This repository has been archived by the owner on Feb 11, 2021. It is now read-only.
/
SDL_blit_A.c
2300 lines (2060 loc) · 73.9 KB
1
2
/*
SDL - Simple DirectMedia Layer
3
Copyright (C) 1997-2006 Sam Lantinga
4
5
This library is free software; you can redistribute it and/or
6
modify it under the terms of the GNU Lesser General Public
7
License as published by the Free Software Foundation; either
8
version 2.1 of the License, or (at your option) any later version.
9
10
11
12
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13
Lesser General Public License for more details.
14
15
16
17
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
18
19
Sam Lantinga
20
slouken@libsdl.org
21
*/
22
#include "SDL_config.h"
23
24
25
26
27
28
29
#include "SDL_video.h"
#include "SDL_blit.h"
/* Functions to perform alpha blended blitting */
/* N->1 blending with per-surface alpha */
30
31
static void
BlitNto1SurfaceAlpha(SDL_BlitInfo * info)
32
{
33
34
35
int width = info->dst_w;
int height = info->dst_h;
Uint8 *src = info->src;
36
int srcskip = info->src_skip;
37
Uint8 *dst = info->dst;
38
int dstskip = info->dst_skip;
39
Uint8 *palmap = info->table;
40
41
SDL_PixelFormat *srcfmt = info->src_fmt;
SDL_PixelFormat *dstfmt = info->dst_fmt;
42
43
int srcbpp = srcfmt->BytesPerPixel;
44
const unsigned A = info->a;
45
46
47
while (height--) {
/* *INDENT-OFF* */
48
49
DUFFS_LOOP4(
{
50
Uint32 Pixel;
51
52
53
54
55
56
unsigned sR;
unsigned sG;
unsigned sB;
unsigned dR;
unsigned dG;
unsigned dB;
57
DISEMBLE_RGB(src, srcbpp, srcfmt, Pixel, sR, sG, sB);
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
dR = dstfmt->palette->colors[*dst].r;
dG = dstfmt->palette->colors[*dst].g;
dB = dstfmt->palette->colors[*dst].b;
ALPHA_BLEND(sR, sG, sB, A, dR, dG, dB);
dR &= 0xff;
dG &= 0xff;
dB &= 0xff;
/* Pack RGB into 8bit pixel */
if ( palmap == NULL ) {
*dst =((dR>>5)<<(3+2))|
((dG>>5)<<(2))|
((dB>>6)<<(0));
} else {
*dst = palmap[((dR>>5)<<(3+2))|
((dG>>5)<<(2)) |
((dB>>6)<<(0))];
}
dst++;
src += srcbpp;
},
width);
79
80
81
82
/* *INDENT-ON* */
src += srcskip;
dst += dstskip;
}
83
84
85
}
/* N->1 blending with pixel alpha */
86
87
static void
BlitNto1PixelAlpha(SDL_BlitInfo * info)
88
{
89
90
91
int width = info->dst_w;
int height = info->dst_h;
Uint8 *src = info->src;
92
int srcskip = info->src_skip;
93
Uint8 *dst = info->dst;
94
int dstskip = info->dst_skip;
95
Uint8 *palmap = info->table;
96
97
SDL_PixelFormat *srcfmt = info->src_fmt;
SDL_PixelFormat *dstfmt = info->dst_fmt;
98
99
100
101
102
int srcbpp = srcfmt->BytesPerPixel;
/* FIXME: fix alpha bit field expansion here too? */
while (height--) {
/* *INDENT-OFF* */
103
104
DUFFS_LOOP4(
{
105
Uint32 Pixel;
106
107
108
109
110
111
112
unsigned sR;
unsigned sG;
unsigned sB;
unsigned sA;
unsigned dR;
unsigned dG;
unsigned dB;
113
DISEMBLE_RGBA(src,srcbpp,srcfmt,Pixel,sR,sG,sB,sA);
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
dR = dstfmt->palette->colors[*dst].r;
dG = dstfmt->palette->colors[*dst].g;
dB = dstfmt->palette->colors[*dst].b;
ALPHA_BLEND(sR, sG, sB, sA, dR, dG, dB);
dR &= 0xff;
dG &= 0xff;
dB &= 0xff;
/* Pack RGB into 8bit pixel */
if ( palmap == NULL ) {
*dst =((dR>>5)<<(3+2))|
((dG>>5)<<(2))|
((dB>>6)<<(0));
} else {
*dst = palmap[((dR>>5)<<(3+2))|
((dG>>5)<<(2)) |
((dB>>6)<<(0)) ];
}
dst++;
src += srcbpp;
},
width);
135
136
137
138
/* *INDENT-ON* */
src += srcskip;
dst += dstskip;
}
139
140
141
}
/* colorkeyed N->1 blending with per-surface alpha */
142
143
static void
BlitNto1SurfaceAlphaKey(SDL_BlitInfo * info)
144
{
145
146
147
int width = info->dst_w;
int height = info->dst_h;
Uint8 *src = info->src;
148
int srcskip = info->src_skip;
149
Uint8 *dst = info->dst;
150
int dstskip = info->dst_skip;
151
Uint8 *palmap = info->table;
152
153
SDL_PixelFormat *srcfmt = info->src_fmt;
SDL_PixelFormat *dstfmt = info->dst_fmt;
154
int srcbpp = srcfmt->BytesPerPixel;
155
Uint32 ckey = info->colorkey;
156
157
const int A = info->a;
158
