmulhw %mm5, %mm0
487
488
/*\ i = 0x4000 - Mx \*/
489
movl $0x4000, %ebx
490
subl Mx, %ebx
491
jbe 5f
492
jmp 2f
493
1:
494
/*\ p += sow; v += (*p * Cx) >> 10 \*/
495
addl $4, %eax
496
movd (%eax), %mm1
497
punpcklbw %mm7, %mm1
498
psllw $6, %mm1
499
pmulhw %mm4, %mm1
500
paddw %mm1, %mm0
501
502
/*\ i -= Cx; while (i > Cx) \*/
503
subl Cx, %ebx
504
2:
505
cmpl Cx, %ebx
506
jg 1b
507
508
/*\ mm6 = i \*/
509
movd %ebx, %mm6
510
punpcklwd %mm6, %mm6
511
punpckldq %mm6, %mm6
512
513
/*\ p += sow; v += (*p * i) >> 10 \*/
514
addl $4, %eax
515
movd (%eax), %mm1
516
punpcklbw %mm7, %mm1
517
psllw $6, %mm1
518
pmulhw %mm6, %mm1
519
paddw %mm1, %mm0
520
5:
521
movd %mm3, %eax
522
testl %eax, %eax
523
jz 6f
524
/*\ p + sow \*/
525
movl %esi, %eax
526
addl sow_4, %eax
527
/*\ vv = (*p * Mx) >> 10 \*/
528
movd (%eax), %mm2
529
punpcklbw %mm7, %mm2
530
psllw $6, %mm2
531
pmulhw %mm5, %mm2
532
533
/*\ i = 0x4000 - Mx \*/
534
movl $0x4000, %ebx
535
subl Mx, %ebx
536
jbe 5f
537
jmp 2f
538
1:
539
/*\ p += sow; vv += (*p * Cx) >> 10 \*/
540
addl $4, %eax
541
movd (%eax), %mm1
542
punpcklbw %mm7, %mm1
543
psllw $6, %mm1
544
pmulhw %mm4, %mm1
545
paddw %mm1, %mm2
546
547
/*\ i -= Cx; while (i > Cx) \*/
548
subl Cx, %ebx
549
2:
550
cmpl Cx, %ebx
551
jg 1b
552
553
/*\ p += sow; v += (*p * i) >> 10 \*/
554
addl $4, %eax
555
movd (%eax), %mm1
556
punpcklbw %mm7, %mm1
557
psllw $6, %mm1
558
pmulhw %mm6, %mm1
559
paddw %mm1, %mm2
560
5:
561
/*\ v = v + (vv - v) * yap \*/
562
psubw %mm0, %mm2
563
psllw $3, %mm2
564
pmulhw %mm3, %mm2
565
paddw %mm2, %mm0
566
6:
567
/*\ dest[x] = v >> 4 \*/
568
psrlw $4, %mm0
569
packuswb %mm0, %mm0
570
movd %mm0, (%edi, %ecx, 4)
571
572
/*\ while (++x) \*/
573
incl %ecx
574
jnz .down_up_loop_x
575
576
/*\ dptr += dow \*/
577
movl dow, %eax
578
leal (%edi, %eax, 4), %edi
579
/*\ yap++; yp++ \*/
580
addl $4, yap
581
addl $4, yp
582
/*\ while (y--) \*/
583
decl y
584
jnz .down_up_loop_y
585
586
jmp .scale_leave
587
588
589
/*\ Scaling down both ways \*/
590
591
.scale_x_down_y_down:
592
/*\ sow_4 = sow * 4 \*/
593
movl sow, %eax
594
sall $2, %eax
595
movl %eax, sow_4
596
597
.down_down_loop_y:
598
599
/*\ Setup My and Cy \*/
600
movl yap, %eax
601
movzwl (%eax), %ebx
602
movl %ebx, My
603
movzwl 2(%eax), %eax
604
movl %eax, Cy
605
606
/*\ x = -dw \*/
607
movl dw, %ecx
608
negl %ecx
609
.down_down_loop_x:
610
/*\ %esi = *yp + xp[x] \*/
611
movl yp, %eax
612
movl (%eax), %esi
613
movl xp, %eax
614
movl (%eax, %ecx, 4), %eax
615
leal (%esi, %eax, 4), %esi
616
617
/*\ Setup Mx and Cx \*/
618
movl xap, %eax
619
movzwl (%eax, %ecx, 4), %ebx
620
movl %ebx, Mx
621
movzwl 2(%eax, %ecx, 4), %eax
622
movl %eax, Cx
623
624
/*\ mm3 = Cx \*/
625
movd %eax, %mm3
626
punpcklwd %mm3, %mm3
627
punpckldq %mm3, %mm3
628
/*\ mm5 = Mx \*/
629
movd %ebx, %mm5
630
punpcklwd %mm5, %mm5
631
punpckldq %mm5, %mm5
632
633
/*\ p = sptr; v = (*p * Mx) >> 9 \*/
634
movl %esi, %eax
635
movd (%eax), %mm0
636
punpcklbw %mm7, %mm0
637
psllw $7, %mm0
638
pmulhw %mm5, %mm0
639
640
/*\ i = 0x4000 - Mx \*/
641
movl $0x4000, %ebx
642
subl Mx, %ebx
643
jbe 5f
644
jmp 2f
645
1:
646
/*\ v += (*++p * Cx) >> 9 \*/
647
addl $4, %eax
648
movd (%eax), %mm1
649
punpcklbw %mm7, %mm1
