RepoMirrors
/
ffmpeg
mirror of https://git.ffmpeg.org/ffmpeg.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
ffmpeg/libavcodec/x86/hpeldsp.asm

462 lines
11 KiB
NASM
Raw Normal View History

dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
;******************************************************************************
x86/hpeldsp: Fix author attribution This also fixes the project name Original authors fabrice and nick go back to the initial ffmpeg commit Others for example contributed in: (for a complete list please use git blame / show / log) commit e9c0a38ff0a9d4754220ae3432b2cdebe5a1c781 Author: Zdenek Kabelac <kabi@informatics.muni.cz> Date: Tue May 28 16:35:58 2002 +0000 * optimized avg_* functions (except xy2) * minor speedup for put_pixels_x2 & cleanup Originally committed as revision 619 to svn://svn.ffmpeg.org/ffmpeg/trunk commit 607dce96c0225e30ae2e7f3b8de2d00b4f064805 Author: Michael Niedermayer <michaelni@gmx.at> Date: Fri May 17 01:04:14 2002 +0000 hopefully faster mmx2&3dnow MC Originally committed as revision 506 to svn://svn.ffmpeg.org/ffmpeg/trunk Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-27 13:45:19 +00:00
;*
;* Copyright (c) 2000-2001 Fabrice Bellard <fabrice@bellard.org>
;* Copyright (c) Nick Kurshev <nickols_k@mail.ru>
;* Copyright (c) 2002 Michael Niedermayer <michaelni@gmx.at>
;* Copyright (c) 2002 Zdenek Kabelac <kabi@informatics.muni.cz>
avcodec/x86: Add daniels copyright to the recent gcc->yasm convertions he did. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-03 12:50:44 +00:00
;* Copyright (c) 2013 Daniel Kang
x86/hpeldsp: Fix author attribution This also fixes the project name Original authors fabrice and nick go back to the initial ffmpeg commit Others for example contributed in: (for a complete list please use git blame / show / log) commit e9c0a38ff0a9d4754220ae3432b2cdebe5a1c781 Author: Zdenek Kabelac <kabi@informatics.muni.cz> Date: Tue May 28 16:35:58 2002 +0000 * optimized avg_* functions (except xy2) * minor speedup for put_pixels_x2 & cleanup Originally committed as revision 619 to svn://svn.ffmpeg.org/ffmpeg/trunk commit 607dce96c0225e30ae2e7f3b8de2d00b4f064805 Author: Michael Niedermayer <michaelni@gmx.at> Date: Fri May 17 01:04:14 2002 +0000 hopefully faster mmx2&3dnow MC Originally committed as revision 506 to svn://svn.ffmpeg.org/ffmpeg/trunk Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-27 13:45:19 +00:00
;*
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
;* MMX optimized hpel functions
;*
x86/hpeldsp: Fix author attribution This also fixes the project name Original authors fabrice and nick go back to the initial ffmpeg commit Others for example contributed in: (for a complete list please use git blame / show / log) commit e9c0a38ff0a9d4754220ae3432b2cdebe5a1c781 Author: Zdenek Kabelac <kabi@informatics.muni.cz> Date: Tue May 28 16:35:58 2002 +0000 * optimized avg_* functions (except xy2) * minor speedup for put_pixels_x2 & cleanup Originally committed as revision 619 to svn://svn.ffmpeg.org/ffmpeg/trunk commit 607dce96c0225e30ae2e7f3b8de2d00b4f064805 Author: Michael Niedermayer <michaelni@gmx.at> Date: Fri May 17 01:04:14 2002 +0000 hopefully faster mmx2&3dnow MC Originally committed as revision 506 to svn://svn.ffmpeg.org/ffmpeg/trunk Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-27 13:45:19 +00:00
;* This file is part of FFmpeg.
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
;*
x86/hpeldsp: Fix author attribution This also fixes the project name Original authors fabrice and nick go back to the initial ffmpeg commit Others for example contributed in: (for a complete list please use git blame / show / log) commit e9c0a38ff0a9d4754220ae3432b2cdebe5a1c781 Author: Zdenek Kabelac <kabi@informatics.muni.cz> Date: Tue May 28 16:35:58 2002 +0000 * optimized avg_* functions (except xy2) * minor speedup for put_pixels_x2 & cleanup Originally committed as revision 619 to svn://svn.ffmpeg.org/ffmpeg/trunk commit 607dce96c0225e30ae2e7f3b8de2d00b4f064805 Author: Michael Niedermayer <michaelni@gmx.at> Date: Fri May 17 01:04:14 2002 +0000 hopefully faster mmx2&3dnow MC Originally committed as revision 506 to svn://svn.ffmpeg.org/ffmpeg/trunk Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-27 13:45:19 +00:00
;* FFmpeg is free software; you can redistribute it and/or
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
;* 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.