159
160
while (height--) {
/* *INDENT-OFF* */
161
162
DUFFS_LOOP(
{
163
Uint32 Pixel;
164
165
166
167
168
169
unsigned sR;
unsigned sG;
unsigned sB;
unsigned dR;
unsigned dG;
unsigned dB;
170
171
DISEMBLE_RGB(src, srcbpp, srcfmt, Pixel, sR, sG, sB);
if ( Pixel != ckey ) {
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
dR = dstfmt->palette->colors[*dst].r;
dG = dstfmt->palette->colors[*dst].g;
dB = dstfmt->palette->colors[*dst].b;
ALPHA_BLEND(sR, sG, sB, A, dR, dG, dB);
dR &= 0xff;
dG &= 0xff;
dB &= 0xff;
/* Pack RGB into 8bit pixel */
if ( palmap == NULL ) {
*dst =((dR>>5)<<(3+2))|
((dG>>5)<<(2)) |
((dB>>6)<<(0));
} else {
*dst = palmap[((dR>>5)<<(3+2))|
((dG>>5)<<(2)) |
((dB>>6)<<(0)) ];
}
}
dst++;
src += srcbpp;
},
width);
194
195
196
197
/* *INDENT-ON* */
src += srcskip;
dst += dstskip;
}
198
199
}
200
#ifdef __MMX__
201
202
/* fast RGB888->(A)RGB888 blending with surface alpha=128 special case */
203
204
static void
BlitRGBtoRGBSurfaceAlpha128MMX(SDL_BlitInfo * info)
205
{
206
207
208
int width = info->dst_w;
int height = info->dst_h;
Uint32 *srcp = (Uint32 *) info->src;
209
int srcskip = info->src_skip >> 2;
210
Uint32 *dstp = (Uint32 *) info->dst;
211
212
int dstskip = info->dst_skip >> 2;
Uint32 dalpha = info->dst_fmt->Amask;
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
__m64 src1, src2, dst1, dst2, lmask, hmask, dsta;
hmask = _mm_set_pi32(0x00fefefe, 0x00fefefe); /* alpha128 mask -> hmask */
lmask = _mm_set_pi32(0x00010101, 0x00010101); /* !alpha128 mask -> lmask */
dsta = _mm_set_pi32(dalpha, dalpha); /* dst alpha mask -> dsta */
while (height--) {
int n = width;
if (n & 1) {
Uint32 s = *srcp++;
Uint32 d = *dstp;
*dstp++ = ((((s & 0x00fefefe) + (d & 0x00fefefe)) >> 1)
+ (s & d & 0x00010101)) | dalpha;
n--;
}
for (n >>= 1; n > 0; --n) {
dst1 = *(__m64 *) dstp; /* 2 x dst -> dst1(ARGBARGB) */
dst2 = dst1; /* 2 x dst -> dst2(ARGBARGB) */
src1 = *(__m64 *) srcp; /* 2 x src -> src1(ARGBARGB) */
src2 = src1; /* 2 x src -> src2(ARGBARGB) */
dst2 = _mm_and_si64(dst2, hmask); /* dst & mask -> dst2 */
src2 = _mm_and_si64(src2, hmask); /* src & mask -> src2 */
src2 = _mm_add_pi32(src2, dst2); /* dst2 + src2 -> src2 */
src2 = _mm_srli_pi32(src2, 1); /* src2 >> 1 -> src2 */
dst1 = _mm_and_si64(dst1, src1); /* src & dst -> dst1 */
dst1 = _mm_and_si64(dst1, lmask); /* dst1 & !mask -> dst1 */
dst1 = _mm_add_pi32(dst1, src2); /* src2 + dst1 -> dst1 */
dst1 = _mm_or_si64(dst1, dsta); /* dsta(full alpha) | dst1 -> dst1 */
*(__m64 *) dstp = dst1; /* dst1 -> 2 x dst pixels */
dstp += 2;
srcp += 2;
}
srcp += srcskip;
dstp += dstskip;
}
_mm_empty();
256
257
258
}
/* fast RGB888->(A)RGB888 blending with surface alpha */
259
260
static void
BlitRGBtoRGBSurfaceAlphaMMX(SDL_BlitInfo * info)
261
{
262
SDL_PixelFormat *df = info->dst_fmt;
263
Uint32 chanmask = df->Rmask | df->Gmask | df->Bmask;
264
unsigned alpha = info->a;
265
266
267
268
269
if (alpha == 128 && (df->Rmask | df->Gmask | df->Bmask) == 0x00FFFFFF) {
/* only call a128 version when R,G,B occupy lower bits */
BlitRGBtoRGBSurfaceAlpha128MMX(info);
} else {
270
271
272
int width = info->dst_w;
int height = info->dst_h;
Uint32 *srcp = (Uint32 *) info->src;
273
int srcskip = info->src_skip >> 2;
274
Uint32 *dstp = (Uint32 *) info->dst;
275
int dstskip = info->dst_skip >> 2;
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
Uint32 dalpha = df->Amask;
Uint32 amult;
__m64 src1, src2, dst1, dst2, mm_alpha, mm_zero, dsta;
mm_zero = _mm_setzero_si64(); /* 0 -> mm_zero */
/* form the alpha mult */
amult = alpha | (alpha << 8);
amult = amult | (amult << 16);
chanmask =
(0xff << df->Rshift) | (0xff << df->Gshift) | (0xff << df->
Bshift);
mm_alpha = _mm_set_pi32(0, amult & chanmask); /* 0000AAAA -> mm_alpha, minus 1 chan */
mm_alpha = _mm_unpacklo_pi8(mm_alpha, mm_zero); /* 0A0A0A0A -> mm_alpha, minus 1 chan */
/* at this point mm_alpha can be 000A0A0A or 0A0A0A00 or another combo */
dsta = _mm_set_pi32(dalpha, dalpha); /* dst alpha mask -> dsta */