650
psllw $7, %mm1
651
pmulhw %mm3, %mm1
652
paddw %mm1, %mm0
653
654
/*\ i -= Cx; while (i > Cx) \*/
655
subl Cx, %ebx
656
2:
657
cmpl Cx, %ebx
658
jg 1b
659
660
/*\ mm6 = i \*/
661
movd %ebx, %mm6
662
punpcklwd %mm6, %mm6
663
punpckldq %mm6, %mm6
664
665
/*\ v += (*++p * i) >> 9 \*/
666
addl $4, %eax
667
movd (%eax), %mm1
668
punpcklbw %mm7, %mm1
669
psllw $7, %mm1
670
pmulhw %mm6, %mm1
671
paddw %mm1, %mm0
672
5:
673
/*\ v *= My \*/
674
movd My, %mm4
675
punpcklwd %mm4, %mm4
676
punpckldq %mm4, %mm4
677
psllw $2, %mm0
678
pmulhw %mm4, %mm0
679
680
/*\ j = 0x4000 - My \*/
681
movl $0x4000, %edx
682
subl My, %edx
683
jbe 6f
684
jmp 4f
685
3:
686
/*\ sptr += sow; p = sptr \*/
687
addl sow_4, %esi
688
movl %esi, %eax
689
/*\ vx = (*p * Mx) >> 9 \*/
690
movd (%eax), %mm1
691
punpcklbw %mm7, %mm1
692
psllw $7, %mm1
693
pmulhw %mm5, %mm1
694
695
/*\ i = 0x4000 - Mx \*/
696
movl $0x4000, %ebx
697
subl Mx, %ebx
698
jbe 5f
699
jmp 2f
700
1:
701
/*\ vx += (*++p * Cx) >> 9 \*/
702
addl $4, %eax
703
movd (%eax), %mm2
704
punpcklbw %mm7, %mm2
705
psllw $7, %mm2
706
pmulhw %mm3, %mm2
707
paddw %mm2, %mm1
708
709
/*\ i -= Cx; while (i > Cx) \*/
710
subl Cx, %ebx
711
2:
712
cmpl Cx, %ebx
713
jg 1b
714
715
/*\ vx += (*++p * i) >> 9 \*/
716
addl $4, %eax
717
movd (%eax), %mm2
718
punpcklbw %mm7, %mm2
719
psllw $7, %mm2
720
pmulhw %mm6, %mm2
721
paddw %mm2, %mm1
722
5:
723
/*\ v += (vx * Cy) >> 14 \*/
724
movd Cy, %mm4
725
punpcklwd %mm4, %mm4
726
punpckldq %mm4, %mm4
727
psllw $2, %mm1
728
pmulhw %mm4, %mm1
729
paddw %mm1, %mm0
730
731
/*\ j -= Cy; while (j > Cy) \*/
732
subl Cy, %edx
733
4:
734
cmpl Cy, %edx
735
jg 3b
736
737
/*\ sptr += sow; p = sptr \*/
738
addl sow_4, %esi
739
movl %esi, %eax
740
/*\ vx = (*p * Mx) >> 9 \*/
741
movd (%eax), %mm1
742
punpcklbw %mm7, %mm1
743
psllw $7, %mm1
744
pmulhw %mm5, %mm1
745
746
/*\ i = 0x4000 - Mx \*/
747
movl $0x4000, %ebx
748
subl Mx, %ebx
749
jbe 5f
750
jmp 2f
751
1:
752
/*\ vx += (*++p * Cx) >> 9 \*/
753
addl $4, %eax
754
movd (%eax), %mm2
755
punpcklbw %mm7, %mm2
756
psllw $7, %mm2
757
pmulhw %mm3, %mm2
758
paddw %mm2, %mm1
759
760
/*\ i -= Cx; while (i > Cx) \*/
761
subl Cx, %ebx
762
2:
763
cmpl Cx, %ebx
764
jg 1b
765
766
/*\ vx += (*++p * i) >> 9 \*/
767
addl $4, %eax
768
movd (%eax), %mm2
769
punpcklbw %mm7, %mm2
770
psllw $7, %mm2
771
pmulhw %mm6, %mm2
772
paddw %mm2, %mm1
773
5:
774
/*\ v += (vx * j) >> 14 \*/
775
movd %edx, %mm4
776
punpcklwd %mm4, %mm4
777
punpckldq %mm4, %mm4
778
psllw $2, %mm1
779
pmulhw %mm4, %mm1
780
paddw %mm1, %mm0
781
6:
782
/*\ dptr[x] = mm0 >> 5 \*/
783
psrlw $5, %mm0
784
packuswb %mm0, %mm0
785
movd %mm0, (%edi, %ecx, 4)
786
787
/*\ while (++x) \*/
788
incl %ecx
789
jnz .down_down_loop_x
790
791
/*\ dptr += dow \*/
792
movl dow, %eax
793
leal (%edi, %eax, 4), %edi
794
/*\ yap++; yp++ \*/
795
addl $4, yap
796
addl $4, yp
797
/*\ while (y--) \*/
798
decl y
799
jnz .down_down_loop_y
800
801
jmp .scale_leave
802
803
.scale_leave:
804
emms
805
popl %esi
806
popl %edi
807
popl %edx
808
popl %ecx
809
popl %ebx
810
movl %ebp, %esp
811
popl %ebp
812
ret
813
814
SIZE(qimageScale_mmx_AARGBA)
815
816
#endif
Loggerhead is a web-based interface for Breezy
Version: 2.0.1