;*
x86/hpeldsp: Fix author attribution This also fixes the project name Original authors fabrice and nick go back to the initial ffmpeg commit Others for example contributed in: (for a complete list please use git blame / show / log) commit e9c0a38ff0a9d4754220ae3432b2cdebe5a1c781 Author: Zdenek Kabelac <kabi@informatics.muni.cz> Date: Tue May 28 16:35:58 2002 +0000 * optimized avg_* functions (except xy2) * minor speedup for put_pixels_x2 & cleanup Originally committed as revision 619 to svn://svn.ffmpeg.org/ffmpeg/trunk commit 607dce96c0225e30ae2e7f3b8de2d00b4f064805 Author: Michael Niedermayer <michaelni@gmx.at> Date: Fri May 17 01:04:14 2002 +0000 hopefully faster mmx2&3dnow MC Originally committed as revision 506 to svn://svn.ffmpeg.org/ffmpeg/trunk Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-27 13:45:19 +00:00
;* FFmpeg is distributed in the hope that it will be useful,
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
;* 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
x86/hpeldsp: Fix author attribution This also fixes the project name Original authors fabrice and nick go back to the initial ffmpeg commit Others for example contributed in: (for a complete list please use git blame / show / log) commit e9c0a38ff0a9d4754220ae3432b2cdebe5a1c781 Author: Zdenek Kabelac <kabi@informatics.muni.cz> Date: Tue May 28 16:35:58 2002 +0000 * optimized avg_* functions (except xy2) * minor speedup for put_pixels_x2 & cleanup Originally committed as revision 619 to svn://svn.ffmpeg.org/ffmpeg/trunk commit 607dce96c0225e30ae2e7f3b8de2d00b4f064805 Author: Michael Niedermayer <michaelni@gmx.at> Date: Fri May 17 01:04:14 2002 +0000 hopefully faster mmx2&3dnow MC Originally committed as revision 506 to svn://svn.ffmpeg.org/ffmpeg/trunk Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-27 13:45:19 +00:00
;* License along with FFmpeg; if not, write to the Free Software
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
;******************************************************************************
%include "libavutil/x86/x86util.asm"
SECTION_RODATA
cextern pb_1
SECTION_TEXT
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; put_pixels8_x2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro PUT_PIXELS8_X2 0
cglobal put_pixels8_x2, 4,5
lea r4, [r2*2]
.loop:
mova m0, [r1]
mova m1, [r1+r2]
PAVGB m0, [r1+1]
PAVGB m1, [r1+r2+1]
mova [r0], m0
mova [r0+r2], m1
add r1, r4
add r0, r4
mova m0, [r1]
mova m1, [r1+r2]
PAVGB m0, [r1+1]
PAVGB m1, [r1+r2+1]
add r1, r4
mova [r0], m0
mova [r0+r2], m1
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
PUT_PIXELS8_X2
INIT_MMX 3dnow
PUT_PIXELS8_X2
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; put_pixels16_x2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro PUT_PIXELS_16 0
cglobal put_pixels16_x2, 4,5
lea r4, [r2*2]
.loop:
mova m0, [r1]
mova m1, [r1+r2]
mova m2, [r1+8]
mova m3, [r1+r2+8]
PAVGB m0, [r1+1]
PAVGB m1, [r1+r2+1]
PAVGB m2, [r1+9]
PAVGB m3, [r1+r2+9]
mova [r0], m0
mova [r0+r2], m1
mova [r0+8], m2
mova [r0+r2+8], m3
add r1, r4
add r0, r4
mova m0, [r1]
mova m1, [r1+r2]
mova m2, [r1+8]
mova m3, [r1+r2+8]
PAVGB m0, [r1+1]
PAVGB m1, [r1+r2+1]
PAVGB m2, [r1+9]
PAVGB m3, [r1+r2+9]
add r1, r4
mova [r0], m0
mova [r0+r2], m1
mova [r0+8], m2
mova [r0+r2+8], m3