while (height--) {
int n = width;
if (n & 1) {
/* One Pixel Blend */
src2 = _mm_cvtsi32_si64(*srcp); /* src(ARGB) -> src2 (0000ARGB) */
src2 = _mm_unpacklo_pi8(src2, mm_zero); /* 0A0R0G0B -> src2 */
dst1 = _mm_cvtsi32_si64(*dstp); /* dst(ARGB) -> dst1 (0000ARGB) */
dst1 = _mm_unpacklo_pi8(dst1, mm_zero); /* 0A0R0G0B -> dst1 */
src2 = _mm_sub_pi16(src2, dst1); /* src2 - dst2 -> src2 */
src2 = _mm_mullo_pi16(src2, mm_alpha); /* src2 * alpha -> src2 */
src2 = _mm_srli_pi16(src2, 8); /* src2 >> 8 -> src2 */
dst1 = _mm_add_pi8(src2, dst1); /* src2 + dst1 -> dst1 */
dst1 = _mm_packs_pu16(dst1, mm_zero); /* 0000ARGB -> dst1 */
dst1 = _mm_or_si64(dst1, dsta); /* dsta | dst1 -> dst1 */
*dstp = _mm_cvtsi64_si32(dst1); /* dst1 -> pixel */
++srcp;
++dstp;
n--;
}
317
318
319
320
321
322
323
for (n >>= 1; n > 0; --n) {
/* Two Pixels Blend */
src1 = *(__m64 *) srcp; /* 2 x src -> src1(ARGBARGB) */
src2 = src1; /* 2 x src -> src2(ARGBARGB) */
src1 = _mm_unpacklo_pi8(src1, mm_zero); /* low - 0A0R0G0B -> src1 */
src2 = _mm_unpackhi_pi8(src2, mm_zero); /* high - 0A0R0G0B -> src2 */
324
325
326
327
328
dst1 = *(__m64 *) dstp; /* 2 x dst -> dst1(ARGBARGB) */
dst2 = dst1; /* 2 x dst -> dst2(ARGBARGB) */
dst1 = _mm_unpacklo_pi8(dst1, mm_zero); /* low - 0A0R0G0B -> dst1 */
dst2 = _mm_unpackhi_pi8(dst2, mm_zero); /* high - 0A0R0G0B -> dst2 */
329
330
331
332
333
src1 = _mm_sub_pi16(src1, dst1); /* src1 - dst1 -> src1 */
src1 = _mm_mullo_pi16(src1, mm_alpha); /* src1 * alpha -> src1 */
src1 = _mm_srli_pi16(src1, 8); /* src1 >> 8 -> src1 */
dst1 = _mm_add_pi8(src1, dst1); /* src1 + dst1(dst1) -> dst1 */
334
335
336
337
338
src2 = _mm_sub_pi16(src2, dst2); /* src2 - dst2 -> src2 */
src2 = _mm_mullo_pi16(src2, mm_alpha); /* src2 * alpha -> src2 */
src2 = _mm_srli_pi16(src2, 8); /* src2 >> 8 -> src2 */
dst2 = _mm_add_pi8(src2, dst2); /* src2 + dst2(dst2) -> dst2 */
339
340
341
342
343
344
345
346
347
348
349
350
351
352
dst1 = _mm_packs_pu16(dst1, dst2); /* 0A0R0G0B(res1), 0A0R0G0B(res2) -> dst1(ARGBARGB) */
dst1 = _mm_or_si64(dst1, dsta); /* dsta | dst1 -> dst1 */
*(__m64 *) dstp = dst1; /* dst1 -> 2 x pixel */
srcp += 2;
dstp += 2;
}
srcp += srcskip;
dstp += dstskip;
}
_mm_empty();
}
353
354
355
}
/* fast ARGB888->(A)RGB888 blending with pixel alpha */
356
357
static void
BlitRGBtoRGBPixelAlphaMMX(SDL_BlitInfo * info)
358
{
359
360
361
int width = info->dst_w;
int height = info->dst_h;
Uint32 *srcp = (Uint32 *) info->src;
362
int srcskip = info->src_skip >> 2;
363
Uint32 *dstp = (Uint32 *) info->dst;
364
365
int dstskip = info->dst_skip >> 2;
SDL_PixelFormat *sf = info->src_fmt;
366
367
368
369
370
371
372
373
Uint32 chanmask = sf->Rmask | sf->Gmask | sf->Bmask;
Uint32 amask = sf->Amask;
Uint32 ashift = sf->Ashift;
Uint64 multmask;
__m64 src1, dst1, mm_alpha, mm_zero, dmask;
mm_zero = _mm_setzero_si64(); /* 0 -> mm_zero */
374
multmask = 0xFFFF;
375
376
multmask <<= (ashift * 2);
multmask = ~multmask;
377
378
379
380
dmask = *(__m64 *) & multmask; /* dst alpha mask -> dmask */
while (height--) {
/* *INDENT-OFF* */
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
DUFFS_LOOP4({
Uint32 alpha = *srcp & amask;
if (alpha == 0) {
/* do nothing */
} else if (alpha == amask) {
/* opaque alpha -- copy RGB, keep dst alpha */
*dstp = (*srcp & chanmask) | (*dstp & ~chanmask);
} else {
src1 = _mm_cvtsi32_si64(*srcp); /* src(ARGB) -> src1 (0000ARGB)*/
src1 = _mm_unpacklo_pi8(src1, mm_zero); /* 0A0R0G0B -> src1 */
dst1 = _mm_cvtsi32_si64(*dstp); /* dst(ARGB) -> dst1 (0000ARGB)*/
dst1 = _mm_unpacklo_pi8(dst1, mm_zero); /* 0A0R0G0B -> dst1 */
mm_alpha = _mm_cvtsi32_si64(alpha); /* alpha -> mm_alpha (0000000A) */
mm_alpha = _mm_srli_si64(mm_alpha, ashift); /* mm_alpha >> ashift -> mm_alpha(0000000A) */
mm_alpha = _mm_unpacklo_pi16(mm_alpha, mm_alpha); /* 00000A0A -> mm_alpha */
mm_alpha = _mm_unpacklo_pi32(mm_alpha, mm_alpha); /* 0A0A0A0A -> mm_alpha */
