yuv2rgb_altivec.c
Go to the documentation of this file.
1 /*
2  * AltiVec acceleration for colorspace conversion
3  *
4  * copyright (C) 2004 Marc Hoffman <marc.hoffman@analog.com>
5  *
6  * This file is part of Libav.
7  *
8  * Libav is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU Lesser General Public
10  * License as published by the Free Software Foundation; either
11  * version 2.1 of the License, or (at your option) any later version.
12  *
13  * Libav is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16  * Lesser General Public License for more details.
17  *
18  * You should have received a copy of the GNU Lesser General Public
19  * License along with Libav; if not, write to the Free Software
20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21  */
22 
23 /*
24  * Convert I420 YV12 to RGB in various formats,
25  * it rejects images that are not in 420 formats,
26  * it rejects images that don't have widths of multiples of 16,
27  * it rejects images that don't have heights of multiples of 2.
28  * Reject defers to C simulation code.
29  *
30  * Lots of optimizations to be done here.
31  *
32  * 1. Need to fix saturation code. I just couldn't get it to fly with packs
33  * and adds, so we currently use max/min to clip.
34  *
35  * 2. The inefficient use of chroma loading needs a bit of brushing up.
36  *
37  * 3. Analysis of pipeline stalls needs to be done. Use shark to identify
38  * pipeline stalls.
39  *
40  *
41  * MODIFIED to calculate coeffs from currently selected color space.
42  * MODIFIED core to be a macro where you specify the output format.
43  * ADDED UYVY conversion which is never called due to some thing in swscale.
44  * CORRECTED algorithim selection to be strict on input formats.
45  * ADDED runtime detection of AltiVec.
46  *
47  * ADDED altivec_yuv2packedX vertical scl + RGB converter
48  *
49  * March 27,2004
50  * PERFORMANCE ANALYSIS
51  *
52  * The C version uses 25% of the processor or ~250Mips for D1 video rawvideo
53  * used as test.
54  * The AltiVec version uses 10% of the processor or ~100Mips for D1 video
55  * same sequence.
56  *
57  * 720 * 480 * 30 ~10MPS
58  *
59  * so we have roughly 10 clocks per pixel. This is too high, something has
60  * to be wrong.
61  *
62  * OPTIMIZED clip codes to utilize vec_max and vec_packs removing the
63  * need for vec_min.
64  *
65  * OPTIMIZED DST OUTPUT cache/DMA controls. We are pretty much guaranteed to
66  * have the input video frame, it was just decompressed so it probably resides
67  * in L1 caches. However, we are creating the output video stream. This needs
68  * to use the DSTST instruction to optimize for the cache. We couple this with
69  * the fact that we are not going to be visiting the input buffer again so we
70  * mark it Least Recently Used. This shaves 25% of the processor cycles off.
71  *
72  * Now memcpy is the largest mips consumer in the system, probably due
73  * to the inefficient X11 stuff.
74  *
75  * GL libraries seem to be very slow on this machine 1.33Ghz PB running
76  * Jaguar, this is not the case for my 1Ghz PB. I thought it might be
77  * a versioning issue, however I have libGL.1.2.dylib for both
78  * machines. (We need to figure this out now.)
79  *
80  * GL2 libraries work now with patch for RGB32.
81  *
82  * NOTE: quartz vo driver ARGB32_to_RGB24 consumes 30% of the processor.
83  *
84  * Integrated luma prescaling adjustment for saturation/contrast/brightness
85  * adjustment.
86  */
87 
88 #include <stdio.h>
89 #include <stdlib.h>
90 #include <string.h>
91 #include <inttypes.h>
92 #include <assert.h>
93 
94 #include "config.h"
95 #include "libswscale/rgb2rgb.h"
96 #include "libswscale/swscale.h"
