Commit Graph

276 Commits

Author SHA1 Message Date
Aurelien Jacobs
1dac8fea05 Enables back the mmx/sse optimized version of the vp3 idct.
It generates different md5sum than the reference C implementation,
but no visual difference, so enabled only when bitexact is not set.

Originally committed as revision 6241 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-09-12 20:58:17 +00:00
Diego Biurrun
04d7f60143 Add official LGPL license headers to the files that were missing them.
Originally committed as revision 6219 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-09-10 14:02:42 +00:00
Måns Rullgård
0e176c3eb5 remove redundant declarations
Originally committed as revision 6153 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-09-02 23:10:28 +00:00
Loren Merritt
3e20143ee7 mmx implementation of deblocking strength decision.
2-3% faster h264.

Originally committed as revision 6113 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-28 09:33:01 +00:00
Loren Merritt
1e4ecf26f5 ff_fft_calc_3dn/3dn2/sse: convert intrinsics to inline asm.
2.5% faster fft, 0.5% faster vorbis.

Originally committed as revision 6023 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-18 23:53:49 +00:00
Michael Niedermayer
cf5aed5bad simplify
Originally committed as revision 6020 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-18 10:43:23 +00:00
Michael Niedermayer
3829a62eae insufficient alignment
Originally committed as revision 6006 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-17 07:49:22 +00:00
Marco Manfredini
6bb9e49249 Fix building with --disable-opts but MMX enabled.
patch by Marco Manfredini mldb %at% gmx %dot% net

Originally committed as revision 5994 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-13 10:10:06 +00:00
John Dalgliesh
4454dc1b6f Support for MacIntel, last part: balign directives
Determines whether .align's arg is power-of-two or not, then defines ASMALIGN appropriately in config.h. Changes all .baligns to ASMALIGNs.
Patch by John Dalgliesh % johnd AH defyne P org %
Original thread:
Date: Aug 11, 2006 8:00 AM
Subject: Re: [Ffmpeg-devel] Mac OS X Intel last part: balign directives

Originally committed as revision 5990 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-12 16:37:31 +00:00
Loren Merritt
069720565c vorbis simd tweaks
Originally committed as revision 5983 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-11 18:19:37 +00:00
Michael Niedermayer
1f1aa1d955 convert vector_fmul_reverse_sse2 and vector_fmul_add_add_sse2 to sse
please complain if they are slower on sse2 cpus ...

Originally committed as revision 5976 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 20:24:58 +00:00
Loren Merritt
eb4825b5d4 sse and 3dnow implementations of float->int conversion and mdct windowing.
15% faster vorbis.

Originally committed as revision 5975 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 19:06:25 +00:00
Luca Barbato
ffad4ed154 Fix x86 SIMD asm and pic, patch from Martin von Gagern <Martin.vGagern@gmx.net>
Originally committed as revision 5973 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 16:05:29 +00:00
John Dalgliesh
347be47226 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 15:26:18 +00:00
John Dalgliesh
0fc256f3d9 Add support for Mac OS X Intel part 2: Assembler macros in fdct_mmx.c
convert gas macros to ccp macros
Patch by John Dalgliesh % johnd AH defyne P org %
Original thread:
Date: Aug 10, 2006 5:39 AM
Subject: Re: [Ffmpeg-devel] Mac OS X Intel part 2: Assembler macros in fdct_mmx.c

Originally committed as revision 5971 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 11:29:57 +00:00
John Dalgliesh
fc48b6fe74 Support for Mac OS X Intel, part 3: binary integer constants:
Apple's assembler only understands the same integer constants as C does: hex, decimal, octal. It doesn't understand binary integer constants (0b...) so this patch replaces binary integer constants with hex ones.
Patch by John Dalgliesh % johnd AH defyne P org %
Original thread:
Date: Aug 10, 2006 8:16 AM
Subject: [Ffmpeg-devel] Mac OS X Intel part 3: binary integer constants

Originally committed as revision 5970 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-10 09:06:06 +00:00
Loren Merritt
ee5df92750 emms -> femms
Originally committed as revision 5965 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-09 06:37:25 +00:00
Loren Merritt
2494bdd90d gcc 2.95 and 3.4.x on x86 32bit without fomit-frame-pointer can't even find 5 registers for asm input.
0.5% slower vorbis.

Originally committed as revision 5964 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-09 06:33:49 +00:00
Loren Merritt
1b87c40245 slightly faster ff_imdct_calc_3dn2() on amd64. (gcc added a bunch of useless movsxd)
Originally committed as revision 5962 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-08 21:47:11 +00:00
Michael Niedermayer
21bb884fb7 change vorbis_inverse_coupling_sse2() so it works on sse1 cpus
Originally committed as revision 5957 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-08 12:03:51 +00:00
Loren Merritt
bcfa3e58ee 3dnow2 implementation of imdct.
6% faster vorbis and wma.

Originally committed as revision 5954 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-08 04:01:04 +00:00
Loren Merritt
cd035a6051 10l, vorbis_inverse_coupling_sse() was really 3dnow
Originally committed as revision 5903 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-03 07:09:29 +00:00
Loren Merritt
2dac4acfc0 sse & sse2 implementations of vorbis channel coupling.
9% faster vorbis (on a K8).

