]> git.sesse.net Git - x264/log
x264
16 years agodon't overwrite pthread* namespace, because system headers might define those functio...
Loren Merritt [Mon, 24 Sep 2007 11:24:28 +0000 (11:24 +0000)]
don't overwrite pthread* namespace, because system headers might define those functions even if we don't want them

git-svn-id: svn://svn.videolan.org/x264/trunk@679 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agofaster 4x4 sad
Loren Merritt [Fri, 21 Sep 2007 20:20:22 +0000 (20:20 +0000)]
faster 4x4 sad

git-svn-id: svn://svn.videolan.org/x264/trunk@678 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agofix an arithmetic overflow in trellis at high qp.
Loren Merritt [Thu, 20 Sep 2007 08:10:45 +0000 (08:10 +0000)]
fix an arithmetic overflow in trellis at high qp.

git-svn-id: svn://svn.videolan.org/x264/trunk@677 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agoimplement multithreaded me=esa
Loren Merritt [Sat, 15 Sep 2007 06:34:05 +0000 (06:34 +0000)]
implement multithreaded me=esa

git-svn-id: svn://svn.videolan.org/x264/trunk@676 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agofix some integer overflows. now vbv size can exceed 2 Gbit.
Loren Merritt [Wed, 12 Sep 2007 05:42:23 +0000 (05:42 +0000)]
fix some integer overflows. now vbv size can exceed 2 Gbit.

git-svn-id: svn://svn.videolan.org/x264/trunk@675 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agoallow --vbv-init to take absolute values (in kbit), in addition to the previous fract...
Loren Merritt [Sun, 9 Sep 2007 03:17:20 +0000 (03:17 +0000)]
allow --vbv-init to take absolute values (in kbit), in addition to the previous fractions of vbv-bufsize.

git-svn-id: svn://svn.videolan.org/x264/trunk@674 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agoremove a bashism
Loren Merritt [Fri, 7 Sep 2007 20:40:13 +0000 (20:40 +0000)]
remove a bashism

git-svn-id: svn://svn.videolan.org/x264/trunk@673 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agoreorder headers so that largefile support is defined before the first copy of stdio
Loren Merritt [Sun, 2 Sep 2007 04:32:17 +0000 (04:32 +0000)]
reorder headers so that largefile support is defined before the first copy of stdio

git-svn-id: svn://svn.videolan.org/x264/trunk@672 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agoregression in r669: broke saving of configure args if make has to re-run configure
Loren Merritt [Mon, 20 Aug 2007 16:44:42 +0000 (16:44 +0000)]
regression in r669: broke saving of configure args if make has to re-run configure

git-svn-id: svn://svn.videolan.org/x264/trunk@671 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agoregression in r669: --enable-shared should imply --enable-pic on some archs.
Loren Merritt [Sat, 18 Aug 2007 01:13:22 +0000 (01:13 +0000)]
regression in r669: --enable-shared should imply --enable-pic on some archs.

git-svn-id: svn://svn.videolan.org/x264/trunk@670 df754926-b1dd-0310-bc7b-ec298dee348c

16 years ago* Add a --host flag to allow overriding config.guess; this is particularly
Loïc Minier [Sun, 12 Aug 2007 12:46:15 +0000 (12:46 +0000)]
* Add a --host flag to allow overriding config.guess; this is particularly
  useful with a 64-bits kernel running a 32-bits userland to build 32-bits
  apps.
* Normalize any host triplet into a quadruplet via config.sub.
* Move option parsing before any use of architecture information.

git-svn-id: svn://svn.videolan.org/x264/trunk@669 df754926-b1dd-0310-bc7b-ec298dee348c

16 years ago* Update config.guess.
Loïc Minier [Sun, 12 Aug 2007 12:36:23 +0000 (12:36 +0000)]
* Update config.guess.

