generic-library/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
4ed5a73a7e
ffmpeg/libavcodec/x86/hpeldsp_rnd_template.c

203 lines
8.2 KiB
C
Raw Normal View History

Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
/*
Update dsputil- and SIMD-related comments to match reality more closely
2014-01-29 13:12:59 +01:00
* SIMD-optimized halfpel functions are compiled twice for rnd/no_rnd
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
* Copyright (c) 2000, 2001 Fabrice Bellard
* Copyright (c) 2003-2004 Michael Niedermayer <michaelni@gmx.at>
*
* MMX optimization by Nick Kurshev <nickols_k@mail.ru>
* mostly rewritten by Michael Niedermayer <michaelni@gmx.at>
* and improved by Zdenek Kabelac <kabi@users.sf.net>
*
* 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
*/
Add missing headers to make template files compile (more) standalone
2014-01-09 10:27:20 +01:00
#include <stddef.h>
#include <stdint.h>
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
// put_pixels
avcodec/x86/hpeldsp_rnd_template: silence -Wunused-function on --disable-mmx This silences some of the -Wunused-function warnings when compiled with --disable-mmx, e.g http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx. Header guards are too brittle and ugly for this case. Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2015-09-19 17:38:59 +02:00
av_unused static void DEF(put, pixels8_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
{
MOVQ_BFE(mm6);
__asm__ volatile(
"lea (%3, %3), %%"REG_a" \n\t"
".p2align 3 \n\t"
"1: \n\t"
"movq (%1), %%mm0 \n\t"
"movq 1(%1), %%mm1 \n\t"
"movq (%1, %3), %%mm2 \n\t"
"movq 1(%1, %3), %%mm3 \n\t"
PAVGBP(%%mm0, %%mm1, %%mm4, %%mm2, %%mm3, %%mm5)
"movq %%mm4, (%2) \n\t"
"movq %%mm5, (%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"movq (%1), %%mm0 \n\t"
"movq 1(%1), %%mm1 \n\t"
"movq (%1, %3), %%mm2 \n\t"
"movq 1(%1, %3), %%mm3 \n\t"
PAVGBP(%%mm0, %%mm1, %%mm4, %%mm2, %%mm3, %%mm5)
"movq %%mm4, (%2) \n\t"
"movq %%mm5, (%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}
avcodec/x86/hpeldsp_rnd_template: silence -Wunused-function on --disable-mmx This silences some of the -Wunused-function warnings when compiled with --disable-mmx, e.g http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx. Header guards are too brittle and ugly for this case. Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2015-09-19 17:38:59 +02:00
av_unused static void DEF(put, pixels16_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
{
MOVQ_BFE(mm6);
__asm__ volatile(
"lea (%3, %3), %%"REG_a" \n\t"
".p2align 3 \n\t"
"1: \n\t"
"movq (%1), %%mm0 \n\t"
"movq 1(%1), %%mm1 \n\t"
"movq (%1, %3), %%mm2 \n\t"
"movq 1(%1, %3), %%mm3 \n\t"
PAVGBP(%%mm0, %%mm1, %%mm4, %%mm2, %%mm3, %%mm5)
"movq %%mm4, (%2) \n\t"
"movq %%mm5, (%2, %3) \n\t"
"movq 8(%1), %%mm0 \n\t"
"movq 9(%1), %%mm1 \n\t"
"movq 8(%1, %3), %%mm2 \n\t"
"movq 9(%1, %3), %%mm3 \n\t"
PAVGBP(%%mm0, %%mm1, %%mm4, %%mm2, %%mm3, %%mm5)
"movq %%mm4, 8(%2) \n\t"
"movq %%mm5, 8(%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"movq (%1), %%mm0 \n\t"
"movq 1(%1), %%mm1 \n\t"
"movq (%1, %3), %%mm2 \n\t"
"movq 1(%1, %3), %%mm3 \n\t"
PAVGBP(%%mm0, %%mm1, %%mm4, %%mm2, %%mm3, %%mm5)
"movq %%mm4, (%2) \n\t"
"movq %%mm5, (%2, %3) \n\t"
"movq 8(%1), %%mm0 \n\t"
"movq 9(%1), %%mm1 \n\t"
"movq 8(%1, %3), %%mm2 \n\t"
"movq 9(%1, %3), %%mm3 \n\t"
PAVGBP(%%mm0, %%mm1, %%mm4, %%mm2, %%mm3, %%mm5)
"movq %%mm4, 8(%2) \n\t"
"movq %%mm5, 8(%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}
static void DEF(put, pixels8_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
MOVQ_BFE(mm6);
__asm__ volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"movq (%1), %%mm0 \n\t"
".p2align 3 \n\t"
"1: \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"),%%mm2 \n\t"
