x86: hpeldsp: Move avg_pixels8_x2_mmx() out of hpeldsp_rnd_template.c

The function is only instantiated once, so there is no point
in keeping it in a template file.
This commit is contained in:
Diego Biurrun 2013-04-23 18:36:25 +02:00
parent dc1b328d0d
commit 110796739a
5 changed files with 58 additions and 25 deletions

View File

@ -40,7 +40,8 @@ MMX-OBJS-$(CONFIG_DSPUTIL) += x86/dsputil_mmx.o \
MMX-OBJS-$(CONFIG_ENCODERS) += x86/dsputilenc_mmx.o \
x86/motion_est.o
MMX-OBJS-$(CONFIG_HPELDSP) += x86/fpel_mmx.o
MMX-OBJS-$(CONFIG_HPELDSP) += x86/fpel_mmx.o \
x86/hpeldsp_mmx.o
MMX-OBJS-$(CONFIG_VC1_DECODER) += x86/vc1dsp_mmx.o
YASM-OBJS-$(CONFIG_AAC_DECODER) += x86/sbrdsp.o

View File

@ -173,6 +173,9 @@ void ff_avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels,
void ff_put_pixels16_sse2(uint8_t *block, const uint8_t *pixels,
ptrdiff_t line_size, int h);
void ff_avg_pixels8_x2_mmx(uint8_t *block, const uint8_t *pixels,
ptrdiff_t line_size, int h);
void ff_put_rv40_qpel8_mc33_mmx(uint8_t *block, uint8_t *pixels, ptrdiff_t stride);
void ff_put_rv40_qpel16_mc33_mmx(uint8_t *block, uint8_t *pixels, ptrdiff_t stride);
void ff_avg_rv40_qpel8_mc33_mmx(uint8_t *block, uint8_t *pixels, ptrdiff_t stride);

View File

@ -75,6 +75,7 @@ void ff_avg_pixels8_xy2_3dnow(uint8_t *block, const uint8_t *pixels,
ptrdiff_t line_size, int h);
#define avg_pixels8_mmx ff_avg_pixels8_mmx
#define avg_pixels8_x2_mmx ff_avg_pixels8_x2_mmx
#define avg_pixels16_mmx ff_avg_pixels16_mmx
#define put_pixels8_mmx ff_put_pixels8_mmx
#define put_pixels16_mmx ff_put_pixels16_mmx
@ -86,7 +87,6 @@ void ff_avg_pixels8_xy2_3dnow(uint8_t *block, const uint8_t *pixels,
/***********************************/
/* MMX no rounding */
#define NO_RND 1
#define DEF(x, y) x ## _no_rnd_ ## y ## _mmx
#define SET_RND MOVQ_WONE
#define PAVGBP(a, b, c, d, e, f) PAVGBP_MMX_NO_RND(a, b, c, d, e, f)
@ -99,7 +99,6 @@ void ff_avg_pixels8_xy2_3dnow(uint8_t *block, const uint8_t *pixels,
#undef SET_RND
#undef PAVGBP
#undef PAVGB
#undef NO_RND
/***********************************/
/* MMX rounding */

View File

@ -0,0 +1,52 @@
/*
* MMX-optimized avg/put pixel routines
*
* Copyright (c) 2001 Fabrice Bellard
*
* This file is part of Libav.
*
* Libav 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.
*
* Libav 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 Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include <stddef.h>
#include <stdint.h>
#include "config.h"
#include "dsputil_mmx.h"
#if HAVE_MMX_INLINE
void ff_avg_pixels8_x2_mmx(uint8_t *block, const uint8_t *pixels,
ptrdiff_t line_size, int h)
{
MOVQ_BFE(mm6);
JUMPALIGN();
do {
__asm__ volatile(
"movq %1, %%mm0 \n\t"
"movq 1%1, %%mm1 \n\t"
"movq %0, %%mm3 \n\t"
PAVGB_MMX(%%mm0, %%mm1, %%mm2, %%mm6)
PAVGB_MMX(%%mm3, %%mm2, %%mm0, %%mm6)
"movq %%mm0, %0 \n\t"
:"+m"(*block)
:"m"(*pixels)
:"memory");
pixels += line_size;
block += line_size;
} while (--h);
}
#endif /* HAVE_MMX_INLINE */

View File

@ -134,28 +134,6 @@ static void DEF(put, pixels8_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_
:REG_a, "memory");
}
#ifndef NO_RND
static void DEF(avg, pixels8_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
MOVQ_BFE(mm6);
JUMPALIGN();
do {
__asm__ volatile(
"movq %1, %%mm0 \n\t"
"movq 1%1, %%mm1 \n\t"
"movq %0, %%mm3 \n\t"
PAVGB(%%mm0, %%mm1, %%mm2, %%mm6)
OP_AVG(%%mm3, %%mm2, %%mm0, %%mm6)
"movq %%mm0, %0 \n\t"
:"+m"(*block)
:"m"(*pixels)
:"memory");
pixels += line_size;
block += line_size;
} while (--h);
}
#endif // NO_RND
static void DEF(avg, pixels16_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
MOVQ_BFE(mm6);