98 #include "libavutil/attributes.h"
99 #include "libavutil/cpu.h"
100 #include "yuv2rgb_altivec.h"
101 
102 #undef PROFILE_THE_BEAST
103 #undef INC_SCALING
104 
105 typedef unsigned char ubyte;
106 typedef signed char sbyte;
107 
108 /* RGB interleaver, 16 planar pels 8-bit samples per channel in
109  * homogeneous vector registers x0,x1,x2 are interleaved with the
110  * following technique:
111  *
112  * o0 = vec_mergeh(x0, x1);
113  * o1 = vec_perm(o0, x2, perm_rgb_0);
114  * o2 = vec_perm(o0, x2, perm_rgb_1);
115  * o3 = vec_mergel(x0, x1);
116  * o4 = vec_perm(o3, o2, perm_rgb_2);
117  * o5 = vec_perm(o3, o2, perm_rgb_3);
118  *
119  * perm_rgb_0: o0(RG).h v1(B) --> o1*
120  * 0 1 2 3 4
121  * rgbr|gbrg|brgb|rgbr
122  * 0010 0100 1001 0010
123  * 0102 3145 2673 894A
124  *
125  * perm_rgb_1: o0(RG).h v1(B) --> o2
126  * 0 1 2 3 4
127  * gbrg|brgb|bbbb|bbbb
128  * 0100 1001 1111 1111
129  * B5CD 6EF7 89AB CDEF
130  *
131  * perm_rgb_2: o3(RG).l o2(rgbB.l) --> o4*
132  * 0 1 2 3 4
133  * gbrg|brgb|rgbr|gbrg
134  * 1111 1111 0010 0100
135  * 89AB CDEF 0182 3945
136  *
137  * perm_rgb_2: o3(RG).l o2(rgbB.l) ---> o5*
138  * 0 1 2 3 4
139  * brgb|rgbr|gbrg|brgb
140  * 1001 0010 0100 1001
141  * a67b 89cA BdCD eEFf
142  *
143  */
144 static const vector unsigned char
145  perm_rgb_0 = { 0x00, 0x01, 0x10, 0x02, 0x03, 0x11, 0x04, 0x05,
146  0x12, 0x06, 0x07, 0x13, 0x08, 0x09, 0x14, 0x0a },
147  perm_rgb_1 = { 0x0b, 0x15, 0x0c, 0x0d, 0x16, 0x0e, 0x0f, 0x17,
148  0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f },
149  perm_rgb_2 = { 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
150  0x00, 0x01, 0x18, 0x02, 0x03, 0x19, 0x04, 0x05 },
151  perm_rgb_3 = { 0x1a, 0x06, 0x07, 0x1b, 0x08, 0x09, 0x1c, 0x0a,
152  0x0b, 0x1d, 0x0c, 0x0d, 0x1e, 0x0e, 0x0f, 0x1f };
153 
154 #define vec_merge3(x2, x1, x0, y0, y1, y2) \
155  do { \
156  __typeof__(x0) o0, o2, o3; \
157  o0 = vec_mergeh(x0, x1); \
158  y0 = vec_perm(o0, x2, perm_rgb_0); \
159  o2 = vec_perm(o0, x2, perm_rgb_1); \
160  o3 = vec_mergel(x0, x1); \
161  y1 = vec_perm(o3, o2, perm_rgb_2); \
162  y2 = vec_perm(o3, o2, perm_rgb_3); \
163  } while (0)
164 
165 #define vec_mstbgr24(x0, x1, x2, ptr) \
166  do { \
167  __typeof__(x0) _0, _1, _2; \
168  vec_merge3(x0, x1, x2, _0, _1, _2); \
169  vec_st(_0, 0, ptr++); \
170  vec_st(_1, 0, ptr++); \
171  vec_st(_2, 0, ptr++); \
172  } while (0)
173 
174 #define vec_mstrgb24(x0, x1, x2, ptr) \
175  do { \
176  __typeof__(x0) _0, _1, _2; \
177  vec_merge3(x2, x1, x0, _0, _1, _2); \
178  vec_st(_0, 0, ptr++); \
179  vec_st(_1, 0, ptr++); \
180  vec_st(_2, 0, ptr++); \
181  } while (0)
182 
183 /* pack the pixels in rgb0 format
184  * msb R
185  * lsb 0
186  */
187 #define vec_mstrgb32(T, x0, x1, x2, x3, ptr) \
188  do { \
189  T _0, _1, _2, _3; \
190  _0 = vec_mergeh(x0, x1); \
191  _1 = vec_mergeh(x2, x3); \
192  _2 = (T) vec_mergeh((vector unsigned short) _0, \
193  (vector unsigned short) _1); \
194  _3 = (T) vec_mergel((vector unsigned short) _0, \
195  (vector unsigned short) _1); \
196  vec_st(_2, 0 * 16, (T *) ptr); \
197  vec_st(_3, 1 * 16, (T *) ptr); \
198  _0 = vec_mergel(x0, x1); \
199  _1 = vec_mergel(x2, x3); \
200  _2 = (T) vec_mergeh((vector unsigned short) _0, \
201  (vector unsigned short) _1); \
202  _3 = (T) vec_mergel((vector unsigned short) _0, \
203  (vector unsigned short) _1); \
204  vec_st(_2, 2 * 16, (T *) ptr); \
205  vec_st(_3, 3 * 16, (T *) ptr); \
206  ptr += 4; \
207  } while (0)
208 
209 /*
210  * 1 0 1.4021 | | Y |
211  * 1 -0.3441 -0.7142 |x| Cb|
212  * 1 1.7718 0 | | Cr|
213  *
214  *
215  * Y: [-128 127]
216  * Cb/Cr : [-128 127]
217  *
218  * typical YUV conversion works on Y: 0-255 this version has been
219  * optimized for JPEG decoding.
