virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-21 10:55:51 +02:00
Code Issues Releases Activity
6f2309ed2e
FFmpeg/libavcodec/x86/simple_idct_mmx.c

1297 lines
71 KiB
C
Raw Normal View History

new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
/*
license/copyright change Originally committed as revision 599 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-26 01:45:33 +03:00
* Simple IDCT MMX
*
* Copyright (c) 2001, 2002 Michael Niedermayer <michaelni@gmx.at>
*
Replace FFmpeg with Libav in licence headers Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-03-18 19:35:10 +02:00
* This file is part of Libav.
Change license headers to say 'FFmpeg' instead of 'this program/this library' and fix GPL/LGPL version mismatches. Originally committed as revision 6577 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-10-07 18:30:46 +03:00
*
Replace FFmpeg with Libav in licence headers Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-03-18 19:35:10 +02:00
* Libav is free software; you can redistribute it and/or
license/copyright change Originally committed as revision 599 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-26 01:45:33 +03:00
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
Change license headers to say 'FFmpeg' instead of 'this program/this library' and fix GPL/LGPL version mismatches. Originally committed as revision 6577 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-10-07 18:30:46 +03:00
* version 2.1 of the License, or (at your option) any later version.
license/copyright change Originally committed as revision 599 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-26 01:45:33 +03:00
*
Replace FFmpeg with Libav in licence headers Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-03-18 19:35:10 +02:00
* Libav is distributed in the hope that it will be useful,
license/copyright change Originally committed as revision 599 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-26 01:45:33 +03:00
* 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
Replace FFmpeg with Libav in licence headers Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-03-18 19:35:10 +02:00
* License along with Libav; if not, write to the Free Software
Update licensing information: The FSF changed postal address. Originally committed as revision 4842 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-01-13 00:43:26 +02:00
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
license/copyright change Originally committed as revision 599 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-26 01:45:33 +03:00
*/
Use full path for #includes from another directory. Originally committed as revision 13098 to svn://svn.ffmpeg.org/ffmpeg/trunk
2008-05-09 14:56:36 +03:00
#include "libavcodec/dsputil.h"
#include "libavcodec/simple_idct.h"
Move declarations of some mmx functions to dsputil_mmx.h Originally committed as revision 19739 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-08-29 19:55:50 +03:00
#include "dsputil_mmx.h"
license/copyright change Originally committed as revision 599 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-26 01:45:33 +03:00
set C4 to 16383 Originally committed as revision 432 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-02 15:51:10 +03:00
/*
23170.475006
22725.260826
21406.727617
19265.545870
16384.000000
12872.826198
8866.956905
4520.335430
*/
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
#define C0 23170 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
#define C1 22725 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
#define C2 21407 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
#define C3 19266 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
set C4 to 16383 Originally committed as revision 432 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-02 15:51:10 +03:00
#if 0
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
#define C4 16384 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
set C4 to 16383 Originally committed as revision 432 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-02 15:51:10 +03:00
#else
#define C4 16383 //cos(i*M_PI/16)*sqrt(2)*(1<<14) - 0.5
#endif
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
#define C5 12873 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
#define C6 8867 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
#define C7 4520 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
#define ROW_SHIFT 11
#define COL_SHIFT 20 // 6
Add and use DECLARE_ASM_CONST for constants used in assembler code. Should make it easier to work around compilation problems with e.g. ICC. Originally committed as revision 11641 to svn://svn.ffmpeg.org/ffmpeg/trunk
2008-01-27 22:50:16 +02:00
DECLARE_ASM_CONST(8, uint64_t, wm1010)= 0xFFFF0000FFFF0000ULL;
DECLARE_ASM_CONST(8, uint64_t, d40000)= 0x0000000000040000ULL;
another non const static, maybe its thread save now Originally committed as revision 2276 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-09-15 22:18:28 +03:00
Move array specifiers outside DECLARE_ALIGNED() invocations Originally committed as revision 21377 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-22 05:25:11 +02:00
DECLARE_ALIGNED(8, static const int16_t, coeffs)[]= {
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
1<<(ROW_SHIFT-1), 0, 1<<(ROW_SHIFT-1), 0,
// 1<<(COL_SHIFT-1), 0, 1<<(COL_SHIFT-1), 0,
// 0, 1<<(COL_SHIFT-1-16), 0, 1<<(COL_SHIFT-1-16),
1<<(ROW_SHIFT-1), 1, 1<<(ROW_SHIFT-1), 0,
// the 1 = ((1<<(COL_SHIFT-1))/C4)<<ROW_SHIFT :)
// 0, 0, 0, 0,
// 0, 0, 0, 0,
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
C4, C4, C4, C4,
C4, -C4, C4, -C4,
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
C2, C6, C2, C6,
C6, -C2, C6, -C2,
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
C1, C3, C1, C3,
C5, C7, C5, C7,
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
C3, -C7, C3, -C7,
-C1, -C5, -C1, -C5,
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
C5, -C1, C5, -C1,
C7, C3, C7, C3,
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
C7, -C5, C7, -C5,
C3, -C1, C3, -C1
};
- Advanced Intra Coding (AIC) support for H.263+ encoder, just DC by now. - Bug fix H.263+ AIC tables. - Warning fixes. Originally committed as revision 431 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-02 07:39:45 +03:00