git-svn-id: svn://svn.videolan.org/x264/trunk@668 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agomingw doesn't have strtok_r
Loren Merritt [Tue, 17 Jul 2007 11:24:26 +0000 (11:24 +0000)]
mingw doesn't have strtok_r

git-svn-id: svn://svn.videolan.org/x264/trunk@667 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agomove os/compiler specific defines to their own header
Loren Merritt [Tue, 17 Jul 2007 11:11:19 +0000 (11:11 +0000)]
move os/compiler specific defines to their own header

git-svn-id: svn://svn.videolan.org/x264/trunk@666 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agoextend zones to support (some) encoding parameters in addition to ratecontrol.
Loren Merritt [Thu, 12 Jul 2007 23:48:23 +0000 (23:48 +0000)]
extend zones to support (some) encoding parameters in addition to ratecontrol.

git-svn-id: svn://svn.videolan.org/x264/trunk@665 df754926-b1dd-0310-bc7b-ec298dee348c

16 years agocosmetics
Loren Merritt [Fri, 6 Jul 2007 17:08:26 +0000 (17:08 +0000)]
cosmetics

git-svn-id: svn://svn.videolan.org/x264/trunk@664 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agolimit vertical motion vectors to +/-512, since some decoders actually depend on that...
Loren Merritt [Thu, 28 Jun 2007 21:26:21 +0000 (21:26 +0000)]
limit vertical motion vectors to +/-512, since some decoders actually depend on that limit.

git-svn-id: svn://svn.videolan.org/x264/trunk@663 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoAdd vertical and horizontal luma deblocking accelerated with Altivec,
Guillaume Poirier [Fri, 22 Jun 2007 21:42:41 +0000 (21:42 +0000)]
Add vertical and horizontal luma deblocking accelerated with Altivec,
based on Graham Booker's code written for FFmpeg with slight modifications
to re-use x264's macros

git-svn-id: svn://svn.videolan.org/x264/trunk@662 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocosmetics in cpu detection
Loren Merritt [Sat, 16 Jun 2007 04:09:01 +0000 (04:09 +0000)]
cosmetics in cpu detection

git-svn-id: svn://svn.videolan.org/x264/trunk@661 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix compilation without asm on x86_32 (r658 worked only on x86_64).
Loren Merritt [Sat, 16 Jun 2007 04:02:48 +0000 (04:02 +0000)]
fix compilation without asm on x86_32 (r658 worked only on x86_64).

git-svn-id: svn://svn.videolan.org/x264/trunk@660 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoexempt 1080p from the non-mod16 warning.
Loren Merritt [Sun, 10 Jun 2007 23:46:31 +0000 (23:46 +0000)]
exempt 1080p from the non-mod16 warning.

git-svn-id: svn://svn.videolan.org/x264/trunk@659 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoallow compiling without yasm/nasm on x86 and x86-64 platforms
Alex Izvorski [Tue, 5 Jun 2007 18:38:31 +0000 (18:38 +0000)]
allow compiling without yasm/nasm on x86 and x86-64 platforms

git-svn-id: svn://svn.videolan.org/x264/trunk@658 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoupdated MS VC8/VC7 build, patch by Gabriel Bouvigne
Alex Izvorski [Tue, 5 Jun 2007 18:32:13 +0000 (18:32 +0000)]
updated MS VC8/VC7 build, patch by Gabriel Bouvigne

git-svn-id: svn://svn.videolan.org/x264/trunk@657 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoreplace alloca with malloc everywhere. per manpage, use of alloca is discouraged...
Alex Izvorski [Sat, 26 May 2007 03:13:08 +0000 (03:13 +0000)]
replace alloca with malloc everywhere. per manpage, use of alloca is discouraged. this may have a minor effect on the speed of ssim and esa, but that appears too small to measure.

git-svn-id: svn://svn.videolan.org/x264/trunk@656 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agorequire a ratecontrol method to be specified, it no longer defaults to cqp=26.
Loren Merritt [Wed, 2 May 2007 21:33:43 +0000 (21:33 +0000)]
require a ratecontrol method to be specified, it no longer defaults to cqp=26.