mm_alpha = _mm_and_si64(mm_alpha, dmask); /* 000A0A0A -> mm_alpha, preserve dst alpha on add */
/* blend */
src1 = _mm_sub_pi16(src1, dst1);/* src1 - dst1 -> src1 */
src1 = _mm_mullo_pi16(src1, mm_alpha); /* (src1 - dst1) * alpha -> src1 */
src1 = _mm_srli_pi16(src1, 8); /* src1 >> 8 -> src1(000R0G0B) */
dst1 = _mm_add_pi8(src1, dst1); /* src1 + dst1 -> dst1(0A0R0G0B) */
dst1 = _mm_packs_pu16(dst1, mm_zero); /* 0000ARGB -> dst1 */
*dstp = _mm_cvtsi64_si32(dst1); /* dst1 -> pixel */
}
++srcp;
++dstp;
}, width);
413
414
415
416
417
/* *INDENT-ON* */
srcp += srcskip;
dstp += dstskip;
}
_mm_empty();
418
}
419
420
#endif /* __MMX__ */
421
422
#if SDL_ALTIVEC_BLITTERS
423
424
425
#if __MWERKS__
#pragma altivec_model on
#endif
426
#if HAVE_ALTIVEC_H
427
#include <altivec.h>
428
#endif
429
#include <assert.h>
430
431
#if (defined(__MACOSX__) && (__GNUC__ < 4))
432
#define VECUINT8_LITERAL(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p) \
433
(vector unsigned char) ( a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p )
434
#define VECUINT16_LITERAL(a,b,c,d,e,f,g,h) \
435
436
(vector unsigned short) ( a,b,c,d,e,f,g,h )
#else
437
#define VECUINT8_LITERAL(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p) \
438
(vector unsigned char) { a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p }
439
#define VECUINT16_LITERAL(a,b,c,d,e,f,g,h) \
440
441
442
(vector unsigned short) { a,b,c,d,e,f,g,h }
#endif
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
#define UNALIGNED_PTR(x) (((size_t) x) & 0x0000000F)
#define VECPRINT(msg, v) do { \
vector unsigned int tmpvec = (vector unsigned int)(v); \
unsigned int *vp = (unsigned int *)&tmpvec; \
printf("%s = %08X %08X %08X %08X\n", msg, vp[0], vp[1], vp[2], vp[3]); \
} while (0)
/* the permuation vector that takes the high bytes out of all the appropriate shorts
(vector unsigned char)(
0x00, 0x10, 0x02, 0x12,
0x04, 0x14, 0x06, 0x16,
0x08, 0x18, 0x0A, 0x1A,
0x0C, 0x1C, 0x0E, 0x1E );
*/
#define VEC_MERGE_PERMUTE() (vec_add(vec_lvsl(0, (int*)NULL), (vector unsigned char)vec_splat_u16(0x0F)))
#define VEC_U32_24() (vec_add(vec_splat_u32(12), vec_splat_u32(12)))
#define VEC_ALPHA_MASK() ((vector unsigned char)vec_sl((vector unsigned int)vec_splat_s8(-1), VEC_U32_24()))
#define VEC_ALIGNER(src) ((UNALIGNED_PTR(src)) \
? vec_lvsl(0, src) \
: vec_add(vec_lvsl(8, src), vec_splat_u8(8)))
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
#define VEC_MULTIPLY_ALPHA(vs, vd, valpha, mergePermute, v1_16, v8_16) do { \
/* vtemp1 contains source AAGGAAGGAAGGAAGG */ \
vector unsigned short vtemp1 = vec_mule(vs, valpha); \
/* vtemp2 contains source RRBBRRBBRRBBRRBB */ \
vector unsigned short vtemp2 = vec_mulo(vs, valpha); \
/* valpha2 is 255-alpha */ \
vector unsigned char valpha2 = vec_nor(valpha, valpha); \
/* vtemp3 contains dest AAGGAAGGAAGGAAGG */ \
vector unsigned short vtemp3 = vec_mule(vd, valpha2); \
/* vtemp4 contains dest RRBBRRBBRRBBRRBB */ \
vector unsigned short vtemp4 = vec_mulo(vd, valpha2); \
/* add source and dest */ \
vtemp1 = vec_add(vtemp1, vtemp3); \
vtemp2 = vec_add(vtemp2, vtemp4); \
/* vtemp1 = (vtemp1 + 1) + ((vtemp1 + 1) >> 8) */ \
vtemp1 = vec_add(vtemp1, v1_16); \
vtemp3 = vec_sr(vtemp1, v8_16); \
vtemp1 = vec_add(vtemp1, vtemp3); \
/* vtemp2 = (vtemp2 + 1) + ((vtemp2 + 1) >> 8) */ \
vtemp2 = vec_add(vtemp2, v1_16); \
vtemp4 = vec_sr(vtemp2, v8_16); \
vtemp2 = vec_add(vtemp2, vtemp4); \
/* (>>8) and get ARGBARGBARGBARGB */ \
vd = (vector unsigned char)vec_perm(vtemp1, vtemp2, mergePermute); \
} while (0)
490
491
/* Calculate the permute vector used for 32->32 swizzling */
492
493
static vector unsigned char
calc_swizzle32(const SDL_PixelFormat * srcfmt, const SDL_PixelFormat * dstfmt)
494
495
496
497
498
499
500
501
502
503
504
505
{
/*
* We have to assume that the bits that aren't used by other
* colors is alpha, and it's one complete byte, since some formats
* leave alpha with a zero mask, but we should still swizzle the bits.