220  */
221 
222 #define vec_unh(x) \
223  (vector signed short) \
224  vec_perm(x, (__typeof__(x)) { 0 }, \
225  ((vector unsigned char) { \
226  0x10, 0x00, 0x10, 0x01, 0x10, 0x02, 0x10, 0x03, \
227  0x10, 0x04, 0x10, 0x05, 0x10, 0x06, 0x10, 0x07 }))
228 
229 #define vec_unl(x) \
230  (vector signed short) \
231  vec_perm(x, (__typeof__(x)) { 0 }, \
232  ((vector unsigned char) { \
233  0x10, 0x08, 0x10, 0x09, 0x10, 0x0A, 0x10, 0x0B, \
234  0x10, 0x0C, 0x10, 0x0D, 0x10, 0x0E, 0x10, 0x0F }))
235 
236 #define vec_clip_s16(x) \
237  vec_max(vec_min(x, ((vector signed short) { \
238  235, 235, 235, 235, 235, 235, 235, 235 })), \
239  ((vector signed short) { 16, 16, 16, 16, 16, 16, 16, 16 }))
240 
241 #define vec_packclp(x, y) \
242  (vector unsigned char) \
243  vec_packs((vector unsigned short) \
244  vec_max(x, ((vector signed short) { 0 })), \
245  (vector unsigned short) \
246  vec_max(y, ((vector signed short) { 0 })))
247 
248 //#define out_pixels(a, b, c, ptr) vec_mstrgb32(__typeof__(a), ((__typeof__(a)) { 255 }), a, a, a, ptr)
249 
250 static inline void cvtyuvtoRGB(SwsContext *c, vector signed short Y,
251  vector signed short U, vector signed short V,
252  vector signed short *R, vector signed short *G,
253  vector signed short *B)
254 {
255  vector signed short vx, ux, uvx;
256 
257  Y = vec_mradds(Y, c->CY, c->OY);
258  U = vec_sub(U, (vector signed short)
259  vec_splat((vector signed short) { 128 }, 0));
260  V = vec_sub(V, (vector signed short)
261  vec_splat((vector signed short) { 128 }, 0));
262 
263  // ux = (CBU * (u << c->CSHIFT) + 0x4000) >> 15;
264  ux = vec_sl(U, c->CSHIFT);
265  *B = vec_mradds(ux, c->CBU, Y);
266 
267  // vx = (CRV * (v << c->CSHIFT) + 0x4000) >> 15;
268  vx = vec_sl(V, c->CSHIFT);
269  *R = vec_mradds(vx, c->CRV, Y);
270 
271  // uvx = ((CGU * u) + (CGV * v)) >> 15;
272  uvx = vec_mradds(U, c->CGU, Y);
273  *G = vec_mradds(V, c->CGV, uvx);
274 }
275 
276 /*
277  * ------------------------------------------------------------------------------
278  * CS converters
279  * ------------------------------------------------------------------------------
280  */
281 
282 #define DEFCSP420_CVT(name, out_pixels) \
283 static int altivec_ ## name(SwsContext *c, const unsigned char **in, \
284  int *instrides, int srcSliceY, int srcSliceH, \
285  unsigned char **oplanes, int *outstrides) \
286 { \
287  int w = c->srcW; \
288  int h = srcSliceH; \
289  int i, j; \
290  int instrides_scl[3]; \
291  vector unsigned char y0, y1; \
292  \
293  vector signed char u, v; \
294  \
295  vector signed short Y0, Y1, Y2, Y3; \
296  vector signed short U, V; \
297  vector signed short vx, ux, uvx; \
298  vector signed short vx0, ux0, uvx0; \
299  vector signed short vx1, ux1, uvx1; \
300  vector signed short R0, G0, B0; \
301  vector signed short R1, G1, B1; \
302  vector unsigned char R, G, B; \
303  \
304  const vector unsigned char *y1ivP, *y2ivP, *uivP, *vivP; \
305  vector unsigned char align_perm; \
306  \
307  vector signed short lCY = c->CY; \
308  vector signed short lOY = c->OY; \
309  vector signed short lCRV = c->CRV; \
310  vector signed short lCBU = c->CBU; \
311  vector signed short lCGU = c->CGU; \
312  vector signed short lCGV = c->CGV; \
313  vector unsigned short lCSHIFT = c->CSHIFT; \
314  \
315  const ubyte *y1i = in[0]; \
316  const ubyte *y2i = in[0] + instrides[0]; \
317  const ubyte *ui = in[1]; \
318  const ubyte *vi = in[2]; \
319  \
320  vector unsigned char *oute = \
321  (vector unsigned char *) \