git-svn-id: svn://svn.videolan.org/x264/trunk@655 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix nnz computation in cavlc+8x8dct+deblock. (regression in r607)
Loren Merritt [Mon, 23 Apr 2007 08:34:52 +0000 (08:34 +0000)]
fix nnz computation in cavlc+8x8dct+deblock. (regression in r607)

git-svn-id: svn://svn.videolan.org/x264/trunk@654 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix the computation of bits used for vbv. (regression in r651)
Loren Merritt [Mon, 23 Apr 2007 07:09:30 +0000 (07:09 +0000)]
fix the computation of bits used for vbv. (regression in r651)

git-svn-id: svn://svn.videolan.org/x264/trunk@653 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoc89 compile fix
Loren Merritt [Sun, 22 Apr 2007 03:37:56 +0000 (03:37 +0000)]
c89 compile fix

git-svn-id: svn://svn.videolan.org/x264/trunk@652 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocabac: use bytestream instead of bitstream.
Loren Merritt [Sat, 21 Apr 2007 11:32:34 +0000 (11:32 +0000)]
cabac: use bytestream instead of bitstream.
35% faster cabac, 20% faster overall lossless, ~1% faster overall at normal bitrates.

git-svn-id: svn://svn.videolan.org/x264/trunk@651 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoremove the restriction on number of threads as a function of resolution (it was wrong...
Loren Merritt [Wed, 11 Apr 2007 22:21:15 +0000 (22:21 +0000)]
remove the restriction on number of threads as a function of resolution (it was wrong anyway in the presence of B-frames), and raise the max number of threads in general (though more will have to be done before it can really scale to lots of cores).

git-svn-id: svn://svn.videolan.org/x264/trunk@650 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agotweak ssse3 quant
Loren Merritt [Tue, 10 Apr 2007 22:37:18 +0000 (22:37 +0000)]
tweak ssse3 quant

git-svn-id: svn://svn.videolan.org/x264/trunk@649 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agochange some tables from int to int8_t. 13KB smaller executable.
Loren Merritt [Sat, 7 Apr 2007 04:53:16 +0000 (04:53 +0000)]
change some tables from int to int8_t. 13KB smaller executable.

git-svn-id: svn://svn.videolan.org/x264/trunk@648 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofaster cabac rdo. up to 10% faster at q0, but negligible at normal bitrates.
Loren Merritt [Fri, 6 Apr 2007 21:45:33 +0000 (21:45 +0000)]
faster cabac rdo. up to 10% faster at q0, but negligible at normal bitrates.

git-svn-id: svn://svn.videolan.org/x264/trunk@647 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoworkaround gcc's inability to align variables on the stack.
Loren Merritt [Fri, 6 Apr 2007 21:17:34 +0000 (21:17 +0000)]
workaround gcc's inability to align variables on the stack.
this crash was introduced in r642, but only because previous versions didn't use sse2 on the stack.

git-svn-id: svn://svn.videolan.org/x264/trunk@646 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago32bit version of ssse3 satd.
Loren Merritt [Thu, 5 Apr 2007 16:11:03 +0000 (16:11 +0000)]
32bit version of ssse3 satd.
switch default assembler to yasm. it will still fallback to nasm if you don't have yasm.

git-svn-id: svn://svn.videolan.org/x264/trunk@645 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agosimplify trellis
Loren Merritt [Wed, 4 Apr 2007 19:34:02 +0000 (19:34 +0000)]
simplify trellis

git-svn-id: svn://svn.videolan.org/x264/trunk@644 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix an arithmetic overflow in trellis with QP >= 42
Loren Merritt [Wed, 4 Apr 2007 18:59:20 +0000 (18:59 +0000)]
fix an arithmetic overflow in trellis with QP >= 42

git-svn-id: svn://svn.videolan.org/x264/trunk@643 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago2x faster quant. 2% overall.
Loren Merritt [Wed, 4 Apr 2007 18:45:25 +0000 (18:45 +0000)]
2x faster quant. 2% overall.
side effects:
not bit-identical to the previous algorithm.
while the new algorithm covers a wider range of cqms than the previous one did,
I couldn't find a good way to fallback to a general version for the extreme
cqms. so now it refuses to encode extreme cqms instead of just being slower.
lays a framework for custom deadzone matrices, though I didn't add an api.

