aboutsummaryrefslogtreecommitdiffstats
path: root/libswscale/ppc/swscale_altivec_template.c
blob: eee7bdd0a57cead4150aa81757dea4320362e724 (plain) (blame)
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
/*
 * AltiVec-enhanced yuv2yuvX
 *
 * Copyright (C) 2004 Romain Dolbeau <romain@dolbeau.org>
 * based on the equivalent C code in swscale.c
 *
 * This file is part of Libav.
 *
 * Libav is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * Libav 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
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with Libav; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#define vzero vec_splat_s32(0)

static inline void
altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW)
{
    register int i;
    vector unsigned int altivec_vectorShiftInt19 =
        vec_add(vec_splat_u32(10), vec_splat_u32(9));
    if ((unsigned long)dest % 16) {
        /* badly aligned store, we force store alignment */
        /* and will handle load misalignment on val w/ vec_perm */
        vector unsigned char perm1;
        vector signed int v1;
        for (i = 0 ; (i < dstW) &&
            (((unsigned long)dest + i) % 16) ; i++) {
                int t = val[i] >> 19;
                dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
        }
        perm1 = vec_lvsl(i << 2, val);
        v1 = vec_ld(i << 2, val);
        for ( ; i < (dstW - 15); i+=16) {
            int offset = i << 2;
            vector signed int v2 = vec_ld(offset + 16, val);
            vector signed int v3 = vec_ld(offset + 32, val);
            vector signed int v4 = vec_ld(offset + 48, val);
            vector signed int v5 = vec_ld(offset + 64, val);
            vector signed int v12 = vec_perm(v1, v2, perm1);
            vector signed int v23 = vec_perm(v2, v3, perm1);
            vector signed int v34 = vec_perm(v3, v4, perm1);
            vector signed int v45 = vec_perm(v4, v5, perm1);

            vector signed int vA = vec_sra(v12, altivec_vectorShiftInt19);
            vector signed int vB = vec_sra(v23, altivec_vectorShiftInt19);
            vector signed int vC = vec_sra(v34, altivec_vectorShiftInt19);
            vector signed int vD = vec_sra(v45, altivec_vectorShiftInt19);
            vector unsigned short vs1 = vec_packsu(vA, vB);
            vector unsigned short vs2 = vec_packsu(vC, vD);
            vector unsigned char vf = vec_packsu(vs1, vs2);
            vec_st(vf, i, dest);
            v1 = v5;
        }
    } else { // dest is properly aligned, great
        for (i = 0; i < (dstW - 15); i+=16) {
            int offset = i << 2;
            vector signed int v1 = vec_ld(offset, val);
            vector signed int v2 = vec_ld(offset + 16, val);
            vector signed int v3 = vec_ld(offset + 32, val);
            vector signed int v4 = vec_ld(offset + 48, val);
            vector signed int v5 = vec_sra(v1, altivec_vectorShiftInt19);
            vector signed int v6 = vec_sra(v2, altivec_vectorShiftInt19);
            vector signed int v7 = vec_sra(v3, altivec_vectorShiftInt19);
            vector signed int v8 = vec_sra(v4, altivec_vectorShiftInt19);
            vector unsigned short vs1 = vec_packsu(v5, v6);
            vector unsigned short vs2 = vec_packsu(v7, v8);
            vector unsigned char vf = vec_packsu(vs1, vs2);
            vec_st(vf, i, dest);
        }
    }
    for ( ; i < dstW ; i++) {
        int t = val[i] >> 19;
        dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
    }
}

static inline void
yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc,
                      int lumFilterSize, const int16_t *chrFilter,
                      const int16_t **chrUSrc, const int16_t **chrVSrc,
                      int chrFilterSize, uint8_t *dest, uint8_t *uDest,
                      uint8_t *vDest, int dstW, int chrDstW)
{
    const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)};
    register int i, j;
    {
        DECLARE_ALIGNED(16, int, val)[dstW];

        for (i = 0; i < (dstW -7); i+=4) {
            vec_st(vini, i << 2, val);
        }
        for (; i < dstW; i++) {
            val[i] = (1 << 18);
        }