322  (oplanes[0] + srcSliceY * outstrides[0]); \
323  vector unsigned char *outo = \
324  (vector unsigned char *) \
325  (oplanes[0] + srcSliceY * outstrides[0] + outstrides[0]); \
326  \
327  /* loop moves y{1, 2}i by w */ \
328  instrides_scl[0] = instrides[0] * 2 - w; \
329  /* loop moves ui by w / 2 */ \
330  instrides_scl[1] = instrides[1] - w / 2; \
331  /* loop moves vi by w / 2 */ \
332  instrides_scl[2] = instrides[2] - w / 2; \
333  \
334  for (i = 0; i < h / 2; i++) { \
335  vec_dstst(outo, (0x02000002 | (((w * 3 + 32) / 32) << 16)), 0); \
336  vec_dstst(oute, (0x02000002 | (((w * 3 + 32) / 32) << 16)), 1); \
337  \
338  for (j = 0; j < w / 16; j++) { \
339  y1ivP = (const vector unsigned char *) y1i; \
340  y2ivP = (const vector unsigned char *) y2i; \
341  uivP = (const vector unsigned char *) ui; \
342  vivP = (const vector unsigned char *) vi; \
343  \
344  align_perm = vec_lvsl(0, y1i); \
345  y0 = (vector unsigned char) \
346  vec_perm(y1ivP[0], y1ivP[1], align_perm); \
347  \
348  align_perm = vec_lvsl(0, y2i); \
349  y1 = (vector unsigned char) \
350  vec_perm(y2ivP[0], y2ivP[1], align_perm); \
351  \
352  align_perm = vec_lvsl(0, ui); \
353  u = (vector signed char) \
354  vec_perm(uivP[0], uivP[1], align_perm); \
355  \
356  align_perm = vec_lvsl(0, vi); \
357  v = (vector signed char) \
358  vec_perm(vivP[0], vivP[1], align_perm); \
359  \
360  u = (vector signed char) \
361  vec_sub(u, \
362  (vector signed char) \
363  vec_splat((vector signed char) { 128 }, 0)); \
364  v = (vector signed char) \
365  vec_sub(v, \
366  (vector signed char) \
367  vec_splat((vector signed char) { 128 }, 0)); \
368  \
369  U = vec_unpackh(u); \
370  V = vec_unpackh(v); \
371  \
372  Y0 = vec_unh(y0); \
373  Y1 = vec_unl(y0); \
374  Y2 = vec_unh(y1); \
375  Y3 = vec_unl(y1); \
376  \
377  Y0 = vec_mradds(Y0, lCY, lOY); \
378  Y1 = vec_mradds(Y1, lCY, lOY); \
379  Y2 = vec_mradds(Y2, lCY, lOY); \
380  Y3 = vec_mradds(Y3, lCY, lOY); \
381  \
382  /* ux = (CBU * (u << CSHIFT) + 0x4000) >> 15 */ \
383  ux = vec_sl(U, lCSHIFT); \
384  ux = vec_mradds(ux, lCBU, (vector signed short) { 0 }); \
385  ux0 = vec_mergeh(ux, ux); \
386  ux1 = vec_mergel(ux, ux); \
387  \
388  /* vx = (CRV * (v << CSHIFT) + 0x4000) >> 15; */ \
389  vx = vec_sl(V, lCSHIFT); \
390  vx = vec_mradds(vx, lCRV, (vector signed short) { 0 }); \
391  vx0 = vec_mergeh(vx, vx); \
392  vx1 = vec_mergel(vx, vx); \
393  \
394  /* uvx = ((CGU * u) + (CGV * v)) >> 15 */ \
395  uvx = vec_mradds(U, lCGU, (vector signed short) { 0 }); \
396  uvx = vec_mradds(V, lCGV, uvx); \
397  uvx0 = vec_mergeh(uvx, uvx); \
398  uvx1 = vec_mergel(uvx, uvx); \
399  \
400  R0 = vec_add(Y0, vx0); \
401  G0 = vec_add(Y0, uvx0); \
402  B0 = vec_add(Y0, ux0); \
403  R1 = vec_add(Y1, vx1); \
404  G1 = vec_add(Y1, uvx1); \
405  B1 = vec_add(Y1, ux1); \
406  \
407  R = vec_packclp(R0, R1); \
408  G = vec_packclp(G0, G1); \
409  B = vec_packclp(B0, B1); \
410  \
411  out_pixels(R, G, B, oute); \
412  \
413  R0 = vec_add(Y2, vx0); \
414  G0 = vec_add(Y2, uvx0); \
415  B0 = vec_add(Y2, ux0); \
416  R1 = vec_add(Y3, vx1); \
417  G1 = vec_add(Y3, uvx1); \
418  B1 = vec_add(Y3, ux1); \
419  R = vec_packclp(R0, R1); \
420  G = vec_packclp(G0, G1); \
421  B = vec_packclp(B0, B1); \
422  \
423  \
424  out_pixels(R, G, B, outo); \
425  \
426  y1i += 16; \
427  y2i += 16; \
428  ui += 8; \
429  vi += 8; \
430  } \
431  \
432  outo += (outstrides[0]) >> 4; \
433  oute += (outstrides[0]) >> 4; \
434  \
435  ui += instrides_scl[1]; \