#if 0
Add missing 'void' keyword to parameterless function declarations. Originally committed as revision 16436 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-01-05 15:57:43 +02:00
static void unused_var_killer(void)
{
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
int a= wm1010 + d40000;
temp[0]=a;
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
}
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
static void inline idctCol (int16_t * col, int16_t *input)
{
#undef C0
#undef C1
#undef C2
#undef C3
#undef C4
#undef C5
#undef C6
#undef C7
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
int a0, a1, a2, a3, b0, b1, b2, b3;
const int C0 = 23170; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C1 = 22725; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C2 = 21407; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C3 = 19266; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C4 = 16383; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C5 = 12873; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C6 = 8867; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C7 = 4520; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
/*
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
if( !(col[8*1] | col[8*2] |col[8*3] |col[8*4] |col[8*5] |col[8*6] | col[8*7])) {
col[8*0] = col[8*1] = col[8*2] = col[8*3] = col[8*4] =
col[8*5] = col[8*6] = col[8*7] = col[8*0]<<3;
return;
}*/
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
col[8*0] = input[8*0 + 0];
col[8*1] = input[8*2 + 0];
col[8*2] = input[8*0 + 1];
col[8*3] = input[8*2 + 1];
col[8*4] = input[8*4 + 0];
col[8*5] = input[8*6 + 0];
col[8*6] = input[8*4 + 1];
col[8*7] = input[8*6 + 1];
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
a0 = C4*col[8*0] + C2*col[8*2] + C4*col[8*4] + C6*col[8*6] + (1<<(COL_SHIFT-1));
a1 = C4*col[8*0] + C6*col[8*2] - C4*col[8*4] - C2*col[8*6] + (1<<(COL_SHIFT-1));
a2 = C4*col[8*0] - C6*col[8*2] - C4*col[8*4] + C2*col[8*6] + (1<<(COL_SHIFT-1));
a3 = C4*col[8*0] - C2*col[8*2] + C4*col[8*4] - C6*col[8*6] + (1<<(COL_SHIFT-1));
b0 = C1*col[8*1] + C3*col[8*3] + C5*col[8*5] + C7*col[8*7];
b1 = C3*col[8*1] - C7*col[8*3] - C1*col[8*5] - C5*col[8*7];
b2 = C5*col[8*1] - C1*col[8*3] + C7*col[8*5] + C3*col[8*7];
b3 = C7*col[8*1] - C5*col[8*3] + C3*col[8*5] - C1*col[8*7];
col[8*0] = (a0 + b0) >> COL_SHIFT;
col[8*1] = (a1 + b1) >> COL_SHIFT;
col[8*2] = (a2 + b2) >> COL_SHIFT;
col[8*3] = (a3 + b3) >> COL_SHIFT;
col[8*4] = (a3 - b3) >> COL_SHIFT;
col[8*5] = (a2 - b2) >> COL_SHIFT;
col[8*6] = (a1 - b1) >> COL_SHIFT;
col[8*7] = (a0 - b0) >> COL_SHIFT;
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
}
static void inline idctRow (int16_t * output, int16_t * input)
{
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
int16_t row[8];
int a0, a1, a2, a3, b0, b1, b2, b3;
const int C0 = 23170; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C1 = 22725; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C2 = 21407; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C3 = 19266; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C4 = 16383; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C5 = 12873; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C6 = 8867; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
const int C7 = 4520; //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
row[0] = input[0];
row[2] = input[1];
row[4] = input[4];
row[6] = input[5];
row[1] = input[8];
row[3] = input[9];
row[5] = input[12];
row[7] = input[13];
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
if( !(row[1] | row[2] |row[3] |row[4] |row[5] |row[6] | row[7]) ) {
row[0] = row[1] = row[2] = row[3] = row[4] =
row[5] = row[6] = row[7] = row[0]<<3;
output[0] = row[0];
output[2] = row[1];
output[4] = row[2];
output[6] = row[3];
output[8] = row[4];
output[10] = row[5];
output[12] = row[6];
output[14] = row[7];
return;
}
a0 = C4*row[0] + C2*row[2] + C4*row[4] + C6*row[6] + (1<<(ROW_SHIFT-1));
a1 = C4*row[0] + C6*row[2] - C4*row[4] - C2*row[6] + (1<<(ROW_SHIFT-1));
a2 = C4*row[0] - C6*row[2] - C4*row[4] + C2*row[6] + (1<<(ROW_SHIFT-1));
a3 = C4*row[0] - C2*row[2] + C4*row[4] - C6*row[6] + (1<<(ROW_SHIFT-1));
b0 = C1*row[1] + C3*row[3] + C5*row[5] + C7*row[7];
b1 = C3*row[1] - C7*row[3] - C1*row[5] - C5*row[7];
b2 = C5*row[1] - C1*row[3] + C7*row[5] + C3*row[7];
b3 = C7*row[1] - C5*row[3] + C3*row[5] - C1*row[7];
row[0] = (a0 + b0) >> ROW_SHIFT;
row[1] = (a1 + b1) >> ROW_SHIFT;
row[2] = (a2 + b2) >> ROW_SHIFT;
row[3] = (a3 + b3) >> ROW_SHIFT;
row[4] = (a3 - b3) >> ROW_SHIFT;
row[5] = (a2 - b2) >> ROW_SHIFT;
row[6] = (a1 - b1) >> ROW_SHIFT;
row[7] = (a0 - b0) >> ROW_SHIFT;
output[0] = row[0];
output[2] = row[1];
output[4] = row[2];
output[6] = row[3];
output[8] = row[4];
output[10] = row[5];
output[12] = row[6];
output[14] = row[7];
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
}
#endif
static inline void idct(int16_t *block)
{
Move array specifiers outside DECLARE_ALIGNED() invocations Originally committed as revision 21377 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-22 05:25:11 +02:00
DECLARE_ALIGNED(8, int64_t, align_tmp)[16];
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
int16_t * const temp= (int16_t*)align_tmp;
another non const static, maybe its thread save now Originally committed as revision 2276 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-09-15 22:18:28 +03:00
Convert asm keyword into __asm__. Neither the asm() nor the __asm__() keyword is part of the C99 standard, but while GCC accepts the former in C89 syntax, it is not accepted in C99 unless GNU extensions are turned on (with -fasm). The latter form is accepted in any syntax as an extension (without requiring further command-line options). Sun Studio C99 compiler also does not accept asm() while accepting __asm__(), albeit reporting warnings that it's not valid C99 syntax. Originally committed as revision 15627 to svn://svn.ffmpeg.org/ffmpeg/trunk
2008-10-16 16:34:09 +03:00
__asm__ volatile(
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
#if 0 //Alternative, simpler variant