        for (j = 0; j < lumFilterSize; j++) {
            vector signed short l1, vLumFilter = vec_ld(j << 1, lumFilter);
            vector unsigned char perm, perm0 = vec_lvsl(j << 1, lumFilter);
            vLumFilter = vec_perm(vLumFilter, vLumFilter, perm0);
            vLumFilter = vec_splat(vLumFilter, 0); // lumFilter[j] is loaded 8 times in vLumFilter

            perm = vec_lvsl(0, lumSrc[j]);
            l1 = vec_ld(0, lumSrc[j]);

            for (i = 0; i < (dstW - 7); i+=8) {
                int offset = i << 2;
                vector signed short l2 = vec_ld((i << 1) + 16, lumSrc[j]);

                vector signed int v1 = vec_ld(offset, val);
                vector signed int v2 = vec_ld(offset + 16, val);

                vector signed short ls = vec_perm(l1, l2, perm); // lumSrc[j][i] ... lumSrc[j][i+7]

                vector signed int i1 = vec_mule(vLumFilter, ls);
                vector signed int i2 = vec_mulo(vLumFilter, ls);

                vector signed int vf1 = vec_mergeh(i1, i2);
                vector signed int vf2 = vec_mergel(i1, i2); // lumSrc[j][i] * lumFilter[j] ... lumSrc[j][i+7] * lumFilter[j]

                vector signed int vo1 = vec_add(v1, vf1);
                vector signed int vo2 = vec_add(v2, vf2);

                vec_st(vo1, offset, val);
                vec_st(vo2, offset + 16, val);

                l1 = l2;
            }
            for ( ; i < dstW; i++) {
                val[i] += lumSrc[j][i] * lumFilter[j];
            }
        }
        altivec_packIntArrayToCharArray(val, dest, dstW);
    }
    if (uDest != 0) {
        DECLARE_ALIGNED(16, int, u)[chrDstW];
        DECLARE_ALIGNED(16, int, v)[chrDstW];

        for (i = 0; i < (chrDstW -7); i+=4) {
            vec_st(vini, i << 2, u);
            vec_st(vini, i << 2, v);
        }
        for (; i < chrDstW; i++) {
            u[i] = (1 << 18);
            v[i] = (1 << 18);
        }

        for (j = 0; j < chrFilterSize; j++) {
            vector signed short l1, l1_V, vChrFilter = vec_ld(j << 1, chrFilter);
            vector unsigned char perm, perm0 = vec_lvsl(j << 1, chrFilter);
            vChrFilter = vec_perm(vChrFilter, vChrFilter, perm0);
            vChrFilter = vec_splat(vChrFilter, 0); // chrFilter[j] is loaded 8 times in vChrFilter

            perm = vec_lvsl(0, chrUSrc[j]);
            l1 = vec_ld(0, chrUSrc[j]);
            l1_V = vec_ld(0, chrVSrc[j]);

            for (i = 0; i < (chrDstW - 7); i+=8) {
                int offset = i << 2;
                vector signed short l2 = vec_ld((i << 1) + 16, chrUSrc[j]);
                vector signed short l2_V = vec_ld((i << 1) + 16, chrVSrc[j]);

                vector signed int v1 = vec_ld(offset, u);
                vector signed int v2 = vec_ld(offset + 16, u);
                vector signed int v1_V = vec_ld(offset, v);
                vector signed int v2_V = vec_ld(offset + 16, v);

                vector signed short ls = vec_perm(l1, l2, perm); // chrUSrc[j][i] ... chrUSrc[j][i+7]
                vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrVSrc[j][i] ... chrVSrc[j][i]

                vector signed int i1 = vec_mule(vChrFilter, ls);
                vector signed int i2 = vec_mulo(vChrFilter, ls);
                vector signed int i1_V = vec_mule(vChrFilter, ls_V);
                vector signed int i2_V = vec_mulo(vChrFilter, ls_V);

                vector signed int vf1 = vec_mergeh(i1, i2);
                vector signed int vf2 = vec_mergel(i1, i2); // chrUSrc[j][i] * chrFilter[j] ... chrUSrc[j][i+7] * chrFilter[j]
                vector signed int vf1_V = vec_mergeh(i1_V, i2_V);
                vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrVSrc[j][i] * chrFilter[j] ... chrVSrc[j][i+7] * chrFilter[j]

