ssd_int8_vs_int16_altivec, not completely benchmarkedwith svq1

Originally committed as revision 8706 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Luca Barbato 2007-04-10 09:47:37 +00:00
parent 6db140ddb4
commit 1f1aadc37c
3 changed files with 82 additions and 0 deletions

View File

@ -390,6 +390,7 @@ OBJS-$(TARGET_ALTIVEC) += ppc/dsputil_altivec.o \
ppc/gmc_altivec.o \
ppc/fdct_altivec.o \
ppc/float_altivec.o \
ppc/int_altivec.o \
ifeq ($(TARGET_ALTIVEC),yes)
OBJS-$(CONFIG_H264_DECODER) += ppc/h264_altivec.o

View File

@ -39,6 +39,7 @@ void dsputil_init_altivec(DSPContext* c, AVCodecContext *avctx);
void vc1dsp_init_altivec(DSPContext* c, AVCodecContext *avctx);
void snow_init_altivec(DSPContext* c, AVCodecContext *avctx);
void float_init_altivec(DSPContext* c, AVCodecContext *avctx);
void int_init_altivec(DSPContext* c, AVCodecContext *avctx);
#endif
@ -284,6 +285,7 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
if(ENABLE_VC1_DECODER || ENABLE_WMV3_DECODER)
vc1dsp_init_altivec(c, avctx);
float_init_altivec(c, avctx);
int_init_altivec(c, avctx);
c->gmc1 = gmc1_altivec;
#ifdef CONFIG_ENCODERS

View File

@ -0,0 +1,79 @@
/*
* 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.
**/
#include "../dsputil.h"
#include "gcc_fixes.h"
#include "dsputil_altivec.h"
static int ssd_int8_vs_int16_altivec(int8_t *pix1, int16_t *pix2, int size) {
int i, size16;
vector signed char vpix1;
vector signed short vpix2, vdiff, vpix1l,vpix1h;
union { vector signed int vscore;
int32_t score[4];
} u;
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];
}
void int_init_altivec(DSPContext* c, AVCodecContext *avctx)
{
c->ssd_int8_vs_int16 = ssd_int8_vs_int16_altivec;
}