mirror of
https://git.videolan.org/git/ffmpeg.git
synced 2024-08-25 10:51:49 +02:00
![Michael Niedermayer](/assets/img/avatar_default.png)
* qatar/master: Fix even more missing includes after the common.h removal build: Factor out rangecoder dependencies to CONFIG_RANGECODER build: Factor out error resilience dependencies to CONFIG_ERROR_RESILIENCE x86: avcodec: Consistently name all init files Add more missing includes after removing the implicit common.h Add some more missing includes after removing the implicit common.h Don't include common.h from avutil.h rtmp: Automatically compute the hash for SWFVerification Conflicts: configure doc/APIchanges doc/examples/decoding_encoding.c libavcodec/Makefile libavcodec/assdec.c libavcodec/audio_frame_queue.c libavcodec/avpacket.c libavcodec/dv_profile.c libavcodec/dwt.c libavcodec/libtheoraenc.c libavcodec/rawdec.c libavcodec/rv40dsp.c libavcodec/tiff.c libavcodec/tiffenc.c libavcodec/v210dec.h libavcodec/vc1dsp.c libavcodec/x86/Makefile libavfilter/asrc_anullsrc.c libavfilter/avfilter.c libavfilter/buffer.c libavfilter/formats.c libavfilter/vf_ass.c libavfilter/vf_drawtext.c libavfilter/vf_fade.c libavfilter/vf_select.c libavfilter/video.c libavfilter/vsrc_testsrc.c libavformat/version.h libavutil/audioconvert.c libavutil/error.h libavutil/version.h Merged-by: Michael Niedermayer <michaelni@gmx.at>
132 lines
4.7 KiB
C
132 lines
4.7 KiB
C
/*
|
|
* GMC (Global Motion Compensation)
|
|
* AltiVec-enabled
|
|
* Copyright (c) 2003 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 "libavutil/mem.h"
|
|
#include "libavutil/ppc/types_altivec.h"
|
|
#include "libavutil/ppc/util_altivec.h"
|
|
#include "libavcodec/dsputil.h"
|
|
#include "dsputil_altivec.h"
|
|
|
|
/*
|
|
altivec-enhanced gmc1. ATM this code assume stride is a multiple of 8,
|
|
to preserve proper dst alignment.
|
|
*/
|
|
void ff_gmc1_altivec(uint8_t *dst /* align 8 */, uint8_t *src /* align1 */, int stride, int h, int x16, int y16, int rounder)
|
|
{
|
|
const DECLARE_ALIGNED(16, unsigned short, rounder_a) = rounder;
|
|
const DECLARE_ALIGNED(16, unsigned short, ABCD)[8] =
|
|
{
|
|
(16-x16)*(16-y16), /* A */
|
|
( x16)*(16-y16), /* B */
|
|
(16-x16)*( y16), /* C */
|
|
( x16)*( y16), /* D */
|
|
0, 0, 0, 0 /* padding */
|
|
};
|
|
register const vector unsigned char vczero = (const vector unsigned char)vec_splat_u8(0);
|
|
register const vector unsigned short vcsr8 = (const vector unsigned short)vec_splat_u16(8);
|
|
register vector unsigned char dstv, dstv2, src_0, src_1, srcvA, srcvB, srcvC, srcvD;
|
|
register vector unsigned short Av, Bv, Cv, Dv, rounderV, tempA, tempB, tempC, tempD;
|
|
int i;
|
|
unsigned long dst_odd = (unsigned long)dst & 0x0000000F;
|
|
unsigned long src_really_odd = (unsigned long)src & 0x0000000F;
|
|
|
|
tempA = vec_ld(0, (const unsigned short*)ABCD);
|
|
Av = vec_splat(tempA, 0);
|
|
Bv = vec_splat(tempA, 1);
|
|
Cv = vec_splat(tempA, 2);
|
|
Dv = vec_splat(tempA, 3);
|
|
|
|
rounderV = vec_splat((vec_u16)vec_lde(0, &rounder_a), 0);
|
|
|
|
// we'll be able to pick-up our 9 char elements
|
|
// at src from those 32 bytes
|
|
// we load the first batch here, as inside the loop
|
|
// we can re-use 'src+stride' from one iteration
|
|
// as the 'src' of the next.
|
|
src_0 = vec_ld(0, src);
|
|
src_1 = vec_ld(16, src);
|
|
srcvA = vec_perm(src_0, src_1, vec_lvsl(0, src));
|
|
|
|
if (src_really_odd != 0x0000000F) {
|
|
// if src & 0xF == 0xF, then (src+1) is properly aligned
|
|
// on the second vector.
|
|
srcvB = vec_perm(src_0, src_1, vec_lvsl(1, src));
|
|
} else {
|
|
srcvB = src_1;
|
|
}
|
|
srcvA = vec_mergeh(vczero, srcvA);
|
|
srcvB = vec_mergeh(vczero, srcvB);
|
|
|
|
for(i=0; i<h; i++) {
|
|
dst_odd = (unsigned long)dst & 0x0000000F;
|
|
src_really_odd = (((unsigned long)src) + stride) & 0x0000000F;
|
|
|
|
dstv = vec_ld(0, dst);
|
|
|
|
// we we'll be able to pick-up our 9 char elements
|
|
// at src + stride from those 32 bytes
|
|
// then reuse the resulting 2 vectors srvcC and srcvD
|
|
// as the next srcvA and srcvB
|
|
src_0 = vec_ld(stride + 0, src);
|
|
src_1 = vec_ld(stride + 16, src);
|
|
srcvC = vec_perm(src_0, src_1, vec_lvsl(stride + 0, src));
|
|
|
|
if (src_really_odd != 0x0000000F) {
|
|
// if src & 0xF == 0xF, then (src+1) is properly aligned
|
|
// on the second vector.
|
|
srcvD = vec_perm(src_0, src_1, vec_lvsl(stride + 1, src));
|
|
} else {
|
|
srcvD = src_1;
|
|
}
|
|
|
|
srcvC = vec_mergeh(vczero, srcvC);
|
|
srcvD = vec_mergeh(vczero, srcvD);
|
|
|
|
|
|
// OK, now we (finally) do the math :-)
|
|
// those four instructions replaces 32 int muls & 32 int adds.
|
|
// isn't AltiVec nice ?
|
|
tempA = vec_mladd((vector unsigned short)srcvA, Av, rounderV);
|
|
tempB = vec_mladd((vector unsigned short)srcvB, Bv, tempA);
|
|
tempC = vec_mladd((vector unsigned short)srcvC, Cv, tempB);
|
|
tempD = vec_mladd((vector unsigned short)srcvD, Dv, tempC);
|
|
|
|
srcvA = srcvC;
|
|
srcvB = srcvD;
|
|
|
|
tempD = vec_sr(tempD, vcsr8);
|
|
|
|
dstv2 = vec_pack(tempD, (vector unsigned short)vczero);
|
|
|
|
if (dst_odd) {
|
|
dstv2 = vec_perm(dstv, dstv2, vcprm(0,1,s0,s1));
|
|
} else {
|
|
dstv2 = vec_perm(dstv, dstv2, vcprm(s0,s1,2,3));
|
|
}
|
|
|
|
vec_st(dstv2, 0, dst);
|
|
|
|
dst += stride;
|
|
src += stride;
|
|
}
|
|
}
|