436  vi += instrides_scl[2]; \
437  y1i += instrides_scl[0]; \
438  y2i += instrides_scl[0]; \
439  } \
440  return srcSliceH; \
441 }
442 
443 #define out_abgr(a, b, c, ptr) \
444  vec_mstrgb32(__typeof__(a), ((__typeof__(a)) { 255 }), c, b, a, ptr)
445 #define out_bgra(a, b, c, ptr) \
446  vec_mstrgb32(__typeof__(a), c, b, a, ((__typeof__(a)) { 255 }), ptr)
447 #define out_rgba(a, b, c, ptr) \
448  vec_mstrgb32(__typeof__(a), a, b, c, ((__typeof__(a)) { 255 }), ptr)
449 #define out_argb(a, b, c, ptr) \
450  vec_mstrgb32(__typeof__(a), ((__typeof__(a)) { 255 }), a, b, c, ptr)
451 #define out_rgb24(a, b, c, ptr) vec_mstrgb24(a, b, c, ptr)
452 #define out_bgr24(a, b, c, ptr) vec_mstbgr24(a, b, c, ptr)
453 
454 DEFCSP420_CVT(yuv2_abgr, out_abgr)
455 DEFCSP420_CVT(yuv2_bgra, out_bgra)
456 DEFCSP420_CVT(yuv2_rgba, out_rgba)
457 DEFCSP420_CVT(yuv2_argb, out_argb)
458 DEFCSP420_CVT(yuv2_rgb24, out_rgb24)
459 DEFCSP420_CVT(yuv2_bgr24, out_bgr24)
460 
461 // uyvy|uyvy|uyvy|uyvy
462 // 0123 4567 89ab cdef
463 static const vector unsigned char
464  demux_u = { 0x10, 0x00, 0x10, 0x00,
465  0x10, 0x04, 0x10, 0x04,
466  0x10, 0x08, 0x10, 0x08,
467  0x10, 0x0c, 0x10, 0x0c },
468  demux_v = { 0x10, 0x02, 0x10, 0x02,
469  0x10, 0x06, 0x10, 0x06,
470  0x10, 0x0A, 0x10, 0x0A,
471  0x10, 0x0E, 0x10, 0x0E },
472  demux_y = { 0x10, 0x01, 0x10, 0x03,
473  0x10, 0x05, 0x10, 0x07,
474  0x10, 0x09, 0x10, 0x0B,
475  0x10, 0x0D, 0x10, 0x0F };
476 
477 /*
478  * this is so I can play live CCIR raw video
479  */
480 static int altivec_uyvy_rgb32(SwsContext *c, const unsigned char **in,
481  int *instrides, int srcSliceY, int srcSliceH,
482  unsigned char **oplanes, int *outstrides)
483 {
484  int w = c->srcW;
485  int h = srcSliceH;
486  int i, j;
487  vector unsigned char uyvy;
488  vector signed short Y, U, V;
489  vector signed short R0, G0, B0, R1, G1, B1;
490  vector unsigned char R, G, B;
491  vector unsigned char *out;
492  const ubyte *img;
493 
494  img = in[0];
495  out = (vector unsigned char *) (oplanes[0] + srcSliceY * outstrides[0]);
496 
497  for (i = 0; i < h; i++)
498  for (j = 0; j < w / 16; j++) {
499  uyvy = vec_ld(0, img);
500 
501  U = (vector signed short)
502  vec_perm(uyvy, (vector unsigned char) { 0 }, demux_u);
503  V = (vector signed short)
504  vec_perm(uyvy, (vector unsigned char) { 0 }, demux_v);
505  Y = (vector signed short)
506  vec_perm(uyvy, (vector unsigned char) { 0 }, demux_y);
507 
508  cvtyuvtoRGB(c, Y, U, V, &R0, &G0, &B0);
509 
510  uyvy = vec_ld(16, img);
511 
512  U = (vector signed short)
513  vec_perm(uyvy, (vector unsigned char) { 0 }, demux_u);
514  V = (vector signed short)
515  vec_perm(uyvy, (vector unsigned char) { 0 }, demux_v);
516  Y = (vector signed short)
517  vec_perm(uyvy, (vector unsigned char) { 0 }, demux_y);
518 
519  cvtyuvtoRGB(c, Y, U, V, &R1, &G1, &B1);
520 
521  R = vec_packclp(R0, R1);
522  G = vec_packclp(G0, G1);
523  B = vec_packclp(B0, B1);
524 
525  // vec_mstbgr24 (R,G,B, out);
526  out_rgba(R, G, B, out);
527 
528  img += 32;
529  }
530  return srcSliceH;
531 }
532 
533 /* Ok currently the acceleration routine only supports
534  * inputs of widths a multiple of 16
535  * and heights a multiple 2
536  *
537  * So we just fall back to the C codes for this.
538  */
540 {
542  return NULL;
543 
544  /*
545  * and this seems not to matter too much I tried a bunch of
546  * videos with abnormal widths and MPlayer crashes elsewhere.
547  * mplayer -vo x11 -rawvideo on:w=350:h=240 raw-350x240.eyuv
548  * boom with X11 bad match.
