yuv2yuv_altivec.c
Go to the documentation of this file.
1 /*
2  * AltiVec-enhanced yuv-to-yuv convertion routines.
3  *
4  * Copyright (C) 2004 Romain Dolbeau <romain@dolbeau.org>
5  * based on the equivalent C code in swscale.c
6  *
7  * This file is part of Libav.
8  *
9  * Libav is free software; you can redistribute it and/or
10  * modify it under the terms of the GNU Lesser General Public
11  * License as published by the Free Software Foundation; either
12  * version 2.1 of the License, or (at your option) any later version.
13  *
14  * Libav is distributed in the hope that it will be useful,
15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17  * Lesser General Public License for more details.
18  *
19  * You should have received a copy of the GNU Lesser General Public
20  * License along with Libav; if not, write to the Free Software
21  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22  */
23 
24 #include <inttypes.h>
25 #include "config.h"
26 #include "libswscale/swscale.h"
28 #include "libavutil/cpu.h"
29 
30 static int yv12toyuy2_unscaled_altivec(SwsContext *c, const uint8_t* src[],
31  int srcStride[], int srcSliceY,
32  int srcSliceH, uint8_t* dstParam[],
33  int dstStride_a[])
34 {
35  uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
36  // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
37  const uint8_t *ysrc = src[0];
38  const uint8_t *usrc = src[1];
39  const uint8_t *vsrc = src[2];
40  const int width = c->srcW;
41  const int height = srcSliceH;
42  const int lumStride = srcStride[0];
43  const int chromStride = srcStride[1];
44  const int dstStride = dstStride_a[0];
45  const vector unsigned char yperm = vec_lvsl(0, ysrc);
46  const int vertLumPerChroma = 2;
47  register unsigned int y;
48 
49  /* This code assumes:
50 
51  1) dst is 16 bytes-aligned
52  2) dstStride is a multiple of 16
53  3) width is a multiple of 16
54  4) lum & chrom stride are multiples of 8
55  */
56 
57  for (y=0; y<height; y++) {
58  int i;
59  for (i = 0; i < width - 31; i+= 32) {
60  const unsigned int j = i >> 1;
61  vector unsigned char v_yA = vec_ld(i, ysrc);
62  vector unsigned char v_yB = vec_ld(i + 16, ysrc);
63  vector unsigned char v_yC = vec_ld(i + 32, ysrc);
64  vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
65  vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
66  vector unsigned char v_uA = vec_ld(j, usrc);
67  vector unsigned char v_uB = vec_ld(j + 16, usrc);
68  vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
69  vector unsigned char v_vA = vec_ld(j, vsrc);
70  vector unsigned char v_vB = vec_ld(j + 16, vsrc);
71  vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
72  vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
73  vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
74  vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
75  vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
76  vector unsigned char v_yuy2_2 = vec_mergeh(v_y2, v_uv_b);
77  vector unsigned char v_yuy2_3 = vec_mergel(v_y2, v_uv_b);
78  vec_st(v_yuy2_0, (i << 1), dst);
79  vec_st(v_yuy2_1, (i << 1) + 16, dst);
80  vec_st(v_yuy2_2, (i << 1) + 32, dst);
81  vec_st(v_yuy2_3, (i << 1) + 48, dst);
82  }
83  if (i < width) {
84  const unsigned int j = i >> 1;
85  vector unsigned char v_y1 = vec_ld(i, ysrc);
86  vector unsigned char v_u = vec_ld(j, usrc);
87  vector unsigned char v_v = vec_ld(j, vsrc);
88  vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
89  vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
90  vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
91  vec_st(v_yuy2_0, (i << 1), dst);
92  vec_st(v_yuy2_1, (i << 1) + 16, dst);
93  }
94  if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
95  usrc += chromStride;
96  vsrc += chromStride;
97  }
98  ysrc += lumStride;
99  dst += dstStride;
100  }
101 
102  return srcSliceH;
103 }
104 
105 static int yv12touyvy_unscaled_altivec(SwsContext *c, const uint8_t* src[],
106  int srcStride[], int srcSliceY,
107  int srcSliceH, uint8_t* dstParam[],
108  int dstStride_a[])
109 {
110  uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
111  // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
112  const uint8_t *ysrc = src[0];
113  const uint8_t *usrc = src[1];
114  const uint8_t *vsrc = src[2];
115  const int width = c->srcW;
116  const int height = srcSliceH;
117  const int lumStride = srcStride[0];
118  const int chromStride = srcStride[1];
119  const int dstStride = dstStride_a[0];
120  const int vertLumPerChroma = 2;
121  const vector unsigned char yperm = vec_lvsl(0, ysrc);
122  register unsigned int y;
123 
124  /* This code assumes:
125 
126  1) dst is 16 bytes-aligned
127  2) dstStride is a multiple of 16
128  3) width is a multiple of 16
129  4) lum & chrom stride are multiples of 8
130  */
131 
132  for (y=0; y<height; y++) {
133  int i;
134  for (i = 0; i < width - 31; i+= 32) {
135  const unsigned int j = i >> 1;
136  vector unsigned char v_yA = vec_ld(i, ysrc);
137  vector unsigned char v_yB = vec_ld(i + 16, ysrc);
138  vector unsigned char v_yC = vec_ld(i + 32, ysrc);
139  vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
140  vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
141  vector unsigned char v_uA = vec_ld(j, usrc);
142  vector unsigned char v_uB = vec_ld(j + 16, usrc);
143  vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
144  vector unsigned char v_vA = vec_ld(j, vsrc);
145  vector unsigned char v_vB = vec_ld(j + 16, vsrc);
146  vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
147  vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
148  vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
149  vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
150  vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
151  vector unsigned char v_uyvy_2 = vec_mergeh(v_uv_b, v_y2);
152  vector unsigned char v_uyvy_3 = vec_mergel(v_uv_b, v_y2);
153  vec_st(v_uyvy_0, (i << 1), dst);
154  vec_st(v_uyvy_1, (i << 1) + 16, dst);
155  vec_st(v_uyvy_2, (i << 1) + 32, dst);
156  vec_st(v_uyvy_3, (i << 1) + 48, dst);
157  }
158  if (i < width) {
159  const unsigned int j = i >> 1;
160  vector unsigned char v_y1 = vec_ld(i, ysrc);
161  vector unsigned char v_u = vec_ld(j, usrc);
162  vector unsigned char v_v = vec_ld(j, vsrc);
163  vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
164  vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
165  vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
166  vec_st(v_uyvy_0, (i << 1), dst);
167  vec_st(v_uyvy_1, (i << 1) + 16, dst);
168  }
169  if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
170  usrc += chromStride;
171  vsrc += chromStride;
172  }
173  ysrc += lumStride;
174  dst += dstStride;
175  }
176  return srcSliceH;
177 }
178 
180 {
181  if ((av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) && !(c->srcW & 15) &&
182  !(c->flags & SWS_BITEXACT) && c->srcFormat == PIX_FMT_YUV420P) {
183  enum PixelFormat dstFormat = c->dstFormat;
184 
185  // unscaled YV12 -> packed YUV, we want speed
186  if (dstFormat == PIX_FMT_YUYV422)
188  else if (dstFormat == PIX_FMT_UYVY422)
190  }
191 }