                vector signed int vo1 = vec_add(v1, vf1);
                vector signed int vo2 = vec_add(v2, vf2);
                vector signed int vo1_V = vec_add(v1_V, vf1_V);
                vector signed int vo2_V = vec_add(v2_V, vf2_V);

                vec_st(vo1, offset, u);
                vec_st(vo2, offset + 16, u);
                vec_st(vo1_V, offset, v);
                vec_st(vo2_V, offset + 16, v);

                l1 = l2;
                l1_V = l2_V;
            }
            for ( ; i < chrDstW; i++) {
                u[i] += chrUSrc[j][i] * chrFilter[j];
                v[i] += chrVSrc[j][i] * chrFilter[j];
            }
        }
        altivec_packIntArrayToCharArray(u, uDest, chrDstW);
        altivec_packIntArrayToCharArray(v, vDest, chrDstW);
    }
}

static inline void hScale_altivec_real(int16_t *dst, int dstW,
                                       const uint8_t *src, int srcW,
                                       int xInc, const int16_t *filter,
                                       const int16_t *filterPos, int filterSize)
{
    register int i;
    DECLARE_ALIGNED(16, int, tempo)[4];

    if (filterSize % 4) {
        for (i=0; i<dstW; i++) {
            register int j;
            register int srcPos = filterPos[i];
            register int val = 0;
            for (j=0; j<filterSize; j++) {
                val += ((int)src[srcPos + j])*filter[filterSize*i + j];
            }
            dst[i] = FFMIN(val>>7, (1<<15)-1);
        }
    }
    else
    switch (filterSize) {
    case 4:
    {
    for (i=0; i<dstW; i++) {
        register int srcPos = filterPos[i];

        vector unsigned char src_v0 = vec_ld(srcPos, src);
        vector unsigned char src_v1, src_vF;
        vector signed short src_v, filter_v;
        vector signed int val_vEven, val_s;
        if ((((int)src + srcPos)% 16) > 12) {
            src_v1 = vec_ld(srcPos + 16, src);
        }
        src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));

        src_v = // vec_unpackh sign-extends...
            (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
        // now put our elements in the even slots
        src_v = vec_mergeh(src_v, (vector signed short)vzero);

        filter_v = vec_ld(i << 3, filter);
        // The 3 above is 2 (filterSize == 4) + 1 (sizeof(short) == 2).

        // The neat trick: We only care for half the elements,
        // high or low depending on (i<<3)%16 (it's 0 or 8 here),
        // and we're going to use vec_mule, so we choose
        // carefully how to "unpack" the elements into the even slots.
        if ((i << 3) % 16)
            filter_v = vec_mergel(filter_v, (vector signed short)vzero);
        else
            filter_v = vec_mergeh(filter_v, (vector signed short)vzero);

        val_vEven = vec_mule(src_v, filter_v);
        val_s = vec_sums(val_vEven, vzero);
        vec_st(val_s, 0, tempo);
        dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
    }
    }
    break;

    case 8:
    {
    for (i=0; i<dstW; i++) {
        register int srcPos = filterPos[i];

        vector unsigned char src_v0 = vec_ld(srcPos, src);
        vector unsigned char src_v1, src_vF;
        vector signed short src_v, filter_v;
        vector signed int val_v, val_s;
        if ((((int)src + srcPos)% 16) > 8) {
            src_v1 = vec_ld(srcPos + 16, src);
        }
        src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));

        src_v = // vec_unpackh sign-extends...
            (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
        filter_v = vec_ld(i << 4, filter);
        // the 4 above is 3 (filterSize == 8) + 1 (sizeof(short) == 2)

        val_v = vec_msums(src_v, filter_v, (vector signed int)vzero);
        val_s = vec_sums(val_v, vzero);
        vec_st(val_s, 0, tempo);
        dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
    }
    }
    break;

    case 16:
    {
        for (i=0; i<dstW; i++) {
            register int srcPos = filterPos[i];

            vector unsigned char src_v0 = vec_ld(srcPos, src);
            vector unsigned char src_v1 = vec_ld(srcPos + 16, src);
            vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));

            vector signed short src_vA = // vec_unpackh sign-extends...
                (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
            vector signed short src_vB = // vec_unpackh sign-extends...
                (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));