549  *
550  */
551  if ((c->srcW & 0xf) != 0)
552  return NULL;
553 
554  switch (c->srcFormat) {
555  case AV_PIX_FMT_YUV410P:
556  case AV_PIX_FMT_YUV420P:
557  /*case IMGFMT_CLPL: ??? */
558  case AV_PIX_FMT_GRAY8:
559  case AV_PIX_FMT_NV12:
560  case AV_PIX_FMT_NV21:
561  if ((c->srcH & 0x1) != 0)
562  return NULL;
563 
564  switch (c->dstFormat) {
565  case AV_PIX_FMT_RGB24:
566  av_log(c, AV_LOG_WARNING, "ALTIVEC: Color Space RGB24\n");
567  return altivec_yuv2_rgb24;
568  case AV_PIX_FMT_BGR24:
569  av_log(c, AV_LOG_WARNING, "ALTIVEC: Color Space BGR24\n");
570  return altivec_yuv2_bgr24;
571  case AV_PIX_FMT_ARGB:
572  av_log(c, AV_LOG_WARNING, "ALTIVEC: Color Space ARGB\n");
573  return altivec_yuv2_argb;
574  case AV_PIX_FMT_ABGR:
575  av_log(c, AV_LOG_WARNING, "ALTIVEC: Color Space ABGR\n");
576  return altivec_yuv2_abgr;
577  case AV_PIX_FMT_RGBA:
578  av_log(c, AV_LOG_WARNING, "ALTIVEC: Color Space RGBA\n");
579  return altivec_yuv2_rgba;
580  case AV_PIX_FMT_BGRA:
581  av_log(c, AV_LOG_WARNING, "ALTIVEC: Color Space BGRA\n");
582  return altivec_yuv2_bgra;
583  default: return NULL;
584  }
585  break;
586 
587  case AV_PIX_FMT_UYVY422:
588  switch (c->dstFormat) {
589  case AV_PIX_FMT_BGR32:
590  av_log(c, AV_LOG_WARNING, "ALTIVEC: Color Space UYVY -> RGB32\n");
591  return altivec_uyvy_rgb32;
592  default: return NULL;
593  }
594  break;
595  }
596  return NULL;
597 }
598 
600  const int inv_table[4],
601  int brightness,
602  int contrast,
603  int saturation)
604 {
605  union {
606  DECLARE_ALIGNED(16, signed short, tmp)[8];
607  vector signed short vec;
608  } buf;
609 
610  buf.tmp[0] = ((0xffffLL) * contrast >> 8) >> 9; // cy
611  buf.tmp[1] = -256 * brightness; // oy
612  buf.tmp[2] = (inv_table[0] >> 3) * (contrast >> 16) * (saturation >> 16); // crv
613  buf.tmp[3] = (inv_table[1] >> 3) * (contrast >> 16) * (saturation >> 16); // cbu
614  buf.tmp[4] = -((inv_table[2] >> 1) * (contrast >> 16) * (saturation >> 16)); // cgu
615  buf.tmp[5] = -((inv_table[3] >> 1) * (contrast >> 16) * (saturation >> 16)); // cgv
616 
617  c->CSHIFT = (vector unsigned short) vec_splat_u16(2);
618  c->CY = vec_splat((vector signed short) buf.vec, 0);
619  c->OY = vec_splat((vector signed short) buf.vec, 1);
620  c->CRV = vec_splat((vector signed short) buf.vec, 2);
621  c->CBU = vec_splat((vector signed short) buf.vec, 3);
622  c->CGU = vec_splat((vector signed short) buf.vec, 4);
623  c->CGV = vec_splat((vector signed short) buf.vec, 5);
624  return;
625 }
626 
628  const int16_t *lumFilter,
629  const int16_t **lumSrc,
630  int lumFilterSize,
631  const int16_t *chrFilter,
632  const int16_t **chrUSrc,
633  const int16_t **chrVSrc,
634  int chrFilterSize,
635  const int16_t **alpSrc,
636  uint8_t *dest,
637  int dstW, int dstY,
638  enum AVPixelFormat target)
639 {
640  int i, j;
641  vector signed short X, X0, X1, Y0, U0, V0, Y1, U1, V1, U, V;
642  vector signed short R0, G0, B0, R1, G1, B1;
643 
644  vector unsigned char R, G, B;
645  vector unsigned char *out, *nout;
646 
647  vector signed short RND = vec_splat_s16(1 << 3);
648  vector unsigned short SCL = vec_splat_u16(4);
649  DECLARE_ALIGNED(16, unsigned int, scratch)[16];
650 
651  vector signed short *YCoeffs, *CCoeffs;
652 
653  YCoeffs = c->vYCoeffsBank + dstY * lumFilterSize;
654  CCoeffs = c->vCCoeffsBank + dstY * chrFilterSize;
655 
656  out = (vector unsigned char *) dest;
657 
658  for (i = 0; i < dstW; i += 16) {
659  Y0 = RND;
660  Y1 = RND;
661  /* extract 16 coeffs from lumSrc */
662  for (j = 0; j < lumFilterSize; j++) {
663  X0 = vec_ld(0, &lumSrc[j][i]);
664  X1 = vec_ld(16, &lumSrc[j][i]);
665  Y0 = vec_mradds(X0, YCoeffs[j], Y0);
666  Y1 = vec_mradds(X1, YCoeffs[j], Y1);
667  }
668 
669  U = RND;
670  V = RND;
671  /* extract 8 coeffs from U,V */
672  for (j = 0; j < chrFilterSize; j++) {
673  X = vec_ld(0, &chrUSrc[j][i / 2]);
674  U = vec_mradds(X, CCoeffs[j], U);
675  X = vec_ld(0, &chrVSrc[j][i / 2]);