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#define ROW_IDCT(src0, src4, src1, src5, dst, rounder, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
#rounder ", %%mm4 \n\t"\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq 56(%2), %%mm5 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
#rounder ", %%mm0 \n\t"\
"paddd %%mm0, %%mm1 \n\t" /* A1 a1 */\
"paddd %%mm0, %%mm0 \n\t" \
"psubd %%mm1, %%mm0 \n\t" /* A2 a2 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm5, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm5 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm5 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm1, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm5, %%mm1 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm5, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm1, %%mm7 \n\t" /* A1+B1 a1+b1 A0+B0 a0+b0 */\
"packssdw %%mm4, %%mm2 \n\t" /* A0-B0 a0-b0 A1-B1 a1-b1 */\
"movq %%mm7, " #dst " \n\t"\
"movq " #src1 ", %%mm1 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"movq %%mm2, 24+" #dst " \n\t"\
"pmaddwd %%mm1, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm1 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm0, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm0 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm0 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm1, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"packssdw %%mm6, %%mm2 \n\t" /* A3+B3 a3+b3 A2+B2 a2+b2 */\
"movq %%mm2, 8+" #dst " \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm0, %%mm4 \n\t" /* A2-B2 a2-b2 A3-B3 a3-b3 */\
"movq %%mm4, 16+" #dst " \n\t"\
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define COL_IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"paddd %%mm1, %%mm0 \n\t" /* A1 a1 */\
"psubd %%mm1, %%mm5 \n\t" /* A2 a2 */\
"movq 56(%2), %%mm1 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm1, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm1 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm1 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm0, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm1, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm1, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm0 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm7, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"movd %%mm7, " #dst " \n\t"\
"packssdw %%mm0, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"movd %%mm0, 16+" #dst " \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"movd %%mm2, 96+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"movd %%mm4, 112+" #dst " \n\t"\
"movq " #src1 ", %%mm0 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm0 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm5, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm5 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm0, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"packssdw %%mm6, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movd %%mm2, 32+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A3-B3 a3-b3 */\
"packssdw %%mm5, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movd %%mm6, 48+" #dst " \n\t"\
"movd %%mm4, 64+" #dst " \n\t"\
"movd %%mm5, 80+" #dst " \n\t"\
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#define DC_COND_ROW_IDCT(src0, src4, src1, src5, dst, rounder, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq "MANGLE(wm1010)", %%mm4 \n\t"\
"pand %%mm0, %%mm4 \n\t"\
"por %%mm1, %%mm4 \n\t"\
"por %%mm2, %%mm4 \n\t"\
"por %%mm3, %%mm4 \n\t"\
"packssdw %%mm4,%%mm4 \n\t"\
"movd %%mm4, %%eax \n\t"\
"orl %%eax, %%eax \n\t"\
"jz 1f \n\t"\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
#rounder ", %%mm4 \n\t"\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq 56(%2), %%mm5 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
#rounder ", %%mm0 \n\t"\
"paddd %%mm0, %%mm1 \n\t" /* A1 a1 */\
"paddd %%mm0, %%mm0 \n\t" \
"psubd %%mm1, %%mm0 \n\t" /* A2 a2 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm5, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm5 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm5 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm1, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm5, %%mm1 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm5, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm1, %%mm7 \n\t" /* A1+B1 a1+b1 A0+B0 a0+b0 */\
"packssdw %%mm4, %%mm2 \n\t" /* A0-B0 a0-b0 A1-B1 a1-b1 */\
"movq %%mm7, " #dst " \n\t"\
"movq " #src1 ", %%mm1 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"movq %%mm2, 24+" #dst " \n\t"\
"pmaddwd %%mm1, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm1 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm0, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm0 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm0 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm1, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"packssdw %%mm6, %%mm2 \n\t" /* A3+B3 a3+b3 A2+B2 a2+b2 */\
"movq %%mm2, 8+" #dst " \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm0, %%mm4 \n\t" /* A2-B2 a2-b2 A3-B3 a3-b3 */\
"movq %%mm4, 16+" #dst " \n\t"\
"jmp 2f \n\t"\
"1: \n\t"\
"pslld $16, %%mm0 \n\t"\
"#paddd "MANGLE(d40000)", %%mm0 \n\t"\
"psrad $13, %%mm0 \n\t"\
"packssdw %%mm0, %%mm0 \n\t"\
"movq %%mm0, " #dst " \n\t"\
"movq %%mm0, 8+" #dst " \n\t"\
"movq %%mm0, 16+" #dst " \n\t"\
"movq %%mm0, 24+" #dst " \n\t"\
"2: \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
//IDCT( src0, src4, src1, src5, dst, rounder, shift)
ROW_IDCT( (%0), 8(%0), 16(%0), 24(%0), 0(%1),paddd 8(%2), 11)
/*ROW_IDCT( 32(%0), 40(%0), 48(%0), 56(%0), 32(%1), paddd (%2), 11)
ROW_IDCT( 64(%0), 72(%0), 80(%0), 88(%0), 64(%1), paddd (%2), 11)
ROW_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1), paddd (%2), 11)*/
DC_COND_ROW_IDCT( 32(%0), 40(%0), 48(%0), 56(%0), 32(%1),paddd (%2), 11)
DC_COND_ROW_IDCT( 64(%0), 72(%0), 80(%0), 88(%0), 64(%1),paddd (%2), 11)
DC_COND_ROW_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11)
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
COL_IDCT( (%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
COL_IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
COL_IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