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
PUT_PIXELS_16
INIT_MMX 3dnow
PUT_PIXELS_16
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; put_no_rnd_pixels8_x2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro PUT_NO_RND_PIXELS8_X2 0
cglobal put_no_rnd_pixels8_x2, 4,5
mova m6, [pb_1]
lea r4, [r2*2]
.loop:
mova m0, [r1]
mova m2, [r1+r2]
mova m1, [r1+1]
mova m3, [r1+r2+1]
add r1, r4
psubusb m0, m6
psubusb m2, m6
PAVGB m0, m1
PAVGB m2, m3
mova [r0], m0
mova [r0+r2], m2
mova m0, [r1]
mova m1, [r1+1]
mova m2, [r1+r2]
mova m3, [r1+r2+1]
add r0, r4
add r1, r4
psubusb m0, m6
psubusb m2, m6
PAVGB m0, m1
PAVGB m2, m3
mova [r0], m0
mova [r0+r2], m2
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
PUT_NO_RND_PIXELS8_X2
INIT_MMX 3dnow
PUT_NO_RND_PIXELS8_X2
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; put_no_rnd_pixels8_x2_exact(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro PUT_NO_RND_PIXELS8_X2_EXACT 0
cglobal put_no_rnd_pixels8_x2_exact, 4,5
lea r4, [r2*3]
pcmpeqb m6, m6
.loop:
mova m0, [r1]
mova m2, [r1+r2]
mova m1, [r1+1]
mova m3, [r1+r2+1]
pxor m0, m6
pxor m2, m6
pxor m1, m6
pxor m3, m6
PAVGB m0, m1
PAVGB m2, m3
pxor m0, m6
pxor m2, m6
mova [r0], m0
mova [r0+r2], m2
mova m0, [r1+r2*2]
mova m1, [r1+r2*2+1]
mova m2, [r1+r4]
mova m3, [r1+r4+1]
pxor m0, m6
pxor m1, m6
pxor m2, m6
pxor m3, m6
PAVGB m0, m1
PAVGB m2, m3
pxor m0, m6
pxor m2, m6
mova [r0+r2*2], m0
mova [r0+r4], m2
lea r1, [r1+r2*4]
lea r0, [r0+r2*4]
sub r3d, 4
jg .loop
REP_RET
%endmacro
INIT_MMX mmxext
PUT_NO_RND_PIXELS8_X2_EXACT
INIT_MMX 3dnow
PUT_NO_RND_PIXELS8_X2_EXACT
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; put_pixels8_y2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro PUT_PIXELS8_Y2 0
cglobal put_pixels8_y2, 4,5
lea r4, [r2*2]
mova m0, [r1]
sub r0, r2
.loop:
mova m1, [r1+r2]
mova m2, [r1+r4]
add r1, r4
PAVGB m0, m1
PAVGB m1, m2
mova [r0+r2], m0
mova [r0+r4], m1
mova m1, [r1+r2]
mova m0, [r1+r4]
add r0, r4
add r1, r4
PAVGB m2, m1
PAVGB m1, m0
mova [r0+r2], m2
mova [r0+r4], m1
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
PUT_PIXELS8_Y2
INIT_MMX 3dnow
PUT_PIXELS8_Y2
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; put_no_rnd_pixels8_y2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro PUT_NO_RND_PIXELS8_Y2 0
cglobal put_no_rnd_pixels8_y2, 4,5
mova m6, [pb_1]
lea r4, [r2+r2]
mova m0, [r1]
sub r0, r2
.loop:
mova m1, [r1+r2]
mova m2, [r1+r4]
add r1, r4
psubusb m1, m6
PAVGB m0, m1
PAVGB m1, m2
mova [r0+r2], m0
mova [r0+r4], m1
mova m1, [r1+r2]
mova m0, [r1+r4]
add r0, r4
add r1, r4
psubusb m1, m6
PAVGB m2, m1
PAVGB m1, m0
mova [r0+r2], m2
mova [r0+r4], m1
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
PUT_NO_RND_PIXELS8_Y2
INIT_MMX 3dnow
PUT_NO_RND_PIXELS8_Y2
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; put_no_rnd_pixels8_y2_exact(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro PUT_NO_RND_PIXELS8_Y2_EXACT 0
cglobal put_no_rnd_pixels8_y2_exact, 4,5
lea r4, [r2*3]
mova m0, [r1]
pcmpeqb m6, m6
add r1, r2
pxor m0, m6
.loop:
mova m1, [r1]
mova m2, [r1+r2]
pxor m1, m6
pxor m2, m6
PAVGB m0, m1
PAVGB m1, m2
pxor m0, m6
pxor m1, m6
mova [r0], m0
mova [r0+r2], m1
mova m1, [r1+r2*2]
mova m0, [r1+r4]
pxor m1, m6
pxor m0, m6
PAVGB m2, m1
PAVGB m1, m0
pxor m2, m6
pxor m1, m6
mova [r0+r2*2], m2
