pixblockdsp_altivec.c

Go to the documentation of this file.

/*
 * Copyright (c) 2002 Brian Foley
 * Copyright (c) 2002 Dieter Shirley
 * Copyright (c) 2003-2004 Romain Dolbeau <romain@dolbeau.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
 */
 
#include "config.h"
 
#include "libavutil/attributes.h"
#include "libavutil/ppc/util_altivec.h"
 
#include "libavcodec/pixblockdsp.h"
 
#include "pixblockdsp_altivec.h"
 
#if HAVE_VSX
static void get_pixels_altivec(int16_t *restrict block, const uint8_t *pixels,
                               ptrdiff_t stride)
{
    int i;
    vector unsigned char perm =
        (vector unsigned char) {0x00,0x10, 0x01,0x11,0x02,0x12,0x03,0x13,\
            0x04,0x14,0x05,0x15,0x06,0x16,0x07,0x17};
    const vector unsigned char zero =
        (const vector unsigned char) vec_splat_u8(0);
 
    for (i = 0; i < 8; i++) {
        /* Read potentially unaligned pixels.
         * We're reading 16 pixels, and actually only want 8,
         * but we simply ignore the extras. */
        vector unsigned char bytes = vec_vsx_ld(0, pixels);
 
        // Convert the bytes into shorts.
        //vector signed short shorts = (vector signed short) vec_perm(zero, bytes, perm);
        vector signed short shorts = (vector signed short) vec_perm(bytes, zero, perm);
 
        // Save the data to the block, we assume the block is 16-byte aligned.
        vec_vsx_st(shorts, i * 16, (vector signed short *) block);
 
        pixels += stride;
    }
}
#else
static void get_pixels_altivec(int16_t *restrict block, const uint8_t *pixels,
                               ptrdiff_t stride)
{
    int i;
    const vec_u8 zero = (const vec_u8)vec_splat_u8(0);
 
    for (i = 0; i < 8; i++) {
        vec_u8 perm = vec_lvsl(0, pixels);
        /* Read potentially unaligned pixels.
         * We're reading 16 pixels, and actually only want 8,
         * but we simply ignore the extras. */
        vec_u8 pixl = vec_ld(0, pixels);
        vec_u8 pixr = vec_ld(7, pixels);
        vec_u8 bytes = vec_perm(pixl, pixr, perm);
 
        // Convert the bytes into shorts.
        vec_s16 shorts = (vec_s16)vec_mergeh(zero, bytes);
 
        // Save the data to the block, we assume the block is 16-byte aligned.
        vec_st(shorts, i * 16, (vec_s16 *)block);
 
        pixels += stride;
    }
}
 
#endif /* HAVE_VSX */
 
#if HAVE_VSX
static void diff_pixels_altivec(int16_t *restrict block, const uint8_t *s1,
                                const uint8_t *s2, ptrdiff_t stride)
{
  int i;
  const vector unsigned char zero =
    (const vector unsigned char) vec_splat_u8(0);
  vector signed short shorts1, shorts2;
 
  for (i = 0; i < 4; i++) {
    /* Read potentially unaligned pixels.
     * We're reading 16 pixels, and actually only want 8,
     * but we simply ignore the extras. */
    vector unsigned char bytes = vec_vsx_ld(0,  s1);
 
    // Convert the bytes into shorts.
    shorts1 = (vector signed short) vec_mergeh(bytes, zero);
 
    // Do the same for the second block of pixels.
    bytes =vec_vsx_ld(0,  s2);
 
    // Convert the bytes into shorts.
    shorts2 = (vector signed short) vec_mergeh(bytes, zero);
 
    // Do the subtraction.
    shorts1 = vec_sub(shorts1, shorts2);
 
    // Save the data to the block, we assume the block is 16-byte aligned.
    vec_vsx_st(shorts1, 0, (vector signed short *) block);
 
    s1    += stride;
    s2    += stride;
    block += 8;
 