git-svn-id: svn://svn.videolan.org/x264/trunk@642 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agowhen encoding with a cqm, probe_skip now also uses the cqm, instead of the flat matrix
Loren Merritt [Wed, 4 Apr 2007 18:35:51 +0000 (18:35 +0000)]
when encoding with a cqm, probe_skip now also uses the cqm, instead of the flat matrix

git-svn-id: svn://svn.videolan.org/x264/trunk@641 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocosmetics in asm macros
Loren Merritt [Wed, 4 Apr 2007 00:48:55 +0000 (00:48 +0000)]
cosmetics in asm macros

git-svn-id: svn://svn.videolan.org/x264/trunk@640 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agouse only c-style comments in public header (patch by Vincent Torres)
Alex Izvorski [Tue, 3 Apr 2007 17:18:17 +0000 (17:18 +0000)]
use only c-style comments in public header (patch by Vincent Torres)

git-svn-id: svn://svn.videolan.org/x264/trunk@639 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoin hpel search, merge two 16x16 mc calls into one 16x17. 15% faster hpel, .3% overall.
Loren Merritt [Mon, 2 Apr 2007 23:56:09 +0000 (23:56 +0000)]
in hpel search, merge two 16x16 mc calls into one 16x17. 15% faster hpel, .3% overall.

git-svn-id: svn://svn.videolan.org/x264/trunk@638 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoCompile fix
Christophe Mutricy [Mon, 2 Apr 2007 19:17:28 +0000 (19:17 +0000)]
Compile fix

git-svn-id: svn://svn.videolan.org/x264/trunk@637 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoremove private stuff from public headers. no more need for -D__X264__
Loren Merritt [Fri, 30 Mar 2007 20:20:36 +0000 (20:20 +0000)]
remove private stuff from public headers. no more need for -D__X264__

git-svn-id: svn://svn.videolan.org/x264/trunk@636 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoadjust bitstream buffer sizes for very large frames
Loren Merritt [Sat, 24 Mar 2007 12:58:27 +0000 (12:58 +0000)]
adjust bitstream buffer sizes for very large frames

git-svn-id: svn://svn.videolan.org/x264/trunk@635 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoconflate HAVE_MMXEXT with HAVE_SSE2, since they were never used distinctly.
Loren Merritt [Wed, 14 Mar 2007 22:41:50 +0000 (22:41 +0000)]
conflate HAVE_MMXEXT with HAVE_SSE2, since they were never used distinctly.

git-svn-id: svn://svn.videolan.org/x264/trunk@634 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago * Made -DNEED_ALTIVEC unnecessary, thanks to Guillaume Poirier.
Sam Hocevar [Wed, 14 Mar 2007 21:53:47 +0000 (21:53 +0000)]
  * Made -DNEED_ALTIVEC unnecessary, thanks to Guillaume Poirier.

git-svn-id: svn://svn.videolan.org/x264/trunk@633 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago * check x264_cpu_detect() before calling AltiVec functions.
Sam Hocevar [Wed, 14 Mar 2007 21:31:50 +0000 (21:31 +0000)]
  * check x264_cpu_detect() before calling AltiVec functions.

git-svn-id: svn://svn.videolan.org/x264/trunk@632 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agossse3 detection. x86_64 ssse3 satd and quant.
Loren Merritt [Wed, 14 Mar 2007 21:11:11 +0000 (21:11 +0000)]
ssse3 detection. x86_64 ssse3 satd and quant.
requires yasm >= 0.6.0

git-svn-id: svn://svn.videolan.org/x264/trunk@631 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago * Use -maltivec when building dependencies, or <altivec.h> cannot be used.
Sam Hocevar [Wed, 14 Mar 2007 20:40:24 +0000 (20:40 +0000)]
  * Use -maltivec when building dependencies, or <altivec.h> cannot be used.
  * Do not declare vectors in non-AltiVec files.