*/
/* ARGB */
const static struct SDL_PixelFormat default_pixel_format = {
NULL, 0, 0,
0, 0, 0, 0,
16, 8, 0, 24,
0x00FF0000, 0x0000FF00, 0x000000FF, 0xFF000000,
506
507
0, 0
};
508
509
510
511
512
513
if (!srcfmt) {
srcfmt = &default_pixel_format;
}
if (!dstfmt) {
dstfmt = &default_pixel_format;
}
514
515
516
517
518
const vector unsigned char plus = VECUINT8_LITERAL(0x00, 0x00, 0x00, 0x00,
0x04, 0x04, 0x04, 0x04,
0x08, 0x08, 0x08, 0x08,
0x0C, 0x0C, 0x0C,
0x0C);
519
520
521
522
523
524
525
526
527
vector unsigned char vswiz;
vector unsigned int srcvec;
#define RESHIFT(X) (3 - ((X) >> 3))
Uint32 rmask = RESHIFT(srcfmt->Rshift) << (dstfmt->Rshift);
Uint32 gmask = RESHIFT(srcfmt->Gshift) << (dstfmt->Gshift);
Uint32 bmask = RESHIFT(srcfmt->Bshift) << (dstfmt->Bshift);
Uint32 amask;
/* Use zero for alpha if either surface doesn't have alpha */
if (dstfmt->Amask) {
528
529
530
amask =
((srcfmt->Amask) ? RESHIFT(srcfmt->Ashift) : 0x10) << (dstfmt->
Ashift);
531
} else {
532
533
534
amask =
0x10101010 & ((dstfmt->Rmask | dstfmt->Gmask | dstfmt->Bmask) ^
0xFFFFFFFF);
535
}
536
537
538
539
#undef RESHIFT
((unsigned int *) (char *) &srcvec)[0] = (rmask | gmask | bmask | amask);
vswiz = vec_add(plus, (vector unsigned char) vec_splat(srcvec, 0));
return (vswiz);
540
541
}
542
543
static void
Blit32to565PixelAlphaAltivec(SDL_BlitInfo * info)
544
{
545
546
int height = info->dst_h;
Uint8 *src = (Uint8 *) info->src;
547
int srcskip = info->src_skip;
548
Uint8 *dst = (Uint8 *) info->dst;
549
550
int dstskip = info->dst_skip;
SDL_PixelFormat *srcfmt = info->src_fmt;
551
552
553
554
555
556
557
558
vector unsigned char v0 = vec_splat_u8(0);
vector unsigned short v8_16 = vec_splat_u16(8);
vector unsigned short v1_16 = vec_splat_u16(1);
vector unsigned short v2_16 = vec_splat_u16(2);
vector unsigned short v3_16 = vec_splat_u16(3);
vector unsigned int v8_32 = vec_splat_u32(8);
vector unsigned int v16_32 = vec_add(v8_32, v8_32);
559
560
561
562
563
564
vector unsigned short v3f =
VECUINT16_LITERAL(0x003f, 0x003f, 0x003f, 0x003f,
0x003f, 0x003f, 0x003f, 0x003f);
vector unsigned short vfc =
VECUINT16_LITERAL(0x00fc, 0x00fc, 0x00fc, 0x00fc,
0x00fc, 0x00fc, 0x00fc, 0x00fc);
565
566
/*
567
568
569
570
571
572
573
574
575
576
577
578
579
0x10 - 0x1f is the alpha
0x00 - 0x0e evens are the red
0x01 - 0x0f odds are zero
*/
vector unsigned char vredalpha1 = VECUINT8_LITERAL(0x10, 0x00, 0x01, 0x01,
0x10, 0x02, 0x01, 0x01,
0x10, 0x04, 0x01, 0x01,
0x10, 0x06, 0x01,
0x01);
vector unsigned char vredalpha2 =
(vector unsigned char) (vec_add((vector unsigned int) vredalpha1,
vec_sl(v8_32, v16_32))
);
580
/*
581
582
583
584
585
586
587
588
589
590
0x00 - 0x0f is ARxx ARxx ARxx ARxx
0x11 - 0x0f odds are blue
*/
vector unsigned char vblue1 = VECUINT8_LITERAL(0x00, 0x01, 0x02, 0x11,
0x04, 0x05, 0x06, 0x13,
0x08, 0x09, 0x0a, 0x15,
0x0c, 0x0d, 0x0e, 0x17);
vector unsigned char vblue2 =
(vector unsigned char) (vec_add((vector unsigned int) vblue1, v8_32)
);
591
/*
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
0x00 - 0x0f is ARxB ARxB ARxB ARxB
0x10 - 0x0e evens are green
*/
vector unsigned char vgreen1 = VECUINT8_LITERAL(0x00, 0x01, 0x10, 0x03,
0x04, 0x05, 0x12, 0x07,
0x08, 0x09, 0x14, 0x0b,
0x0c, 0x0d, 0x16, 0x0f);
vector unsigned char vgreen2 =
(vector unsigned
char) (vec_add((vector unsigned int) vgreen1, vec_sl(v8_32, v8_32))
);
vector unsigned char vgmerge = VECUINT8_LITERAL(0x00, 0x02, 0x00, 0x06,
0x00, 0x0a, 0x00, 0x0e,
0x00, 0x12, 0x00, 0x16,
0x00, 0x1a, 0x00, 0x1e);
607
608
vector unsigned char mergePermute = VEC_MERGE_PERMUTE();
vector unsigned char vpermute = calc_swizzle32(srcfmt, NULL);
609
610
vector unsigned char valphaPermute =
vec_and(vec_lvsl(0, (int *) NULL), vec_splat_u8(0xC));
611
612
vector unsigned short vf800 = (vector unsigned short) vec_splat_u8(-7);
613
614
vf800 = vec_sl(vf800, vec_splat_u16(8));
615
while (height--) {
616
617
618
619
int extrawidth;
vector unsigned char valigner;
vector unsigned char vsrc;
