aboutsummaryrefslogtreecommitdiffstats
path: root/libswscale/ppc/yuv2yuv_altivec.c
blob: 08545b3c5d7950ad5d688f56a7bd65cef3300dd0 (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
/*
 * AltiVec-enhanced yuv-to-yuv conversion routines.
 *
 * 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
 */

#include <inttypes.h>

#include "config.h"
#include "libavutil/attributes.h"
#include "libavutil/cpu.h"
#include "libswscale/swscale.h"
#include "libswscale/swscale_internal.h"

#if HAVE_ALTIVEC

static 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;

    /* 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 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;

    /* 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;
}

#endif /* HAVE_ALTIVEC */

av_cold void ff_get_unscaled_swscale_ppc(SwsContext *c)
{
#if HAVE_ALTIVEC
    if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
        return;

    if (!(c->srcW & 15) && !(c->flags & SWS_BITEXACT) &&
        c->srcFormat == AV_PIX_FMT_YUV420P) {
        enum AVPixelFormat dstFormat = c->dstFormat;

        // unscaled YV12 -> packed YUV, we want speed
        if (dstFormat == AV_PIX_FMT_YUYV422)
            c->swscale = yv12toyuy2_unscaled_altivec;
        else if (dstFormat == AV_PIX_FMT_UYVY422)
            c->swscale = yv12touyvy_unscaled_altivec;
    }
#endif /* HAVE_ALTIVEC */
}