git-svn-id: svn://svn.videolan.org/x264/trunk@630 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago * common/cpu.c: runtime AltiVec autodetection on Linux.
Sam Hocevar [Wed, 14 Mar 2007 18:04:06 +0000 (18:04 +0000)]
  * common/cpu.c: runtime AltiVec autodetection on Linux.
  * configure, Makefile: do not build the whole project with -maltivec because
    it generates AltiVec code in weird places.

git-svn-id: svn://svn.videolan.org/x264/trunk@629 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix a small memleak.
Loren Merritt [Mon, 5 Mar 2007 15:35:42 +0000 (15:35 +0000)]
fix a small memleak.
patch by Limin Wang.

git-svn-id: svn://svn.videolan.org/x264/trunk@628 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocompile fix for GCC-3.3 on OSX, based on a patch by
Guillaume Poirier [Sat, 3 Mar 2007 12:59:23 +0000 (12:59 +0000)]
compile fix for GCC-3.3 on OSX, based on a patch by
Patrice Bensoussan % patrice P bensoussan A free P fr%
Note: regression test still do not pass with GCC-3.3,
but they never did as far as I can remember.

git-svn-id: svn://svn.videolan.org/x264/trunk@627 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocosmetics in regression test
Loren Merritt [Sat, 3 Mar 2007 12:12:54 +0000 (12:12 +0000)]
cosmetics in regression test

git-svn-id: svn://svn.videolan.org/x264/trunk@626 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoregression testing, run similar to fprofiled: VIDS='vid_720x480.yuv' make test
Alex Izvorski [Sat, 3 Mar 2007 11:44:01 +0000 (11:44 +0000)]
regression testing, run similar to fprofiled: VIDS='vid_720x480.yuv' make test

git-svn-id: svn://svn.videolan.org/x264/trunk@625 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoadd ability to generate doxygen documentation; make dox
Alex Izvorski [Wed, 28 Feb 2007 18:47:04 +0000 (18:47 +0000)]
add ability to generate doxygen documentation; make dox

git-svn-id: svn://svn.videolan.org/x264/trunk@624 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agooops, scenecut detection failed to activate when using threads and not using B-frames
Loren Merritt [Thu, 22 Feb 2007 05:01:38 +0000 (05:01 +0000)]
oops, scenecut detection failed to activate when using threads and not using B-frames

git-svn-id: svn://svn.videolan.org/x264/trunk@623 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoextras/getopt.c was BSD licensed. replace with a LGPL version (from glibc).
Loren Merritt [Mon, 29 Jan 2007 14:42:42 +0000 (14:42 +0000)]
extras/getopt.c was BSD licensed. replace with a LGPL version (from glibc).

git-svn-id: svn://svn.videolan.org/x264/trunk@622 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoFix build issues on Linux. Only gcc-4.x is supported, as on OSX.
Guillaume Poirier [Thu, 25 Jan 2007 08:32:16 +0000 (08:32 +0000)]
Fix build issues on Linux. Only gcc-4.x is supported, as on OSX.
Cleans up a few inconsistencies in the code too.

git-svn-id: svn://svn.videolan.org/x264/trunk@621 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agotweak block_residual_write_cavlc.
Loren Merritt [Sun, 21 Jan 2007 12:12:04 +0000 (12:12 +0000)]
tweak block_residual_write_cavlc.
up to 1% faster lossless, no difference at normal bitrates.

git-svn-id: svn://svn.videolan.org/x264/trunk@620 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agodon't assume int is exactly 4 bytes
Loren Merritt [Sat, 20 Jan 2007 05:07:44 +0000 (05:07 +0000)]
don't assume int is exactly 4 bytes

git-svn-id: svn://svn.videolan.org/x264/trunk@619 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agomake array_non_zero() compatible with -fstrict-aliasing
Loren Merritt [Thu, 11 Jan 2007 23:55:51 +0000 (23:55 +0000)]
make array_non_zero() compatible with -fstrict-aliasing

git-svn-id: svn://svn.videolan.org/x264/trunk@618 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoHonor CFLAGS and LDFLAGS set by the user
Christophe Mutricy [Tue, 9 Jan 2007 20:25:32 +0000 (20:25 +0000)]
Honor CFLAGS and LDFLAGS set by the user