COL_IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#else
#define DC_COND_IDCT(src0, src4, src1, src5, dst, rounder, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq "MANGLE(wm1010)", %%mm4 \n\t"\
"pand %%mm0, %%mm4 \n\t"\
"por %%mm1, %%mm4 \n\t"\
"por %%mm2, %%mm4 \n\t"\
"por %%mm3, %%mm4 \n\t"\
"packssdw %%mm4,%%mm4 \n\t"\
"movd %%mm4, %%eax \n\t"\
"orl %%eax, %%eax \n\t"\
"jz 1f \n\t"\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
#rounder ", %%mm4 \n\t"\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq 56(%2), %%mm5 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
#rounder ", %%mm0 \n\t"\
"paddd %%mm0, %%mm1 \n\t" /* A1 a1 */\
"paddd %%mm0, %%mm0 \n\t" \
"psubd %%mm1, %%mm0 \n\t" /* A2 a2 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm5, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm5 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm5 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm1, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm5, %%mm1 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm5, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm1, %%mm7 \n\t" /* A1+B1 a1+b1 A0+B0 a0+b0 */\
"packssdw %%mm4, %%mm2 \n\t" /* A0-B0 a0-b0 A1-B1 a1-b1 */\
"movq %%mm7, " #dst " \n\t"\
"movq " #src1 ", %%mm1 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"movq %%mm2, 24+" #dst " \n\t"\
"pmaddwd %%mm1, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm1 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm0, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm0 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm0 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm1, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"packssdw %%mm6, %%mm2 \n\t" /* A3+B3 a3+b3 A2+B2 a2+b2 */\
"movq %%mm2, 8+" #dst " \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm0, %%mm4 \n\t" /* A2-B2 a2-b2 A3-B3 a3-b3 */\
"movq %%mm4, 16+" #dst " \n\t"\
"jmp 2f \n\t"\
"1: \n\t"\
"pslld $16, %%mm0 \n\t"\
"paddd "MANGLE(d40000)", %%mm0 \n\t"\
"psrad $13, %%mm0 \n\t"\
"packssdw %%mm0, %%mm0 \n\t"\
"movq %%mm0, " #dst " \n\t"\
"movq %%mm0, 8+" #dst " \n\t"\
"movq %%mm0, 16+" #dst " \n\t"\
"movq %%mm0, 24+" #dst " \n\t"\
"2: \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#define Z_COND_IDCT(src0, src4, src1, src5, dst, rounder, shift, bt) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq %%mm0, %%mm4 \n\t"\
"por %%mm1, %%mm4 \n\t"\
"por %%mm2, %%mm4 \n\t"\
"por %%mm3, %%mm4 \n\t"\
"packssdw %%mm4,%%mm4 \n\t"\
"movd %%mm4, %%eax \n\t"\
"orl %%eax, %%eax \n\t"\
"jz " #bt " \n\t"\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
#rounder ", %%mm4 \n\t"\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq 56(%2), %%mm5 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
#rounder ", %%mm0 \n\t"\
"paddd %%mm0, %%mm1 \n\t" /* A1 a1 */\
"paddd %%mm0, %%mm0 \n\t" \
"psubd %%mm1, %%mm0 \n\t" /* A2 a2 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm5, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm5 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm5 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm1, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm5, %%mm1 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm5, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm1, %%mm7 \n\t" /* A1+B1 a1+b1 A0+B0 a0+b0 */\
"packssdw %%mm4, %%mm2 \n\t" /* A0-B0 a0-b0 A1-B1 a1-b1 */\
"movq %%mm7, " #dst " \n\t"\
"movq " #src1 ", %%mm1 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"movq %%mm2, 24+" #dst " \n\t"\
"pmaddwd %%mm1, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm1 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm0, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm0 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm0 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm1, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"packssdw %%mm6, %%mm2 \n\t" /* A3+B3 a3+b3 A2+B2 a2+b2 */\
"movq %%mm2, 8+" #dst " \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm0, %%mm4 \n\t" /* A2-B2 a2-b2 A3-B3 a3-b3 */\
"movq %%mm4, 16+" #dst " \n\t"\
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#define ROW_IDCT(src0, src4, src1, src5, dst, rounder, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
#rounder ", %%mm4 \n\t"\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq 56(%2), %%mm5 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
#rounder ", %%mm0 \n\t"\
"paddd %%mm0, %%mm1 \n\t" /* A1 a1 */\
"paddd %%mm0, %%mm0 \n\t" \
"psubd %%mm1, %%mm0 \n\t" /* A2 a2 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm5, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm5 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm5 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm5 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm1, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm5, %%mm1 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm5, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm1, %%mm7 \n\t" /* A1+B1 a1+b1 A0+B0 a0+b0 */\
"packssdw %%mm4, %%mm2 \n\t" /* A0-B0 a0-b0 A1-B1 a1-b1 */\
"movq %%mm7, " #dst " \n\t"\
"movq " #src1 ", %%mm1 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"movq %%mm2, 24+" #dst " \n\t"\
"pmaddwd %%mm1, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm1 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm0, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm0 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm0 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm1, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"packssdw %%mm6, %%mm2 \n\t" /* A3+B3 a3+b3 A2+B2 a2+b2 */\
"movq %%mm2, 8+" #dst " \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm0, %%mm4 \n\t" /* A2-B2 a2-b2 A3-B3 a3-b3 */\
"movq %%mm4, 16+" #dst " \n\t"\
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
//IDCT( src0, src4, src1, src5, dst, rounder, shift)
DC_COND_IDCT( 0(%0), 8(%0), 16(%0), 24(%0), 0(%1),paddd 8(%2), 11)
Z_COND_IDCT( 32(%0), 40(%0), 48(%0), 56(%0), 32(%1),paddd (%2), 11, 4f)
Z_COND_IDCT( 64(%0), 72(%0), 80(%0), 88(%0), 64(%1),paddd (%2), 11, 2f)
Z_COND_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11, 1f)
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"paddd %%mm1, %%mm0 \n\t" /* A1 a1 */\