676  V = vec_mradds(X, CCoeffs[j], V);
677  }
678 
679  /* scale and clip signals */
680  Y0 = vec_sra(Y0, SCL);
681  Y1 = vec_sra(Y1, SCL);
682  U = vec_sra(U, SCL);
683  V = vec_sra(V, SCL);
684 
685  Y0 = vec_clip_s16(Y0);
686  Y1 = vec_clip_s16(Y1);
687  U = vec_clip_s16(U);
688  V = vec_clip_s16(V);
689 
690  /* now we have
691  * Y0 = y0 y1 y2 y3 y4 y5 y6 y7 Y1 = y8 y9 y10 y11 y12 y13 y14 y15
692  * U = u0 u1 u2 u3 u4 u5 u6 u7 V = v0 v1 v2 v3 v4 v5 v6 v7
693  *
694  * Y0 = y0 y1 y2 y3 y4 y5 y6 y7 Y1 = y8 y9 y10 y11 y12 y13 y14 y15
695  * U0 = u0 u0 u1 u1 u2 u2 u3 u3 U1 = u4 u4 u5 u5 u6 u6 u7 u7
696  * V0 = v0 v0 v1 v1 v2 v2 v3 v3 V1 = v4 v4 v5 v5 v6 v6 v7 v7
697  */
698 
699  U0 = vec_mergeh(U, U);
700  V0 = vec_mergeh(V, V);
701 
702  U1 = vec_mergel(U, U);
703  V1 = vec_mergel(V, V);
704 
705  cvtyuvtoRGB(c, Y0, U0, V0, &R0, &G0, &B0);
706  cvtyuvtoRGB(c, Y1, U1, V1, &R1, &G1, &B1);
707 
708  R = vec_packclp(R0, R1);
709  G = vec_packclp(G0, G1);
710  B = vec_packclp(B0, B1);
711 
712  switch (target) {
713  case AV_PIX_FMT_ABGR:
714  out_abgr(R, G, B, out);
715  break;
716  case AV_PIX_FMT_BGRA:
717  out_bgra(R, G, B, out);
718  break;
719  case AV_PIX_FMT_RGBA:
720  out_rgba(R, G, B, out);
721  break;
722  case AV_PIX_FMT_ARGB:
723  out_argb(R, G, B, out);
724  break;
725  case AV_PIX_FMT_RGB24:
726  out_rgb24(R, G, B, out);
727  break;
728  case AV_PIX_FMT_BGR24:
729  out_bgr24(R, G, B, out);
730  break;
731  default:
732  {
733  /* If this is reached, the caller should have called yuv2packedXinC
734  * instead. */
735  static int printed_error_message;
736  if (!printed_error_message) {
737  av_log(c, AV_LOG_ERROR,
738  "altivec_yuv2packedX doesn't support %s output\n",
740  printed_error_message = 1;
741  }
742  return;
743  }
744  }
745  }
746 
747  if (i < dstW) {
748  i -= 16;
749 
750  Y0 = RND;
751  Y1 = RND;
752  /* extract 16 coeffs from lumSrc */
753  for (j = 0; j < lumFilterSize; j++) {
754  X0 = vec_ld(0, &lumSrc[j][i]);
755  X1 = vec_ld(16, &lumSrc[j][i]);
756  Y0 = vec_mradds(X0, YCoeffs[j], Y0);
757  Y1 = vec_mradds(X1, YCoeffs[j], Y1);
758  }
759 
760  U = RND;
761  V = RND;
762  /* extract 8 coeffs from U,V */
763  for (j = 0; j < chrFilterSize; j++) {
764  X = vec_ld(0, &chrUSrc[j][i / 2]);
765  U = vec_mradds(X, CCoeffs[j], U);
766  X = vec_ld(0, &chrVSrc[j][i / 2]);
767  V = vec_mradds(X, CCoeffs[j], V);
768  }
769 
770  /* scale and clip signals */
771  Y0 = vec_sra(Y0, SCL);
772  Y1 = vec_sra(Y1, SCL);
773  U = vec_sra(U, SCL);
774  V = vec_sra(V, SCL);
775 
776  Y0 = vec_clip_s16(Y0);
777  Y1 = vec_clip_s16(Y1);
778  U = vec_clip_s16(U);
779  V = vec_clip_s16(V);
780 
781  /* now we have
782  * Y0 = y0 y1 y2 y3 y4 y5 y6 y7 Y1 = y8 y9 y10 y11 y12 y13 y14 y15
783  * U = u0 u1 u2 u3 u4 u5 u6 u7 V = v0 v1 v2 v3 v4 v5 v6 v7
784  *
785  * Y0 = y0 y1 y2 y3 y4 y5 y6 y7 Y1 = y8 y9 y10 y11 y12 y13 y14 y15
786  * U0 = u0 u0 u1 u1 u2 u2 u3 u3 U1 = u4 u4 u5 u5 u6 u6 u7 u7
787  * V0 = v0 v0 v1 v1 v2 v2 v3 v3 V1 = v4 v4 v5 v5 v6 v6 v7 v7
788  */
789 
790  U0 = vec_mergeh(U, U);
791  V0 = vec_mergeh(V, V);
792 
793  U1 = vec_mergel(U, U);
794  V1 = vec_mergel(V, V);
795 
796  cvtyuvtoRGB(c, Y0, U0, V0, &R0, &G0, &B0);
797  cvtyuvtoRGB(c, Y1, U1, V1, &R1, &G1, &B1);
798 
799  R = vec_packclp(R0, R1);
800  G = vec_packclp(G0, G1);
801  B = vec_packclp(B0, B1);
802 
803  nout = (vector unsigned char *) scratch;
804  switch (target) {
805  case AV_PIX_FMT_ABGR:
806  out_abgr(R, G, B, nout);
807  break;
808  case AV_PIX_FMT_BGRA:
809  out_bgra(R, G, B, nout);
810  break;
811  case AV_PIX_FMT_RGBA:
812  out_rgba(R, G, B, nout);
813  break;
814  case AV_PIX_FMT_ARGB:
815  out_argb(R, G, B, nout);
816  break;
817  case AV_PIX_FMT_RGB24:
818  out_rgb24(R, G, B, nout);
819  break;
820  case AV_PIX_FMT_BGR24:
821  out_bgr24(R, G, B, nout);
822  break;
823  default:
824  /* Unreachable, I think. */
825  av_log(c, AV_LOG_ERROR,