git-svn-id: svn://svn.videolan.org/x264/trunk@617 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoCheck whether 'echo -n' works, otherwise try printf (fixes build on current OS X...
Eric Petit [Tue, 2 Jan 2007 14:51:10 +0000 (14:51 +0000)]
Check whether 'echo -n' works, otherwise try printf (fixes build on current OS X 10.5)

git-svn-id: svn://svn.videolan.org/x264/trunk@616 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoCheck version of nasm on OS X / Intel
Eric Petit [Mon, 1 Jan 2007 22:41:44 +0000 (22:41 +0000)]
Check version of nasm on OS X / Intel

git-svn-id: svn://svn.videolan.org/x264/trunk@615 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agowrong reference frames were used with refs>=14 + pyramid (regression in r607)
Loren Merritt [Wed, 20 Dec 2006 04:22:59 +0000 (04:22 +0000)]
wrong reference frames were used with refs>=14 + pyramid (regression in r607)

git-svn-id: svn://svn.videolan.org/x264/trunk@614 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoenable thread synchronization primitives on linux too
Loren Merritt [Tue, 19 Dec 2006 21:24:47 +0000 (21:24 +0000)]
enable thread synchronization primitives on linux too

git-svn-id: svn://svn.videolan.org/x264/trunk@613 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix a crash with x264_encoder_headers() + threads
Loren Merritt [Tue, 19 Dec 2006 09:35:45 +0000 (09:35 +0000)]
fix a crash with x264_encoder_headers() + threads

git-svn-id: svn://svn.videolan.org/x264/trunk@612 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agodon't skip autodection on configure --enable-pthread
Loren Merritt [Sat, 16 Dec 2006 00:46:37 +0000 (00:46 +0000)]
don't skip autodection on configure --enable-pthread

git-svn-id: svn://svn.videolan.org/x264/trunk@611 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agomore win32threads -> pthreads
Loren Merritt [Sat, 16 Dec 2006 00:32:38 +0000 (00:32 +0000)]
more win32threads -> pthreads

git-svn-id: svn://svn.videolan.org/x264/trunk@610 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocosmetics: rename list operators to be consistent with Perl, and move them to common/
Loren Merritt [Fri, 15 Dec 2006 23:08:57 +0000 (23:08 +0000)]
cosmetics: rename list operators to be consistent with Perl, and move them to common/

git-svn-id: svn://svn.videolan.org/x264/trunk@609 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agowin32: use pthreads instead of win32threads. for some reason, pthreads is much faster.
Loren Merritt [Fri, 15 Dec 2006 23:06:21 +0000 (23:06 +0000)]
win32: use pthreads instead of win32threads. for some reason, pthreads is much faster.

git-svn-id: svn://svn.videolan.org/x264/trunk@608 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoNew threading method:
Loren Merritt [Fri, 15 Dec 2006 23:03:36 +0000 (23:03 +0000)]
New threading method:
Encode multiple frames in prallel instead of dividing each frame into slices.
Improves speed, and reduces the bitrate penalty of threading.

Side effects:
It is no longer possible to re-encode a frame, so threaded scenecut detection
must run in the pre-me pass, which is faster but less precise.
It is now useful to use more threads than you have cpus. --threads=auto has
been updated to use cpus*1.5.
Minor changes to ratecontrol.

New options: --pre-scenecut, --mvrange-thread, --non-deterministic

git-svn-id: svn://svn.videolan.org/x264/trunk@607 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago * Do not assume anything about sizeof(cpu_set_t).
Sam Hocevar [Tue, 12 Dec 2006 02:17:44 +0000 (02:17 +0000)]
  * Do not assume anything about sizeof(cpu_set_t).

git-svn-id: svn://svn.videolan.org/x264/trunk@606 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago * Add support for kFreeBSD (FreeBSD kernel with GNU userland).
Sam Hocevar [Mon, 11 Dec 2006 16:01:49 +0000 (16:01 +0000)]
  * Add support for kFreeBSD (FreeBSD kernel with GNU userland).