"psubd %%mm1, %%mm5 \n\t" /* A2 a2 */\
"movq 56(%2), %%mm1 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm1, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm1 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm1 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm0, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm1, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm1, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm0 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm7, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"movd %%mm7, " #dst " \n\t"\
"packssdw %%mm0, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"movd %%mm0, 16+" #dst " \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"movd %%mm2, 96+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"movd %%mm4, 112+" #dst " \n\t"\
"movq " #src1 ", %%mm0 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm0 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm5, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm5 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm0, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"packssdw %%mm6, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movd %%mm2, 32+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A3-B3 a3-b3 */\
"packssdw %%mm5, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movd %%mm6, 48+" #dst " \n\t"\
"movd %%mm4, 64+" #dst " \n\t"\
"movd %%mm5, 80+" #dst " \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( (%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"jmp 9f \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Replace ASMALIGN() with .p2align This macro has unconditionally used .p2align for a long time and serves no useful purpose.
2011-01-16 00:09:35 +02:00
"# .p2align 4 \n\t"\
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"4: \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Z_COND_IDCT( 64(%0), 72(%0), 80(%0), 88(%0), 64(%1),paddd (%2), 11, 6f)
Z_COND_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11, 5f)
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"paddd %%mm1, %%mm0 \n\t" /* A1 a1 */\
"psubd %%mm1, %%mm5 \n\t" /* A2 a2 */\
"movq 56(%2), %%mm1 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
"movq 72(%2), %%mm7 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm1 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm1, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm0, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm7, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm7, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm0 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm1, %%mm1 \n\t" /* A0+B0 a0+b0 */\
"movd %%mm1, " #dst " \n\t"\
"packssdw %%mm0, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"movd %%mm0, 16+" #dst " \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"movd %%mm2, 96+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"movd %%mm4, 112+" #dst " \n\t"\
"movq 88(%2), %%mm1 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm5, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm1, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm1, %%mm5 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"movq %%mm6, %%mm1 \n\t" /* A3 a3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm1 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"psrad $" #shift ", %%mm1 \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"packssdw %%mm6, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movd %%mm2, 32+" #dst " \n\t"\
"packssdw %%mm1, %%mm1 \n\t" /* A3-B3 a3-b3 */\
"packssdw %%mm5, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movd %%mm6, 48+" #dst " \n\t"\
"movd %%mm1, 64+" #dst " \n\t"\
"movd %%mm5, 80+" #dst " \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( (%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"jmp 9f \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Replace ASMALIGN() with .p2align This macro has unconditionally used .p2align for a long time and serves no useful purpose.
2011-01-16 00:09:35 +02:00
"# .p2align 4 \n\t"\
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"6: \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Z_COND_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11, 7f)
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 56(%2), %%mm1 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
"movq 72(%2), %%mm7 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm1 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm1, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm0, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm7, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm7, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm0 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm1, %%mm1 \n\t" /* A0+B0 a0+b0 */\
"movd %%mm1, " #dst " \n\t"\
"packssdw %%mm0, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"movd %%mm0, 16+" #dst " \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"movd %%mm2, 96+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"movd %%mm4, 112+" #dst " \n\t"\
"movq 88(%2), %%mm1 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm5, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm1, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm1, %%mm5 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"movq %%mm6, %%mm1 \n\t" /* A3 a3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm1 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"psrad $" #shift ", %%mm1 \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"packssdw %%mm6, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movd %%mm2, 32+" #dst " \n\t"\
"packssdw %%mm1, %%mm1 \n\t" /* A3-B3 a3-b3 */\
"packssdw %%mm5, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movd %%mm6, 48+" #dst " \n\t"\
"movd %%mm1, 64+" #dst " \n\t"\
"movd %%mm5, 80+" #dst " \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( (%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"jmp 9f \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Replace ASMALIGN() with .p2align This macro has unconditionally used .p2align for a long time and serves no useful purpose.