vector unsigned char voverflow;
620
int width = info->dst_w;
621
622
623
#define ONE_PIXEL_BLEND(condition, widthvar) \
while (condition) { \
624
Uint32 Pixel; \
625
unsigned sR, sG, sB, dR, dG, dB, sA; \
626
DISEMBLE_RGBA(src, 4, srcfmt, Pixel, sR, sG, sB, sA); \
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
if(sA) { \
unsigned short dstpixel = *((unsigned short *)dst); \
dR = (dstpixel >> 8) & 0xf8; \
dG = (dstpixel >> 3) & 0xfc; \
dB = (dstpixel << 3) & 0xf8; \
ACCURATE_ALPHA_BLEND(sR, sG, sB, sA, dR, dG, dB); \
*((unsigned short *)dst) = ( \
((dR & 0xf8) << 8) | ((dG & 0xfc) << 3) | (dB >> 3) \
); \
} \
src += 4; \
dst += 2; \
widthvar--; \
}
ONE_PIXEL_BLEND((UNALIGNED_PTR(dst)) && (width), width);
extrawidth = (width % 8);
valigner = VEC_ALIGNER(src);
644
vsrc = (vector unsigned char) vec_ld(0, src);
645
646
647
648
649
650
651
652
653
width -= extrawidth;
while (width) {
vector unsigned char valpha;
vector unsigned char vsrc1, vsrc2;
vector unsigned char vdst1, vdst2;
vector unsigned short vR, vG, vB;
vector unsigned short vpixel, vrpixel, vgpixel, vbpixel;
/* Load 8 pixels from src as ARGB */
654
voverflow = (vector unsigned char) vec_ld(15, src);
655
656
657
vsrc = vec_perm(vsrc, voverflow, valigner);
vsrc1 = vec_perm(vsrc, vsrc, vpermute);
src += 16;
658
vsrc = (vector unsigned char) vec_ld(15, src);
659
660
661
662
663
664
voverflow = vec_perm(voverflow, vsrc, valigner);
vsrc2 = vec_perm(voverflow, voverflow, vpermute);
src += 16;
/* Load 8 pixels from dst as XRGB */
voverflow = vec_ld(0, dst);
665
666
vR = vec_and((vector unsigned short) voverflow, vf800);
vB = vec_sl((vector unsigned short) voverflow, v3_16);
667
vG = vec_sl(vB, v2_16);
668
669
670
671
672
673
674
675
676
677
678
679
vdst1 =
(vector unsigned char) vec_perm((vector unsigned char) vR,
(vector unsigned char) vR,
vredalpha1);
vdst1 = vec_perm(vdst1, (vector unsigned char) vB, vblue1);
vdst1 = vec_perm(vdst1, (vector unsigned char) vG, vgreen1);
vdst2 =
(vector unsigned char) vec_perm((vector unsigned char) vR,
(vector unsigned char) vR,
vredalpha2);
vdst2 = vec_perm(vdst2, (vector unsigned char) vB, vblue2);
vdst2 = vec_perm(vdst2, (vector unsigned char) vG, vgreen2);
680
681
682
/* Alpha blend 8 pixels as ARGB */
valpha = vec_perm(vsrc1, v0, valphaPermute);
683
684
VEC_MULTIPLY_ALPHA(vsrc1, vdst1, valpha, mergePermute, v1_16,
v8_16);
685
valpha = vec_perm(vsrc2, v0, valphaPermute);
686
687
VEC_MULTIPLY_ALPHA(vsrc2, vdst2, valpha, mergePermute, v1_16,
v8_16);
688
689
/* Convert 8 pixels to 565 */
690
691
692
693
694
vpixel = (vector unsigned short) vec_packpx((vector unsigned int)
vdst1,
(vector unsigned int)
vdst2);
vgpixel = (vector unsigned short) vec_perm(vdst1, vdst2, vgmerge);
695
696
697
698
699
vgpixel = vec_and(vgpixel, vfc);
vgpixel = vec_sl(vgpixel, v3_16);
vrpixel = vec_sl(vpixel, v1_16);
vrpixel = vec_and(vrpixel, vf800);
vbpixel = vec_and(vpixel, v3f);
700
701
702
703
704
vdst1 =
vec_or((vector unsigned char) vrpixel,
(vector unsigned char) vgpixel);
vdst1 = vec_or(vdst1, (vector unsigned char) vbpixel);
705
706
707
708
709
710
711
712
713
714
715
716
717
/* Store 8 pixels */
vec_st(vdst1, 0, dst);
width -= 8;
dst += 16;
}
ONE_PIXEL_BLEND((extrawidth), extrawidth);
#undef ONE_PIXEL_BLEND
src += srcskip;
dst += dstskip;
}
}
718
719
static void
Blit32to32SurfaceAlphaKeyAltivec(SDL_BlitInfo * info)
720
{
721
722
int height = info->dst_h;
Uint32 *srcp = (Uint32 *) info->src;
723
int srcskip = info->src_skip >> 2;
724
Uint32 *dstp = (Uint32 *) info->dst;
725
726
727
728
int dstskip = info->dst_skip >> 2;
SDL_PixelFormat *srcfmt = info->src_fmt;
SDL_PixelFormat *dstfmt = info->dst_fmt;
unsigned sA = info->a;
729
730
unsigned dA = dstfmt->Amask ? SDL_ALPHA_OPAQUE : 0;
Uint32 rgbmask = srcfmt->Rmask | srcfmt->Gmask | srcfmt->Bmask;
731
Uint32 ckey = info->colorkey;
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
vector unsigned char mergePermute;
vector unsigned char vsrcPermute;