mova [r0+r4], m1
lea r1, [r1+r2*4]
lea r0, [r0+r2*4]
sub r3d, 4
jg .loop
REP_RET
%endmacro
INIT_MMX mmxext
PUT_NO_RND_PIXELS8_Y2_EXACT
INIT_MMX 3dnow
PUT_NO_RND_PIXELS8_Y2_EXACT
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; avg_pixels8(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro AVG_PIXELS8 0
cglobal avg_pixels8, 4,5
dsputil: Fix error by not using redzone and register name Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com> Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 20:28:58 +00:00
lea r4, [r2*2]
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
.loop:
mova m0, [r0]
mova m1, [r0+r2]
PAVGB m0, [r1]
PAVGB m1, [r1+r2]
mova [r0], m0
mova [r0+r2], m1
add r1, r4
add r0, r4
mova m0, [r0]
mova m1, [r0+r2]
PAVGB m0, [r1]
PAVGB m1, [r1+r2]
add r1, r4
mova [r0], m0
mova [r0+r2], m1
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX 3dnow
AVG_PIXELS8
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; avg_pixels8_x2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro AVG_PIXELS8_X2 0
cglobal avg_pixels8_x2, 4,5
lea r4, [r2*2]
.loop:
mova m0, [r1]
mova m2, [r1+r2]
PAVGB m0, [r1+1]
PAVGB m2, [r1+r2+1]
PAVGB m0, [r0]
PAVGB m2, [r0+r2]
add r1, r4
mova [r0], m0
mova [r0+r2], m2
mova m0, [r1]
mova m2, [r1+r2]
PAVGB m0, [r1+1]
PAVGB m2, [r1+r2+1]
add r0, r4
add r1, r4
PAVGB m0, [r0]
PAVGB m2, [r0+r2]
mova [r0], m0
mova [r0+r2], m2
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
AVG_PIXELS8_X2
INIT_MMX 3dnow
AVG_PIXELS8_X2
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; avg_pixels8_y2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro AVG_PIXELS8_Y2 0
cglobal avg_pixels8_y2, 4,5
lea r4, [r2*2]
mova m0, [r1]
sub r0, r2
.loop:
mova m1, [r1+r2]
mova m2, [r1+r4]
add r1, r4
PAVGB m0, m1
PAVGB m1, m2
mova m3, [r0+r2]
mova m4, [r0+r4]
PAVGB m0, m3
PAVGB m1, m4
mova [r0+r2], m0
mova [r0+r4], m1
mova m1, [r1+r2]
mova m0, [r1+r4]
PAVGB m2, m1
PAVGB m1, m0
add r0, r4
add r1, r4
mova m3, [r0+r2]
mova m4, [r0+r4]
PAVGB m2, m3
PAVGB m1, m4
mova [r0+r2], m2
mova [r0+r4], m1
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
AVG_PIXELS8_Y2
INIT_MMX 3dnow
AVG_PIXELS8_Y2
Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter. This avoids SIMD-optimized functions having to sign-extend their line size argument manually to be able to do pointer arithmetic.
2013-01-29 21:13:55 +00:00
; avg_pixels8_xy2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
%macro AVG_PIXELS8_XY2 0
cglobal avg_pixels8_xy2, 4,5
mova m6, [pb_1]
lea r4, [r2*2]
mova m0, [r1]
pavgb m0, [r1+1]
.loop:
mova m2, [r1+r4]
mova m1, [r1+r2]
psubusb m2, m6
pavgb m1, [r1+r2+1]
pavgb m2, [r1+r4+1]
add r1, r4
pavgb m0, m1
pavgb m1, m2
pavgb m0, [r0]
pavgb m1, [r0+r2]
mova [r0], m0
mova [r0+r2], m1
mova m1, [r1+r2]
mova m0, [r1+r4]
pavgb m1, [r1+r2+1]
pavgb m0, [r1+r4+1]
add r0, r4
add r1, r4
pavgb m2, m1
pavgb m1, m0
pavgb m2, [r0]
pavgb m1, [r0+r2]
mova [r0], m2
AVG_PIXELS8_XY2: fix typo, make code actually work Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-27 14:50:26 +00:00
mova [r0+r2], m1
dsputil: x86: Convert mpeg4 qpel and dsputil avg to yasm Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2013-01-27 03:45:43 +00:00
add r0, r4
sub r3d, 4
jne .loop
REP_RET
%endmacro
INIT_MMX mmxext
AVG_PIXELS8_XY2
INIT_MMX 3dnow
AVG_PIXELS8_XY2