git-svn-id: svn://svn.videolan.org/x264/trunk@605 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoAdd Altivec implementations of add8x8_idct8, add16x16_idct8, sa8d_8x8 and sa8d_16x16
Guillaume Poirier [Mon, 27 Nov 2006 21:40:21 +0000 (21:40 +0000)]
Add Altivec implementations of add8x8_idct8, add16x16_idct8, sa8d_8x8 and sa8d_16x16
Note: doesn't take advantage of some possible aligned memory accesses, so there's still room for improvement

git-svn-id: svn://svn.videolan.org/x264/trunk@604 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoForce alignment of the fake .rodata on MacIntel
Eric Petit [Sat, 25 Nov 2006 16:31:24 +0000 (16:31 +0000)]
Force alignment of the fake .rodata on MacIntel

git-svn-id: svn://svn.videolan.org/x264/trunk@603 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agodon't treat vbv_maxrate as a minrate too if it's higher than target average bitrate.
Loren Merritt [Thu, 23 Nov 2006 03:13:18 +0000 (03:13 +0000)]
don't treat vbv_maxrate as a minrate too if it's higher than target average bitrate.

git-svn-id: svn://svn.videolan.org/x264/trunk@602 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoMerges Guillaume Poirier's AltiVec changes:
Eric Petit [Sat, 18 Nov 2006 14:38:07 +0000 (14:38 +0000)]
Merges Guillaume Poirier's AltiVec changes:
 * Adds optimized quant and sub*dct8 routines
 * Faster sub*dct routines
~8% overall speed-up with default settings

git-svn-id: svn://svn.videolan.org/x264/trunk@601 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago10% faster deblock mmx functions. ported from ffmpeg.
Loren Merritt [Mon, 6 Nov 2006 22:49:41 +0000 (22:49 +0000)]
10% faster deblock mmx functions. ported from ffmpeg.

git-svn-id: svn://svn.videolan.org/x264/trunk@600 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocheckasm: ignore insignificant differences in floating-point ssim
Loren Merritt [Mon, 6 Nov 2006 22:38:42 +0000 (22:38 +0000)]
checkasm: ignore insignificant differences in floating-point ssim

git-svn-id: svn://svn.videolan.org/x264/trunk@599 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agodisplay final ratefactor in abr when a loose vbv is applied. (still disabled in true...
Loren Merritt [Mon, 30 Oct 2006 02:31:48 +0000 (02:31 +0000)]
display final ratefactor in abr when a loose vbv is applied. (still disabled in true cbr)

git-svn-id: svn://svn.videolan.org/x264/trunk@598 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix parsing of --deblock %d,%d (beta was ignored)
Loren Merritt [Mon, 30 Oct 2006 00:09:21 +0000 (00:09 +0000)]
fix parsing of --deblock %d,%d  (beta was ignored)

git-svn-id: svn://svn.videolan.org/x264/trunk@597 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocompute chroma_qp only once per mb
Loren Merritt [Sun, 29 Oct 2006 05:48:57 +0000 (05:48 +0000)]
compute chroma_qp only once per mb

git-svn-id: svn://svn.videolan.org/x264/trunk@596 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agord refinement of intra chroma direction (enabled in --subme 7)
Loren Merritt [Sun, 29 Oct 2006 01:17:33 +0000 (01:17 +0000)]
rd refinement of intra chroma direction (enabled in --subme 7)
patch by Alex Wright.