    /* The code below is a copy of the code above...
     * This is a manual unroll. */
 
    /* Read potentially unaligned pixels.
     * We're reading 16 pixels, and actually only want 8,
     * but we simply ignore the extras. */
    bytes = vec_vsx_ld(0,  s1);
 
    // Convert the bytes into shorts.
    shorts1 = (vector signed short) vec_mergeh(bytes, zero);
 
    // Do the same for the second block of pixels.
    bytes = vec_vsx_ld(0,  s2);
 
    // Convert the bytes into shorts.
    shorts2 = (vector signed short) vec_mergeh(bytes, zero);
 
    // Do the subtraction.
    shorts1 = vec_sub(shorts1, shorts2);
 
    // Save the data to the block, we assume the block is 16-byte aligned.
    vec_vsx_st(shorts1, 0, (vector signed short *) block);
 
    s1    += stride;
    s2    += stride;
    block += 8;
  }
}
#else
static void diff_pixels_altivec(int16_t *restrict block, const uint8_t *s1,
                                const uint8_t *s2, ptrdiff_t stride)
{
    int i;
    vec_u8 perm;
    const vec_u8 zero = (const vec_u8)vec_splat_u8(0);
    vec_s16 shorts1, shorts2;
 
    for (i = 0; i < 4; i++) {
        /* Read potentially unaligned pixels.
         * We're reading 16 pixels, and actually only want 8,
         * but we simply ignore the extras. */
        perm = vec_lvsl(0, s1);
        vec_u8 pixl  = vec_ld(0,  s1);
        vec_u8 pixr  = vec_ld(15, s1);
        vec_u8 bytes = vec_perm(pixl, pixr, perm);
 
        // Convert the bytes into shorts.
        shorts1 = (vec_s16)vec_mergeh(zero, bytes);
 
        // Do the same for the second block of pixels.
        perm = vec_lvsl(0, s2);
        pixl  = vec_ld(0,  s2);
        pixr  = vec_ld(15, s2);
        bytes = vec_perm(pixl, pixr, perm);
 
        // Convert the bytes into shorts.
        shorts2 = (vec_s16)vec_mergeh(zero, bytes);
 
        // Do the subtraction.
        shorts1 = vec_sub(shorts1, shorts2);
 
        // Save the data to the block, we assume the block is 16-byte aligned.
        vec_st(shorts1, 0, (vec_s16 *)block);
 
        s1    += stride;
        s2    += stride;
        block += 8;
 
        /* The code below is a copy of the code above...
         * This is a manual unroll. */
 
        /* Read potentially unaligned pixels.
         * We're reading 16 pixels, and actually only want 8,
         * but we simply ignore the extras. */
        perm = vec_lvsl(0, s1);
        pixl  = vec_ld(0,  s1);
        pixr  = vec_ld(15, s1);
        bytes = vec_perm(pixl, pixr, perm);
 
        // Convert the bytes into shorts.
        shorts1 = (vec_s16)vec_mergeh(zero, bytes);
 
        // Do the same for the second block of pixels.
        perm = vec_lvsl(0, s2);
        pixl  = vec_ld(0,  s2);
        pixr  = vec_ld(15, s2);
        bytes = vec_perm(pixl, pixr, perm);
 
        // Convert the bytes into shorts.
        shorts2 = (vec_s16)vec_mergeh(zero, bytes);
 
        // Do the subtraction.
        shorts1 = vec_sub(shorts1, shorts2);
 
        // Save the data to the block, we assume the block is 16-byte aligned.
        vec_st(shorts1, 0, (vec_s16 *)block);
 
        s1    += stride;
        s2    += stride;
        block += 8;
    }
}
 
#endif /* HAVE_VSX */
 
av_cold void ff_pixblockdsp_init_altivec(PixblockDSPContext *c,
                                         unsigned high_bit_depth)
{
    c->diff_pixels = diff_pixels_altivec;
 
    if (!high_bit_depth) {
        c->get_pixels = get_pixels_altivec;
    }
}