PAVGBP(%%mm1, %%mm0, %%mm4, %%mm2, %%mm1, %%mm5)
"movq %%mm4, (%2) \n\t"
"movq %%mm5, (%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"),%%mm0 \n\t"
PAVGBP(%%mm1, %%mm2, %%mm4, %%mm0, %%mm1, %%mm5)
"movq %%mm4, (%2) \n\t"
"movq %%mm5, (%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}
avcodec/x86/hpeldsp_rnd_template: silence -Wunused-function on --disable-mmx This silences some of the -Wunused-function warnings when compiled with --disable-mmx, e.g http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx. Header guards are too brittle and ugly for this case. Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2015-09-19 17:38:59 +02:00
av_unused static void DEF(avg, pixels16_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
{
MOVQ_BFE(mm6);
__asm__ volatile(
x86/hpeldsp: fix loop in {avg,avg_no_rnd}_pixels16_x2_mmx Handle it inside the __asm__() block. Fixes fate-vc1_ilaced_twomv when using the gcc-usan toolchain. Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2014-10-23 05:43:43 +02:00
".p2align 3 \n\t"
"1: \n\t"
"movq (%1), %%mm0 \n\t"
"movq 1(%1), %%mm1 \n\t"
"movq (%2), %%mm3 \n\t"
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
PAVGB(%%mm0, %%mm1, %%mm2, %%mm6)
x86: rnd_template: Eliminate pointless OP_AVG macro indirection
2013-05-07 00:15:41 +02:00
PAVGB_MMX(%%mm3, %%mm2, %%mm0, %%mm6)
x86/hpeldsp: fix loop in {avg,avg_no_rnd}_pixels16_x2_mmx Handle it inside the __asm__() block. Fixes fate-vc1_ilaced_twomv when using the gcc-usan toolchain. Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2014-10-23 05:43:43 +02:00
"movq %%mm0, (%2) \n\t"
"movq 8(%1), %%mm0 \n\t"
"movq 9(%1), %%mm1 \n\t"
"movq 8(%2), %%mm3 \n\t"
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
PAVGB(%%mm0, %%mm1, %%mm2, %%mm6)
x86: rnd_template: Eliminate pointless OP_AVG macro indirection
2013-05-07 00:15:41 +02:00
PAVGB_MMX(%%mm3, %%mm2, %%mm0, %%mm6)
x86/hpeldsp: fix loop in {avg,avg_no_rnd}_pixels16_x2_mmx Handle it inside the __asm__() block. Fixes fate-vc1_ilaced_twomv when using the gcc-usan toolchain. Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2014-10-23 05:43:43 +02:00
"movq %%mm0, 8(%2) \n\t"
"add %3, %1 \n\t"
"add %3, %2 \n\t"
"subl $1, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((x86_reg)line_size)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
:"memory");
}
static void DEF(avg, pixels8_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
MOVQ_BFE(mm6);
__asm__ volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"movq (%1), %%mm0 \n\t"
".p2align 3 \n\t"
"1: \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm2 \n\t"
PAVGBP(%%mm1, %%mm0, %%mm4, %%mm2, %%mm1, %%mm5)
"movq (%2), %%mm3 \n\t"
x86: rnd_template: Eliminate pointless OP_AVG macro indirection
2013-05-07 00:15:41 +02:00
PAVGB_MMX(%%mm3, %%mm4, %%mm0, %%mm6)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
"movq (%2, %3), %%mm3 \n\t"
x86: rnd_template: Eliminate pointless OP_AVG macro indirection
2013-05-07 00:15:41 +02:00
PAVGB_MMX(%%mm3, %%mm5, %%mm1, %%mm6)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
"movq %%mm0, (%2) \n\t"
"movq %%mm1, (%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm0 \n\t"
PAVGBP(%%mm1, %%mm2, %%mm4, %%mm0, %%mm1, %%mm5)
"movq (%2), %%mm3 \n\t"
x86: rnd_template: Eliminate pointless OP_AVG macro indirection
2013-05-07 00:15:41 +02:00
PAVGB_MMX(%%mm3, %%mm4, %%mm2, %%mm6)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
"movq (%2, %3), %%mm3 \n\t"
x86: rnd_template: Eliminate pointless OP_AVG macro indirection
2013-05-07 00:15:41 +02:00
PAVGB_MMX(%%mm3, %%mm5, %%mm1, %%mm6)
Move x86 half-pel assembly from dsputil to hpeldsp.
2013-03-10 23:37:59 +01:00
"movq %%mm2, (%2) \n\t"
"movq %%mm1, (%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}
Reference in New Issue Copy Permalink