ssd_int8_vs_int16_altivec, not completely benchmarkedwith svq1
[libav.git] / libavcodec / ppc / int_altivec.c
1 /*
2 * Copyright (c) 2007 Luca Barbato <lu_zero@gentoo.org>
3 *
4 * This file is part of FFmpeg.
5 *
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21 /**
22 ** @file int_altivec.c
23 ** integer misc ops.
24 **/
25
26 #include "../dsputil.h"
27
28 #include "gcc_fixes.h"
29
30 #include "dsputil_altivec.h"
31
32 static int ssd_int8_vs_int16_altivec(int8_t *pix1, int16_t *pix2, int size) {
33 int i, size16;
34 vector signed char vpix1;
35 vector signed short vpix2, vdiff, vpix1l,vpix1h;
36 union { vector signed int vscore;
37 int32_t score[4];
38 } u;
39 u.vscore = vec_splat_s32(0);
40 //
41 //XXX lazy way, fix it later
42
43 #define vec_unaligned_load(b) \
44 vec_perm(vec_ld(0,b),vec_ld(15,b),vec_lvsl(0, b));
45
46 size16 = size >> 4;
47 while(size16) {
48 // score += (pix1[i]-pix2[i])*(pix1[i]-pix2[i]);
49 //load pix1 and the first batch of pix2
50
51 vpix1 = vec_unaligned_load(pix1);
52 vpix2 = vec_unaligned_load(pix2);
53 pix2 += 8;
54 //unpack
55 vpix1h = vec_unpackh(vpix1);
56 vdiff = vec_sub(vpix1h, vpix2);
57 vpix1l = vec_unpackl(vpix1);
58 // load another batch from pix2
59 vpix2 = vec_unaligned_load(pix2);
60 u.vscore = vec_msum(vdiff, vdiff, u.vscore);
61 vdiff = vec_sub(vpix1l, vpix2);
62 u.vscore = vec_msum(vdiff, vdiff, u.vscore);
63 pix1 += 16;
64 pix2 += 8;
65 size16--;
66 }
67 u.vscore = vec_sums(u.vscore, vec_splat_s32(0));
68
69 size %= 16;
70 for (i = 0; i < size; i++) {
71 u.score[3] += (pix1[i]-pix2[i])*(pix1[i]-pix2[i]);
72 }
73 return u.score[3];
74 }
75
76 void int_init_altivec(DSPContext* c, AVCodecContext *avctx)
77 {
78 c->ssd_int8_vs_int16 = ssd_int8_vs_int16_altivec;
79 }