int_altivec.c 4.17 KB
Newer Older
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
/*
 * Copyright (c) 2007 Luca Barbato <lu_zero@gentoo.org>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg 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.
 *
 * FFmpeg 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 FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
 ** @file int_altivec.c
 ** integer misc ops.
 **/

26
#include "libavcodec/dsputil.h"
27 28 29 30 31

#include "gcc_fixes.h"

#include "dsputil_altivec.h"

32 33
#include "types_altivec.h"

34 35
static int ssd_int8_vs_int16_altivec(const int8_t *pix1, const int16_t *pix2,
                                     int size) {
36 37 38 39 40
    int i, size16;
    vector signed char vpix1;
    vector signed short vpix2, vdiff, vpix1l,vpix1h;
    union { vector signed int vscore;
            int32_t score[4];
41
          } u;
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
    u.vscore = vec_splat_s32(0);
//
//XXX lazy way, fix it later

#define vec_unaligned_load(b) \
    vec_perm(vec_ld(0,b),vec_ld(15,b),vec_lvsl(0, b));

    size16 = size >> 4;
    while(size16) {
//        score += (pix1[i]-pix2[i])*(pix1[i]-pix2[i]);
        //load pix1 and the first batch of pix2

        vpix1 = vec_unaligned_load(pix1);
        vpix2 = vec_unaligned_load(pix2);
        pix2 += 8;
        //unpack
        vpix1h = vec_unpackh(vpix1);
        vdiff  = vec_sub(vpix1h, vpix2);
        vpix1l = vec_unpackl(vpix1);
        // load another batch from pix2
        vpix2 = vec_unaligned_load(pix2);
        u.vscore = vec_msum(vdiff, vdiff, u.vscore);
        vdiff  = vec_sub(vpix1l, vpix2);
        u.vscore = vec_msum(vdiff, vdiff, u.vscore);
        pix1 += 16;
        pix2 += 8;
        size16--;
    }
    u.vscore = vec_sums(u.vscore, vec_splat_s32(0));

    size %= 16;
    for (i = 0; i < size; i++) {
        u.score[3] += (pix1[i]-pix2[i])*(pix1[i]-pix2[i]);
    }
    return u.score[3];
}

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
static void add_int16_altivec(int16_t * v1, int16_t * v2, int order)
{
    int i;
    register vec_s16_t vec, *pv;

    for(i = 0; i < order; i += 8){
        pv = (vec_s16_t*)v2;
        vec = vec_perm(pv[0], pv[1], vec_lvsl(0, v2));
        vec_st(vec_add(vec_ld(0, v1), vec), 0, v1);
        v1 += 8;
        v2 += 8;
    }
}

static void sub_int16_altivec(int16_t * v1, int16_t * v2, int order)
{
    int i;
    register vec_s16_t vec, *pv;

    for(i = 0; i < order; i += 8){
        pv = (vec_s16_t*)v2;
        vec = vec_perm(pv[0], pv[1], vec_lvsl(0, v2));
        vec_st(vec_sub(vec_ld(0, v1), vec), 0, v1);
        v1 += 8;
        v2 += 8;
    }
}

static int32_t scalarproduct_int16_altivec(int16_t * v1, int16_t * v2, int order, const int shift)
{
    int i;
    LOAD_ZERO;
    register vec_s16_t vec1, *pv;
    register vec_s32_t res = vec_splat_s32(0), t;
    register vec_u32_t shifts;
    DECLARE_ALIGNED_16(int32_t, ires);

    shifts = zero_u32v;
    if(shift & 0x10) shifts = vec_add(shifts, vec_sl(vec_splat_u32(0x08), vec_splat_u32(0x1)));
    if(shift & 0x08) shifts = vec_add(shifts, vec_splat_u32(0x08));
    if(shift & 0x04) shifts = vec_add(shifts, vec_splat_u32(0x04));
    if(shift & 0x02) shifts = vec_add(shifts, vec_splat_u32(0x02));
    if(shift & 0x01) shifts = vec_add(shifts, vec_splat_u32(0x01));

    for(i = 0; i < order; i += 8){
        pv = (vec_s16_t*)v1;
        vec1 = vec_perm(pv[0], pv[1], vec_lvsl(0, v1));
        t = vec_msum(vec1, vec_ld(0, v2), zero_s32v);
        t = vec_sr(t, shifts);
        res = vec_sums(t, res);
        v1 += 8;
        v2 += 8;
    }
    res = vec_splat(res, 3);
    vec_ste(res, 0, &ires);
    return ires;
}

137 138 139
void int_init_altivec(DSPContext* c, AVCodecContext *avctx)
{
    c->ssd_int8_vs_int16 = ssd_int8_vs_int16_altivec;
140 141 142
    c->add_int16 = add_int16_altivec;
    c->sub_int16 = sub_int16_altivec;
    c->scalarproduct_int16 = scalarproduct_int16_altivec;
143
}