2011-01-16 00:09:35 +02:00
"# .p2align 4 \n\t"\
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"2: \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Z_COND_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11, 3f)
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq " #src5 ", %%mm3 \n\t" /* R7 R5 r7 r5 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 56(%2), %%mm1 \n\t" /* C7 C5 C7 C5 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* C7R7+C5R5 C7r7+C5r5 */\
"pmaddwd 64(%2), %%mm2 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm1, %%mm7 \n\t" /* B0 b0 */\
"movq 72(%2), %%mm1 \n\t" /* -C5 -C1 -C5 -C1 */\
"pmaddwd %%mm3, %%mm1 \n\t" /* -C5R7-C1R5 -C5r7-C1r5 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"paddd %%mm2, %%mm1 \n\t" /* B1 b1 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm0, %%mm2 \n\t" /* A1 a1 */\
"paddd %%mm1, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm1, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm0 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm7, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"movd %%mm7, " #dst " \n\t"\
"packssdw %%mm0, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"movd %%mm0, 16+" #dst " \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A1-B1 a1-b1 */\
"movd %%mm2, 96+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"movd %%mm4, 112+" #dst " \n\t"\
"movq " #src1 ", %%mm0 \n\t" /* R3 R1 r3 r1 */\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"movq 88(%2), %%mm7 \n\t" /* C3 C7 C3 C7 */\
"pmaddwd 96(%2), %%mm0 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C3R7+C7R5 C3r7+C7r5 */\
"movq %%mm5, %%mm2 \n\t" /* A2 a2 */\
"pmaddwd 104(%2), %%mm3 \n\t" /* -C1R7+C3R5 -C1r7+C3r5 */\
"paddd %%mm7, %%mm4 \n\t" /* B2 b2 */\
"paddd %%mm4, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm5 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm2 \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm0, %%mm3 \n\t" /* B3 b3 */\
"paddd %%mm3, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm3, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm2, %%mm2 \n\t" /* A2+B2 a2+b2 */\
"packssdw %%mm6, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movd %%mm2, 32+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A3-B3 a3-b3 */\
"packssdw %%mm5, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movd %%mm6, 48+" #dst " \n\t"\
"movd %%mm4, 64+" #dst " \n\t"\
"movd %%mm5, 80+" #dst " \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( (%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"jmp 9f \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Replace ASMALIGN() with .p2align This macro has unconditionally used .p2align for a long time and serves no useful purpose.
2011-01-16 00:09:35 +02:00
"# .p2align 4 \n\t"\
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"3: \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 64(%2), %%mm3 \n\t"\
"pmaddwd %%mm2, %%mm3 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm0, %%mm1 \n\t" /* A1 a1 */\
"paddd %%mm3, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm3, %%mm1 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm0 \n\t"\
"psrad $" #shift ", %%mm1 \n\t"\
"packssdw %%mm7, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"movd %%mm7, " #dst " \n\t"\
"packssdw %%mm0, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"movd %%mm0, 16+" #dst " \n\t"\
"packssdw %%mm1, %%mm1 \n\t" /* A1-B1 a1-b1 */\
"movd %%mm1, 96+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"movd %%mm4, 112+" #dst " \n\t"\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"pmaddwd %%mm2, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"pmaddwd 96(%2), %%mm2 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"movq %%mm5, %%mm1 \n\t" /* A2 a2 */\
"paddd %%mm4, %%mm1 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm5 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm1 \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm2, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm2, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm1, %%mm1 \n\t" /* A2+B2 a2+b2 */\
"packssdw %%mm6, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movd %%mm1, 32+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A3-B3 a3-b3 */\
"packssdw %%mm5, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movd %%mm6, 48+" #dst " \n\t"\
"movd %%mm4, 64+" #dst " \n\t"\
"movd %%mm5, 80+" #dst " \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( (%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"jmp 9f \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Replace ASMALIGN() with .p2align This macro has unconditionally used .p2align for a long time and serves no useful purpose.
2011-01-16 00:09:35 +02:00
"# .p2align 4 \n\t"\
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"5: \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"paddd %%mm1, %%mm0 \n\t" /* A1 a1 */\
"psubd %%mm1, %%mm5 \n\t" /* A2 a2 */\
"movq 8+" #src0 ", %%mm2 \n\t" /* R4 R0 r4 r0 */\
"movq 8+" #src4 ", %%mm3 \n\t" /* R6 R2 r6 r2 */\
"movq 16(%2), %%mm1 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm2, %%mm1 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm7 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm7, %%mm2 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm7 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm3, %%mm7 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"pmaddwd 40(%2), %%mm3 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"paddd %%mm1, %%mm7 \n\t" /* A0 a0 */\
"paddd %%mm1, %%mm1 \n\t" /* 2C0 2c0 */\
"psubd %%mm7, %%mm1 \n\t" /* A3 a3 */\
"paddd %%mm2, %%mm3 \n\t" /* A1 a1 */\
"paddd %%mm2, %%mm2 \n\t" /* 2C1 2c1 */\
"psubd %%mm3, %%mm2 \n\t" /* A2 a2 */\
"psrad $" #shift ", %%mm4 \n\t"\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm3 \n\t"\
"packssdw %%mm7, %%mm4 \n\t" /* A0 a0 */\
"movq %%mm4, " #dst " \n\t"\
"psrad $" #shift ", %%mm0 \n\t"\
"packssdw %%mm3, %%mm0 \n\t" /* A1 a1 */\
"movq %%mm0, 16+" #dst " \n\t"\
"movq %%mm0, 96+" #dst " \n\t"\
"movq %%mm4, 112+" #dst " \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"psrad $" #shift ", %%mm6 \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm2, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movq %%mm5, 32+" #dst " \n\t"\
"psrad $" #shift ", %%mm1 \n\t"\
"packssdw %%mm1, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movq %%mm6, 48+" #dst " \n\t"\
"movq %%mm6, 64+" #dst " \n\t"\
"movq %%mm5, 80+" #dst " \n\t"
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( 0(%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
//IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
//IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"jmp 9f \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Replace ASMALIGN() with .p2align This macro has unconditionally used .p2align for a long time and serves no useful purpose.