vector unsigned char vdstPermute;
vector unsigned char vsdstPermute;
vector unsigned char valpha;
vector unsigned char valphamask;
vector unsigned char vbits;
vector unsigned char v0;
vector unsigned short v1;
vector unsigned short v8;
vector unsigned int vckey;
vector unsigned int vrgbmask;
mergePermute = VEC_MERGE_PERMUTE();
v0 = vec_splat_u8(0);
v1 = vec_splat_u16(1);
v8 = vec_splat_u16(8);
/* set the alpha to 255 on the destination surf */
valphamask = VEC_ALPHA_MASK();
vsrcPermute = calc_swizzle32(srcfmt, NULL);
vdstPermute = calc_swizzle32(NULL, dstfmt);
vsdstPermute = calc_swizzle32(dstfmt, NULL);
/* set a vector full of alpha and 255-alpha */
758
((unsigned char *) &valpha)[0] = sA;
759
valpha = vec_splat(valpha, 0);
760
vbits = (vector unsigned char) vec_splat_s8(-1);
761
762
ckey &= rgbmask;
763
((unsigned int *) (char *) &vckey)[0] = ckey;
764
vckey = vec_splat(vckey, 0);
765
((unsigned int *) (char *) &vrgbmask)[0] = rgbmask;
766
767
vrgbmask = vec_splat(vrgbmask, 0);
768
while (height--) {
769
int width = info->dst_w;
770
771
#define ONE_PIXEL_BLEND(condition, widthvar) \
while (condition) { \
772
Uint32 Pixel; \
773
unsigned sR, sG, sB, dR, dG, dB; \
774
775
776
777
RETRIEVE_RGB_PIXEL(((Uint8 *)srcp), 4, Pixel); \
if(sA && Pixel != ckey) { \
RGB_FROM_PIXEL(Pixel, srcfmt, sR, sG, sB); \
DISEMBLE_RGB(((Uint8 *)dstp), 4, dstfmt, Pixel, dR, dG, dB); \
778
779
780
ACCURATE_ALPHA_BLEND(sR, sG, sB, sA, dR, dG, dB); \
ASSEMBLE_RGBA(((Uint8 *)dstp), 4, dstfmt, dR, dG, dB, dA); \
} \
781
782
dstp++; \
srcp++; \
783
784
785
786
787
788
widthvar--; \
}
ONE_PIXEL_BLEND((UNALIGNED_PTR(dstp)) && (width), width);
if (width > 0) {
int extrawidth = (width % 4);
vector unsigned char valigner = VEC_ALIGNER(srcp);
789
vector unsigned char vs = (vector unsigned char) vec_ld(0, srcp);
790
791
792
793
794
795
796
797
width -= extrawidth;
while (width) {
vector unsigned char vsel;
vector unsigned char voverflow;
vector unsigned char vd;
vector unsigned char vd_orig;
/* s = *srcp */
798
voverflow = (vector unsigned char) vec_ld(15, srcp);
799
vs = vec_perm(vs, voverflow, valigner);
800
801
/* vsel is set for items that match the key */
802
803
804
805
806
vsel =
(vector unsigned char) vec_and((vector unsigned int) vs,
vrgbmask);
vsel = (vector unsigned char) vec_cmpeq((vector unsigned int)
vsel, vckey);
807
808
809
810
811
/* permute to source format */
vs = vec_perm(vs, valpha, vsrcPermute);
/* d = *dstp */
812
vd = (vector unsigned char) vec_ld(0, dstp);
813
814
815
816
817
818
819
820
821
vd_orig = vd = vec_perm(vd, v0, vsdstPermute);
VEC_MULTIPLY_ALPHA(vs, vd, valpha, mergePermute, v1, v8);
/* set the alpha channel to full on */
vd = vec_or(vd, valphamask);
/* mask out color key */
vd = vec_sel(vd, vd_orig, vsel);
822
823
824
825
826
/* permute to dest format */
vd = vec_perm(vd, vbits, vdstPermute);
/* *dstp = res */
827
828
vec_st((vector unsigned int) vd, 0, dstp);
829
830
831
832
833
834
835
836
srcp += 4;
dstp += 4;
width -= 4;
vs = voverflow;
}
ONE_PIXEL_BLEND((extrawidth), extrawidth);
}
#undef ONE_PIXEL_BLEND
837
838
839
840
841
842
843
srcp += srcskip;
dstp += dstskip;
}
}
844
845
static void
Blit32to32PixelAlphaAltivec(SDL_BlitInfo * info)
846
{
847
848
849
int width = info->dst_w;
int height = info->dst_h;
Uint32 *srcp = (Uint32 *) info->src;
850
int srcskip = info->src_skip >> 2;
851
Uint32 *dstp = (Uint32 *) info->dst;
852
853
854
int dstskip = info->dst_skip >> 2;
SDL_PixelFormat *srcfmt = info->src_fmt;
SDL_PixelFormat *dstfmt = info->dst_fmt;
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
vector unsigned char mergePermute;
vector unsigned char valphaPermute;
vector unsigned char vsrcPermute;
vector unsigned char vdstPermute;
vector unsigned char vsdstPermute;
vector unsigned char valphamask;
vector unsigned char vpixelmask;
vector unsigned char v0;
vector unsigned short v1;
vector unsigned short v8;
v0 = vec_splat_u8(0);
v1 = vec_splat_u16(1);
v8 = vec_splat_u16(8);