826  "altivec_yuv2packedX doesn't support %s output\n",
828  return;
829  }
830 
831  memcpy(&((uint32_t *) dest)[i], scratch, (dstW - i) / 4);
832  }
833 }
834 
835 #define YUV2PACKEDX_WRAPPER(suffix, pixfmt) \
836 void ff_yuv2 ## suffix ## _X_altivec(SwsContext *c, \
837  const int16_t *lumFilter, \
838  const int16_t **lumSrc, \
839  int lumFilterSize, \
840  const int16_t *chrFilter, \
841  const int16_t **chrUSrc, \
842  const int16_t **chrVSrc, \
843  int chrFilterSize, \
844  const int16_t **alpSrc, \
845  uint8_t *dest, int dstW, int dstY) \
846 { \
847  ff_yuv2packedX_altivec(c, lumFilter, lumSrc, lumFilterSize, \
848  chrFilter, chrUSrc, chrVSrc, \
849  chrFilterSize, alpSrc, \
850  dest, dstW, dstY, pixfmt); \
851 }
852 
packed YUV 4:2:2, 16bpp, Cb Y0 Cr Y1
Definition: pixfmt.h:82
const char * sws_format_name(enum AVPixelFormat format)
Definition: utils.c:189
#define YUV2PACKEDX_WRAPPER(suffix, pixfmt)
av_cold void ff_yuv2rgb_init_tables_altivec(SwsContext *c, const int inv_table[4], int brightness, int contrast, int saturation)
#define out_bgra(a, b, c, ptr)
#define B
Definition: dsputil.c:1897
packed RGB 8:8:8, 24bpp, RGBRGB...
Definition: pixfmt.h:67
static const vector unsigned char demux_v
#define vec_clip_s16(x)
Macro definitions for various function/variable attributes.
int srcH
Height of source luma/alpha planes.
#define B1
Definition: faandct.c:42
uint8_t
#define out_abgr(a, b, c, ptr)
unsigned char ubyte
static int altivec_uyvy_rgb32(SwsContext *c, const unsigned char **in, int *instrides, int srcSliceY, int srcSliceH, unsigned char **oplanes, int *outstrides)
packed ABGR 8:8:8:8, 32bpp, ABGRABGR...
Definition: pixfmt.h:95
av_cold SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c)
#define R
Definition: dsputil.c:1899
external api for the swscale stuff
enum AVPixelFormat dstFormat
Destination pixel format.
static av_always_inline void ff_yuv2packedX_altivec(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t *dest, int dstW, int dstY, enum AVPixelFormat target)
#define R0(v, w, x, y, z, i)
Definition: sha.c:56
packed BGRA 8:8:8:8, 32bpp, BGRABGRA...
Definition: pixfmt.h:96
planar YUV 4:2:0, 12bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (firs...
Definition: pixfmt.h:90
void av_log(void *avcl, int level, const char *fmt,...)
Definition: log.c:146
static void cvtyuvtoRGB(SwsContext *c, vector signed short Y, vector signed short U, vector signed short V, vector signed short *R, vector signed short *G, vector signed short *B)
packed ARGB 8:8:8:8, 32bpp, ARGBARGB...
Definition: pixfmt.h:93
static const vector unsigned char perm_rgb_2
packed RGBA 8:8:8:8, 32bpp, RGBARGBA...
Definition: pixfmt.h:94
Definition: vf_drawbox.c:36
as above, but U and V bytes are swapped
Definition: pixfmt.h:91
static const vector unsigned char demux_y
static const vector unsigned char perm_rgb_3
#define DEFCSP420_CVT(name, out_pixels)
static const vector unsigned char perm_rgb_0
packed RGB 8:8:8, 24bpp, BGRBGR...
Definition: pixfmt.h:68
#define R1
Definition: simple_idct.c:155
NULL
Definition: eval.c:52
static const vector unsigned char demux_u
#define out_rgb24(a, b, c, ptr)
#define out_argb(a, b, c, ptr)
planar YUV 4:1:0, 9bpp, (1 Cr & Cb sample per 4x4 Y samples)
Definition: pixfmt.h:71
int(* SwsFunc)(struct SwsContext *context, const uint8_t *src[], int srcStride[], int srcSliceY, int srcSliceH, uint8_t *dst[], int dstStride[])
#define vec_packclp(x, y)
signed char sbyte
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:25
Definition: vf_drawbox.c:36
#define out_bgr24(a, b, c, ptr)
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:65
Y , 8bpp.
Definition: pixfmt.h:73
enum AVPixelFormat srcFormat
Source pixel format.
#define G
Definition: dsputil.c:1898
#define out_rgba(a, b, c, ptr)
static const vector unsigned char perm_rgb_1
int srcW
Width of source luma/alpha planes.
AVPixelFormat
Pixel format.
Definition: pixfmt.h:63
#define B0
Definition: faandct.c:41