2011-01-16 00:09:35 +02:00
"# .p2align 4 \n\t"\
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"1: \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq " #src4 ", %%mm1 \n\t" /* R6 R2 r6 r2 */\
"movq " #src1 ", %%mm2 \n\t" /* R3 R1 r3 r1 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm5 \n\t" /* C6 C2 C6 C2 */\
"pmaddwd %%mm1, %%mm5 \n\t" /* C6R6+C2R2 C6r6+C2r2 */\
"movq 40(%2), %%mm6 \n\t" /* -C2 C6 -C2 C6 */\
"pmaddwd %%mm6, %%mm1 \n\t" /* -C2R6+C6R2 -C2r6+C6r2 */\
"movq %%mm4, %%mm6 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 48(%2), %%mm7 \n\t" /* C3 C1 C3 C1 */\
"pmaddwd %%mm2, %%mm7 \n\t" /* C3R3+C1R1 C3r3+C1r1 */\
"paddd %%mm5, %%mm4 \n\t" /* A0 a0 */\
"psubd %%mm5, %%mm6 \n\t" /* A3 a3 */\
"movq %%mm0, %%mm5 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"paddd %%mm1, %%mm0 \n\t" /* A1 a1 */\
"psubd %%mm1, %%mm5 \n\t" /* A2 a2 */\
"movq 64(%2), %%mm1 \n\t"\
"pmaddwd %%mm2, %%mm1 \n\t" /* -C7R3+C3R1 -C7r3+C3r1 */\
"paddd %%mm4, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"paddd %%mm4, %%mm4 \n\t" /* 2A0 2a0 */\
"psubd %%mm7, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"psrad $" #shift ", %%mm7 \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"movq %%mm0, %%mm3 \n\t" /* A1 a1 */\
"paddd %%mm1, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"psubd %%mm1, %%mm3 \n\t" /* A1-B1 a1-b1 */\
"psrad $" #shift ", %%mm0 \n\t"\
"psrad $" #shift ", %%mm3 \n\t"\
"packssdw %%mm7, %%mm7 \n\t" /* A0+B0 a0+b0 */\
"movd %%mm7, " #dst " \n\t"\
"packssdw %%mm0, %%mm0 \n\t" /* A1+B1 a1+b1 */\
"movd %%mm0, 16+" #dst " \n\t"\
"packssdw %%mm3, %%mm3 \n\t" /* A1-B1 a1-b1 */\
"movd %%mm3, 96+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A0-B0 a0-b0 */\
"movd %%mm4, 112+" #dst " \n\t"\
"movq 80(%2), %%mm4 \n\t" /* -C1 C5 -C1 C5 */\
"pmaddwd %%mm2, %%mm4 \n\t" /* -C1R3+C5R1 -C1r3+C5r1 */\
"pmaddwd 96(%2), %%mm2 \n\t" /* -C5R3+C7R1 -C5r3+C7r1 */\
"movq %%mm5, %%mm3 \n\t" /* A2 a2 */\
"paddd %%mm4, %%mm3 \n\t" /* A2+B2 a2+b2 */\
"psubd %%mm4, %%mm5 \n\t" /* a2-B2 a2-b2 */\
"psrad $" #shift ", %%mm3 \n\t"\
"psrad $" #shift ", %%mm5 \n\t"\
"movq %%mm6, %%mm4 \n\t" /* A3 a3 */\
"paddd %%mm2, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"psubd %%mm2, %%mm4 \n\t" /* a3-B3 a3-b3 */\
"psrad $" #shift ", %%mm6 \n\t"\
"packssdw %%mm3, %%mm3 \n\t" /* A2+B2 a2+b2 */\
"movd %%mm3, 32+" #dst " \n\t"\
"psrad $" #shift ", %%mm4 \n\t"\
"packssdw %%mm6, %%mm6 \n\t" /* A3+B3 a3+b3 */\
"movd %%mm6, 48+" #dst " \n\t"\
"packssdw %%mm4, %%mm4 \n\t" /* A3-B3 a3-b3 */\
"packssdw %%mm5, %%mm5 \n\t" /* A2-B2 a2-b2 */\
"movd %%mm4, 64+" #dst " \n\t"\
"movd %%mm5, 80+" #dst " \n\t"
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( (%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"jmp 9f \n\t"
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Replace ASMALIGN() with .p2align This macro has unconditionally used .p2align for a long time and serves no useful purpose.