mergePermute = VEC_MERGE_PERMUTE();
valphamask = VEC_ALPHA_MASK();
871
valphaPermute = vec_and(vec_lvsl(0, (int *) NULL), vec_splat_u8(0xC));
872
873
874
875
876
vpixelmask = vec_nor(valphamask, v0);
vsrcPermute = calc_swizzle32(srcfmt, NULL);
vdstPermute = calc_swizzle32(NULL, dstfmt);
vsdstPermute = calc_swizzle32(dstfmt, NULL);
877
while (height--) {
878
width = info->dst_w;
879
#define ONE_PIXEL_BLEND(condition, widthvar) while ((condition)) { \
880
Uint32 Pixel; \
881
unsigned sR, sG, sB, dR, dG, dB, sA, dA; \
882
DISEMBLE_RGBA((Uint8 *)srcp, 4, srcfmt, Pixel, sR, sG, sB, sA); \
883
if(sA) { \
884
DISEMBLE_RGBA((Uint8 *)dstp, 4, dstfmt, Pixel, dR, dG, dB, dA); \
885
886
887
888
889
890
891
892
893
ACCURATE_ALPHA_BLEND(sR, sG, sB, sA, dR, dG, dB); \
ASSEMBLE_RGBA((Uint8 *)dstp, 4, dstfmt, dR, dG, dB, dA); \
} \
++srcp; \
++dstp; \
widthvar--; \
}
ONE_PIXEL_BLEND((UNALIGNED_PTR(dstp)) && (width), width);
if (width > 0) {
894
895
/* vsrcPermute */
/* vdstPermute */
896
897
int extrawidth = (width % 4);
vector unsigned char valigner = VEC_ALIGNER(srcp);
898
vector unsigned char vs = (vector unsigned char) vec_ld(0, srcp);
899
900
901
902
903
904
905
width -= extrawidth;
while (width) {
vector unsigned char voverflow;
vector unsigned char vd;
vector unsigned char valpha;
vector unsigned char vdstalpha;
/* s = *srcp */
906
voverflow = (vector unsigned char) vec_ld(15, srcp);
907
908
909
910
vs = vec_perm(vs, voverflow, valigner);
vs = vec_perm(vs, v0, vsrcPermute);
valpha = vec_perm(vs, v0, valphaPermute);
911
912
/* d = *dstp */
913
vd = (vector unsigned char) vec_ld(0, dstp);
914
915
916
917
918
919
920
921
922
923
924
vd = vec_perm(vd, v0, vsdstPermute);
vdstalpha = vec_and(vd, valphamask);
VEC_MULTIPLY_ALPHA(vs, vd, valpha, mergePermute, v1, v8);
/* set the alpha to the dest alpha */
vd = vec_and(vd, vpixelmask);
vd = vec_or(vd, vdstalpha);
vd = vec_perm(vd, v0, vdstPermute);
/* *dstp = res */
925
926
vec_st((vector unsigned int) vd, 0, dstp);
927
928
929
930
931
932
933
934
srcp += 4;
dstp += 4;
width -= 4;
vs = voverflow;
}
ONE_PIXEL_BLEND((extrawidth), extrawidth);
}
935
936
srcp += srcskip;
dstp += dstskip;
937
#undef ONE_PIXEL_BLEND
938
}
939
940
941
}
/* fast ARGB888->(A)RGB888 blending with pixel alpha */
942
943
static void
BlitRGBtoRGBPixelAlphaAltivec(SDL_BlitInfo * info)
944
{
945
946
947
int width = info->dst_w;
int height = info->dst_h;
Uint32 *srcp = (Uint32 *) info->src;
948
int srcskip = info->src_skip >> 2;
949
Uint32 *dstp = (Uint32 *) info->dst;
950
int dstskip = info->dst_skip >> 2;
951
952
953
954
955
956
957
958
959
960
961
962
vector unsigned char mergePermute;
vector unsigned char valphaPermute;
vector unsigned char valphamask;
vector unsigned char vpixelmask;
vector unsigned char v0;
vector unsigned short v1;
vector unsigned short v8;
v0 = vec_splat_u8(0);
v1 = vec_splat_u16(1);
v8 = vec_splat_u16(8);
mergePermute = VEC_MERGE_PERMUTE();
valphamask = VEC_ALPHA_MASK();
963
964
965
valphaPermute = vec_and(vec_lvsl(0, (int *) NULL), vec_splat_u8(0xC));
966
vpixelmask = vec_nor(valphamask, v0);
967
while (height--) {
968
width = info->dst_w;
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
#define ONE_PIXEL_BLEND(condition, widthvar) \
while ((condition)) { \
Uint32 dalpha; \
Uint32 d; \
Uint32 s1; \
Uint32 d1; \
Uint32 s = *srcp; \
Uint32 alpha = s >> 24; \
if(alpha) { \
if(alpha == SDL_ALPHA_OPAQUE) { \
*dstp = (s & 0x00ffffff) | (*dstp & 0xff000000); \
} else { \
d = *dstp; \
dalpha = d & 0xff000000; \
s1 = s & 0xff00ff; \
d1 = d & 0xff00ff; \
d1 = (d1 + ((s1 - d1) * alpha >> 8)) & 0xff00ff; \
s &= 0xff00; \
d &= 0xff00; \
d = (d + ((s - d) * alpha >> 8)) & 0xff00; \
*dstp = d1 | d | dalpha; \
} \
} \
++srcp; \
++dstp; \
widthvar--; \
}
ONE_PIXEL_BLEND((UNALIGNED_PTR(dstp)) && (width), width);
if (width > 0) {
int extrawidth = (width % 4);
vector unsigned char valigner = VEC_ALIGNER(srcp);
1000
vector unsigned char vs = (vector unsigned char) vec_ld(0, srcp);