            vector signed short filter_v0 = vec_ld(i << 5, filter);
            vector signed short filter_v1 = vec_ld((i << 5) + 16, filter);
            // the 5 above are 4 (filterSize == 16) + 1 (sizeof(short) == 2)

            vector signed int val_acc = vec_msums(src_vA, filter_v0, (vector signed int)vzero);
            vector signed int val_v = vec_msums(src_vB, filter_v1, val_acc);

            vector signed int val_s = vec_sums(val_v, vzero);

            vec_st(val_s, 0, tempo);
            dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
        }
    }
    break;

    default:
    {
    for (i=0; i<dstW; i++) {
        register int j;
        register int srcPos = filterPos[i];

        vector signed int val_s, val_v = (vector signed int)vzero;
        vector signed short filter_v0R = vec_ld(i * 2 * filterSize, filter);
        vector unsigned char permF = vec_lvsl((i * 2 * filterSize), filter);

        vector unsigned char src_v0 = vec_ld(srcPos, src);
        vector unsigned char permS = vec_lvsl(srcPos, src);

        for (j = 0 ; j < filterSize - 15; j += 16) {
            vector unsigned char src_v1 = vec_ld(srcPos + j + 16, src);
            vector unsigned char src_vF = vec_perm(src_v0, src_v1, permS);

            vector signed short src_vA = // vec_unpackh sign-extends...
                (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
            vector signed short src_vB = // vec_unpackh sign-extends...
                (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));

            vector signed short filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
            vector signed short filter_v2R = vec_ld((i * 2 * filterSize) + (j * 2) + 32, filter);
            vector signed short filter_v0  = vec_perm(filter_v0R, filter_v1R, permF);
            vector signed short filter_v1  = vec_perm(filter_v1R, filter_v2R, permF);

            vector signed int val_acc = vec_msums(src_vA, filter_v0, val_v);
            val_v = vec_msums(src_vB, filter_v1, val_acc);

            filter_v0R = filter_v2R;
            src_v0 = src_v1;
        }

        if (j < filterSize-7) {
            // loading src_v0 is useless, it's already done above
            //vector unsigned char src_v0 = vec_ld(srcPos + j, src);
            vector unsigned char src_v1, src_vF;
            vector signed short src_v, filter_v1R, filter_v;
            if ((((int)src + srcPos)% 16) > 8) {
                src_v1 = vec_ld(srcPos + j + 16, src);
            }
            src_vF = vec_perm(src_v0, src_v1, permS);

            src_v = // vec_unpackh sign-extends...
                (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
            // loading filter_v0R is useless, it's already done above
            //vector signed short filter_v0R = vec_ld((i * 2 * filterSize) + j, filter);
            filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
            filter_v = vec_perm(filter_v0R, filter_v1R, permF);

            val_v = vec_msums(src_v, filter_v, val_v);
        }

        val_s = vec_sums(val_v, vzero);

        vec_st(val_s, 0, tempo);
        dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
    }

    }
    }
}

static inline int yv12toyuy2_unscaled_altivec(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
                                              int srcSliceH, uint8_t* dstParam[], int dstStride_a[])
{
    uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
    // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
    const uint8_t *ysrc = src[0];
    const uint8_t *usrc = src[1];
    const uint8_t *vsrc = src[2];
    const int width = c->srcW;
    const int height = srcSliceH;
    const int lumStride = srcStride[0];
    const int chromStride = srcStride[1];
    const int dstStride = dstStride_a[0];
    const vector unsigned char yperm = vec_lvsl(0, ysrc);
    const int vertLumPerChroma = 2;
    register unsigned int y;

    if (width&15) {
        yv12toyuy2(ysrc, usrc, vsrc, dst, c->srcW, srcSliceH, lumStride, chromStride, dstStride);
        return srcSliceH;
    }

    /* This code assumes:

    1) dst is 16 bytes-aligned
    2) dstStride is a multiple of 16
    3) width is a multiple of 16
    4) lum & chrom stride are multiples of 8
    */

    for (y=0; y<height; y++) {
        int i;
        for (i = 0; i < width - 31; i+= 32) {
            const unsigned int j = i >> 1;
            vector unsigned char v_yA = vec_ld(i, ysrc);
            vector unsigned char v_yB = vec_ld(i + 16, ysrc);
            vector unsigned char v_yC = vec_ld(i + 32, ysrc);
            vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
            vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
            vector unsigned char v_uA = vec_ld(j, usrc);
            vector unsigned char v_uB = vec_ld(j + 16, usrc);
            vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
            vector unsigned char v_vA = vec_ld(j, vsrc);
            vector unsigned char v_vB = vec_ld(j + 16, vsrc);
            vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
            vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
            vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
            vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
            vector unsigned char v_yuy2_2 = vec_mergeh(v_y2, v_uv_b);
            vector unsigned char v_yuy2_3 = vec_mergel(v_y2, v_uv_b);
            vec_st(v_yuy2_0, (i << 1), dst);
            vec_st(v_yuy2_1, (i << 1) + 16, dst);
            vec_st(v_yuy2_2, (i << 1) + 32, dst);
            vec_st(v_yuy2_3, (i << 1) + 48, dst);
        }
        if (i < width) {
            const unsigned int j = i >> 1;
            vector unsigned char v_y1 = vec_ld(i, ysrc);
            vector unsigned char v_u = vec_ld(j, usrc);
            vector unsigned char v_v = vec_ld(j, vsrc);
            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
            vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
            vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
            vec_st(v_yuy2_0, (i << 1), dst);
            vec_st(v_yuy2_1, (i << 1) + 16, dst);
        }
        if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
            usrc += chromStride;
            vsrc += chromStride;
        }
        ysrc += lumStride;
        dst += dstStride;
    }

    return srcSliceH;
}

static inline int yv12touyvy_unscaled_altivec(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
                                              int srcSliceH, uint8_t* dstParam[], int dstStride_a[])
{
    uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
    // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
    const uint8_t *ysrc = src[0];
    const uint8_t *usrc = src[1];
    const uint8_t *vsrc = src[2];
    const int width = c->srcW;
    const int height = srcSliceH;
    const int lumStride = srcStride[0];
    const int chromStride = srcStride[1];
    const int dstStride = dstStride_a[0];
    const int vertLumPerChroma = 2;
    const vector unsigned char yperm = vec_lvsl(0, ysrc);
    register unsigned int y;

    if (width&15) {
        yv12touyvy(ysrc, usrc, vsrc, dst, c->srcW, srcSliceH, lumStride, chromStride, dstStride);
        return srcSliceH;
    }

    /* This code assumes:

    1) dst is 16 bytes-aligned
    2) dstStride is a multiple of 16
    3) width is a multiple of 16
    4) lum & chrom stride are multiples of 8
    */

    for (y=0; y<height; y++) {
        int i;
        for (i = 0; i < width - 31; i+= 32) {
            const unsigned int j = i >> 1;
            vector unsigned char v_yA = vec_ld(i, ysrc);
            vector unsigned char v_yB = vec_ld(i + 16, ysrc);
            vector unsigned char v_yC = vec_ld(i + 32, ysrc);
            vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
            vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
            vector unsigned char v_uA = vec_ld(j, usrc);
            vector unsigned char v_uB = vec_ld(j + 16, usrc);
            vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
            vector unsigned char v_vA = vec_ld(j, vsrc);
            vector unsigned char v_vB = vec_ld(j + 16, vsrc);
            vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
            vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
            vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
            vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
            vector unsigned char v_uyvy_2 = vec_mergeh(v_uv_b, v_y2);
            vector unsigned char v_uyvy_3 = vec_mergel(v_uv_b, v_y2);
            vec_st(v_uyvy_0, (i << 1), dst);
            vec_st(v_uyvy_1, (i << 1) + 16, dst);
            vec_st(v_uyvy_2, (i << 1) + 32, dst);
            vec_st(v_uyvy_3, (i << 1) + 48, dst);
        }
        if (i < width) {
            const unsigned int j = i >> 1;
            vector unsigned char v_y1 = vec_ld(i, ysrc);
            vector unsigned char v_u = vec_ld(j, usrc);
            vector unsigned char v_v = vec_ld(j, vsrc);
            vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
            vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
            vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
            vec_st(v_uyvy_0, (i << 1), dst);
            vec_st(v_uyvy_1, (i << 1) + 16, dst);
        }
        if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
            usrc += chromStride;
            vsrc += chromStride;
        }
        ysrc += lumStride;
        dst += dstStride;
    }
    return srcSliceH;
}