git-svn-id: svn://svn.videolan.org/x264/trunk@595 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix a crash in avc2avi
Loren Merritt [Wed, 18 Oct 2006 04:06:44 +0000 (04:06 +0000)]
fix a crash in avc2avi

git-svn-id: svn://svn.videolan.org/x264/trunk@594 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoskip deblocking and motion interpolation when using only I-frames
Loren Merritt [Sun, 15 Oct 2006 23:39:03 +0000 (23:39 +0000)]
skip deblocking and motion interpolation when using only I-frames

git-svn-id: svn://svn.videolan.org/x264/trunk@593 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocosmetics
Loren Merritt [Fri, 13 Oct 2006 23:50:57 +0000 (23:50 +0000)]
cosmetics

git-svn-id: svn://svn.videolan.org/x264/trunk@592 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoallow fractional values of crf
Loren Merritt [Fri, 13 Oct 2006 20:04:58 +0000 (20:04 +0000)]
allow fractional values of crf

git-svn-id: svn://svn.videolan.org/x264/trunk@591 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoprefetch pixels for motion compensation and deblocking.
Loren Merritt [Tue, 10 Oct 2006 21:26:31 +0000 (21:26 +0000)]
prefetch pixels for motion compensation and deblocking.

git-svn-id: svn://svn.videolan.org/x264/trunk@590 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agofix a crash on interlace + >8 reference frames
Loren Merritt [Tue, 10 Oct 2006 19:16:39 +0000 (19:16 +0000)]
fix a crash on interlace + >8 reference frames

git-svn-id: svn://svn.videolan.org/x264/trunk@589 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agono more decoder. it never worked anyway, and the presence of defunct code was confusi...
Loren Merritt [Tue, 10 Oct 2006 05:05:55 +0000 (05:05 +0000)]
no more decoder. it never worked anyway, and the presence of defunct code was confusing people.

git-svn-id: svn://svn.videolan.org/x264/trunk@588 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agocompute pskip_mv only once per macroblock, and store it
Loren Merritt [Mon, 9 Oct 2006 23:31:45 +0000 (23:31 +0000)]
compute pskip_mv only once per macroblock, and store it

git-svn-id: svn://svn.videolan.org/x264/trunk@587 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agoslightly faster chroma_mc_mmx
Loren Merritt [Mon, 9 Oct 2006 20:55:54 +0000 (20:55 +0000)]
slightly faster chroma_mc_mmx

git-svn-id: svn://svn.videolan.org/x264/trunk@586 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agomissing emms in plane_copy_mmx
Loren Merritt [Mon, 9 Oct 2006 17:44:47 +0000 (17:44 +0000)]
missing emms in plane_copy_mmx

git-svn-id: svn://svn.videolan.org/x264/trunk@585 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agomerge center_filter_mmx with horizontal_filter_mmx
Loren Merritt [Fri, 6 Oct 2006 23:25:41 +0000 (23:25 +0000)]
merge center_filter_mmx with horizontal_filter_mmx

git-svn-id: svn://svn.videolan.org/x264/trunk@584 df754926-b1dd-0310-bc7b-ec298dee348c

17 years ago1.5x faster center_filter_mmx (amd64)
Loren Merritt [Fri, 6 Oct 2006 05:43:53 +0000 (05:43 +0000)]
1.5x faster center_filter_mmx (amd64)

git-svn-id: svn://svn.videolan.org/x264/trunk@583 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agommx/prefetch implementation of plane_copy
Loren Merritt [Fri, 6 Oct 2006 00:02:59 +0000 (00:02 +0000)]
mmx/prefetch implementation of plane_copy

git-svn-id: svn://svn.videolan.org/x264/trunk@582 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agono more vfw
Loren Merritt [Thu, 5 Oct 2006 08:15:55 +0000 (08:15 +0000)]
no more vfw

git-svn-id: svn://svn.videolan.org/x264/trunk@581 df754926-b1dd-0310-bc7b-ec298dee348c

17 years agogtk fixes:
Loren Merritt [Thu, 5 Oct 2006 07:44:22 +0000 (07:44 +0000)]
gtk fixes:
in Makefile
- fix datadir for mingw users
- remove the shared lib during the clean rule
- use $(ENCODE_BIN) instead of x264_gtk_encode
- add some $(DESTDIR) and create some directories when necessary
- remove -lintl
statfile_length -> statsfile_length
fix the "sensitivity" of the widget of update_statfile
the logo is now handled correctly on windows
added: beginning of multipass support
patch by Vincent Torri.

git-svn-id: svn://svn.videolan.org/x264/trunk@580 df754926-b1dd-0310-bc7b-ec298dee348c