Originally committed as revision 5898 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-08-03 03:18:47 +00:00
Stefan Gehrer
595e7bd940 some MMX optimizations for the CAVS decoder
Originally committed as revision 5846 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-07-29 08:45:33 +00:00
Michael Niedermayer
5ced7b80ad disable the vp3 mmx and sse2 idcts, their output doesnt match the c idct (tested with -f crc) and the theora spec does not allow different idcts not to mention the difference is quite vissible ...
Originally committed as revision 5788 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-07-19 09:49:21 +00:00
Måns Rullgård
98d417cbcd #define SBUTTERFLY outside CONFIG_ENCODERS
Originally committed as revision 5628 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-07-05 19:31:01 +00:00
Luca Abeni
9c39071d6d Move REG_* macros from libavcodec/i386/mmx.h to libavutil/x86_cpu.h
Originally committed as revision 5595 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-07-03 10:52:07 +00:00
Måns Rullgård
3f8674a902 remove redundant macro definitions
Originally committed as revision 5589 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-07-02 22:01:31 +00:00
Måns Rullgård
8fb0d07339 kill warning
Originally committed as revision 5588 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-07-02 21:53:30 +00:00
Michael Niedermayer
e27b6e62f7 missmatch control for mpeg2 intra dequantization if bitexact=1
Originally committed as revision 5328 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-04-28 17:03:52 +00:00
Zuxy Meng
392f6da897 Remove unused and unsupported Cyrix's "Extended MMX",
Add SSE3 support.
Patch by Zuxy Meng < zuxy POIS meng AH gmail POIS com >
Original thread:
04/26/06 13:13:
[Ffmpeg-devel] [PATCH] Bug fix,	SSE3 support in i386/cputest.c and dsputil.h

Originally committed as revision 5326 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-04-28 07:46:13 +00:00
Wolfram Gloger
f42635f558 gcc-2.95 compile fix, patch by Wolfram Gloger <wmglo A dent PIS med PIS uni-muenchen PIS de>
Originally committed as revision 5298 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-04-18 03:48:30 +00:00
Loren Merritt
75ca1a5f70 gmc_mmx tweaks
Originally committed as revision 5269 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-04-05 04:13:41 +00:00
Loren Merritt
703c8195a8 mmx implementation of 3-point GMC. (5x faster than C)
Originally committed as revision 5265 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-04-04 09:23:45 +00:00
Luca Barbato
22b48b85b6 altivec support for snow
Originally committed as revision 5228 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-27 12:51:19 +00:00
Loren Merritt
5e8b787afa simplified and slightly faster h264_chroma_mc8_mmx
Originally committed as revision 5214 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-25 08:41:14 +00:00
Loren Merritt
513fbd8e5a prefetch pixels for future motion compensation. 2-5% faster h264.
Originally committed as revision 5203 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-23 20:16:36 +00:00
Loren Merritt
5e6a5c4daf 10l
Originally committed as revision 5201 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-22 22:08:28 +00:00
Loren Merritt
fdd3057981 added mmx implementation of h264_chroma_mc2
Originally committed as revision 5200 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-22 22:05:00 +00:00
Robert Edele
e8600e5edc add MMX and SSE versions of ff_snow_inner_add_yblock
Patch by Robert Edele < yartrebo AH earthlink POIS net >
Original Thread:
Date: Mar 22, 2006 3:24 AM
Subject: [Ffmpeg-devel] [PATCH] snow mmx + sse2 part 5

Originally committed as revision 5197 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-22 12:08:35 +00:00
Robert Edele
2c9a0285d4 snow mmx+sse2 optimizations, part 4
Patch by Robert Edele, yartrebo <<at>> earthlink <<dot>> net

Originally committed as revision 5191 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-21 21:51:07 +00:00
Robert Edele
4567b4bdab Add the mmx and sse2 implementations of ff_snow_vertical_compose().
Patch by Robert Edele < yartrebo AH earthlink POIS net >
Original thread:
Date: Mar 20, 2006 5:54 PM
Subject: [Ffmpeg-devel] [PATCH] snow mmx + sse2 part 3

Originally committed as revision 5185 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-20 22:27:59 +00:00
Robert Edele
059715a41c First part of a series of speed-enchancing patches.
This one sets up a snow.h and makes snow use the dsputil function pointer
framework to access the three functions that will be implemented in asm
in the other parts of the patchset.
Patch by Robert Edele < yartrebo AH earthlink POIS net>
Original thread:
Subject: [Ffmpeg-devel] [PATCH] Snow mmx+sse2 asm optimizations
Date: Sun, 05 Feb 2006 12:47:14 -0500

Originally committed as revision 5172 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-16 19:18:18 +00:00
Zuxy Meng
82eb4b0f1b 3DNow! & Extended 3DNow! versions of FFT
Patch by Zuxy Meng, zuxy <<dot>> meng >>at<< gmail <<dot>> com
Minor non-functional diff-related fixes by me.

Originally committed as revision 5125 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-08 04:13:55 +00:00
Loren Merritt
548a1c8a35 h264_idct8_add_mmx
Originally committed as revision 5123 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-07 22:45:56 +00:00
Loren Merritt
6da971f160 h264_idct_add only needs mmx1
Originally committed as revision 5122 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-07 22:33:32 +00:00
Zuxy Meng
2ffb22d2ad use xorps instead of mulps to toggle the sign of a float, as suggested by Software Optimization Guide for AMD64 Processors.
Patch by Zuxy Meng < zuxy POIS meng AH  gmail POIS com > OKed by Michael
Original thread:
Date: Mar 5, 2006 8:15 PM
Subject: [Ffmpeg-devel] [PATCH] Little optimization to fft_sse.c

Originally committed as revision 5112 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-05 20:25:18 +00:00
Loren Merritt
d84f7c61ee gcc2.95 workaround
Originally committed as revision 5111 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-05 19:02:35 +00:00
Loren Merritt
7a5b2fa812 remove some useless instructions
Originally committed as revision 5109 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-04 19:56:01 +00:00
Loren Merritt
6a8eb0f45a 4% faster h264_qpel_mc
Originally committed as revision 5094 to svn://svn.ffmpeg.org/ffmpeg/trunk
2006-03-02 08:21:08 +00:00