2011-01-16 00:09:35 +02:00
"# .p2align 4 \n\t"
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"7: \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
#undef IDCT
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
#define IDCT(src0, src4, src1, src5, dst, shift) \
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
"movq " #src0 ", %%mm0 \n\t" /* R4 R0 r4 r0 */\
"movq 16(%2), %%mm4 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm0, %%mm4 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm5 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm5, %%mm0 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"psrad $" #shift ", %%mm4 \n\t"\
"psrad $" #shift ", %%mm0 \n\t"\
"movq 8+" #src0 ", %%mm2 \n\t" /* R4 R0 r4 r0 */\
"movq 16(%2), %%mm1 \n\t" /* C4 C4 C4 C4 */\
"pmaddwd %%mm2, %%mm1 \n\t" /* C4R4+C4R0 C4r4+C4r0 */\
"movq 24(%2), %%mm7 \n\t" /* -C4 C4 -C4 C4 */\
"pmaddwd %%mm7, %%mm2 \n\t" /* -C4R4+C4R0 -C4r4+C4r0 */\
"movq 32(%2), %%mm7 \n\t" /* C6 C2 C6 C2 */\
"psrad $" #shift ", %%mm1 \n\t"\
"packssdw %%mm1, %%mm4 \n\t" /* A0 a0 */\
"movq %%mm4, " #dst " \n\t"\
"psrad $" #shift ", %%mm2 \n\t"\
"packssdw %%mm2, %%mm0 \n\t" /* A1 a1 */\
"movq %%mm0, 16+" #dst " \n\t"\
"movq %%mm0, 96+" #dst " \n\t"\
"movq %%mm4, 112+" #dst " \n\t"\
"movq %%mm0, 32+" #dst " \n\t"\
"movq %%mm4, 48+" #dst " \n\t"\
"movq %%mm4, 64+" #dst " \n\t"\
"movq %%mm0, 80+" #dst " \n\t"
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
Support for MacIntel, take xx: '/nop' illegal for old versions of GAS Patch by John Dalgliesh % johnd AH defyne P org % Original thread: Date: Aug 8, 2006 8:12 PM Subject: Re: [Ffmpeg-devel] [PATCH] '/nop' illegal for old versions of GAS Originally committed as revision 5972 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 18:26:18 +03:00
//IDCT( src0, src4, src1, src5, dst, shift)
IDCT( 0(%1), 64(%1), 32(%1), 96(%1), 0(%0), 20)
//IDCT( 8(%1), 72(%1), 40(%1), 104(%1), 4(%0), 20)
IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20)
//IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20)
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
#endif
/*
Input
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
00 40 04 44 20 60 24 64
10 30 14 34 50 70 54 74
01 41 03 43 21 61 23 63
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
11 31 13 33 51 71 53 73
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
02 42 06 46 22 62 26 66
12 32 16 36 52 72 56 76
05 45 07 47 25 65 27 67
15 35 17 37 55 75 57 77
COSMETICS: Remove all trailing whitespace. Originally committed as revision 4749 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-17 20:14:38 +02:00
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
Temp
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
00 04 10 14 20 24 30 34
40 44 50 54 60 64 70 74
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
01 03 11 13 21 23 31 33
41 43 51 53 61 63 71 73
(commit by michael) faster simple idct in MMX Originally committed as revision 269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-17 22:00:41 +02:00
02 06 12 16 22 26 32 36
42 46 52 56 62 66 72 76
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
05 07 15 17 25 27 35 37
45 47 55 57 65 67 75 77
*/
"9: \n\t"
COSMETICS: tabs --> spaces, some prettyprinting Originally committed as revision 4764 to svn://svn.ffmpeg.org/ffmpeg/trunk
2005-12-22 03:10:11 +02:00
:: "r" (block), "r" (temp), "r" (coeffs)
: "%eax"
);
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
}
idct permutation cleanup, idct can be selected per context now fixing some threadunsafe code Originally committed as revision 980 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-30 01:44:22 +03:00
void ff_simple_idct_mmx(int16_t *block)
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
{
idct permutation cleanup, idct can be selected per context now fixing some threadunsafe code Originally committed as revision 980 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-30 01:44:22 +03:00
idct(block);
}
//FIXME merge add/put into the idct
* UINTX -> uintx_t INTX -> intx_t Originally committed as revision 1578 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-02-11 18:35:48 +02:00
void ff_simple_idct_put_mmx(uint8_t *dest, int line_size, DCTELEM *block)
idct permutation cleanup, idct can be selected per context now fixing some threadunsafe code Originally committed as revision 980 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-30 01:44:22 +03:00
{
idct(block);
Put ff_ prefix on non-static {put_signed,put,add}_pixels_clamped_mmx() functions. Originally committed as revision 24987 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-08-30 19:22:27 +03:00
ff_put_pixels_clamped_mmx(block, dest, line_size);
idct permutation cleanup, idct can be selected per context now fixing some threadunsafe code Originally committed as revision 980 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-30 01:44:22 +03:00
}
* UINTX -> uintx_t INTX -> intx_t Originally committed as revision 1578 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-02-11 18:35:48 +02:00
void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, DCTELEM *block)
idct permutation cleanup, idct can be selected per context now fixing some threadunsafe code Originally committed as revision 980 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-30 01:44:22 +03:00
{
idct(block);
Put ff_ prefix on non-static {put_signed,put,add}_pixels_clamped_mmx() functions. Originally committed as revision 24987 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-08-30 19:22:27 +03:00
ff_add_pixels_clamped_mmx(block, dest, line_size);
new IDCT code by Michael Niedermayer (michaelni@gmx.at) - #define SIMPLE_IDCT to enable Originally committed as revision 235 to svn://svn.ffmpeg.org/ffmpeg/trunk
2001-12-09 14:39:54 +02:00
}
Reference in New Issue Copy Permalink