x264 - OpenSource AVC/H.264 Video Codec

Es ist soweit :cool:

Zitat

commit dbc5ef040b1f8a83e7491dc8a2fc8943b1e20c07 r1016
Author: Loren Merritt <pengvado@akuvian.org>
Date: Thu Oct 30 00:47:09 2008 -0700

Fix bug in hadamard_ac SSE assembly
Some extreme inputs could cause overflows.

commit a0a1bfac7f4a09159f6ef2bf13fb69548b6c5a02 r1015
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Oct 28 20:35:15 2008 -0700

Full sub8x8 RD mode decision
Small speed penalty with p4x4 enabled, but significant quality gain at subme >= 6
As before, gain is proportional to the amount of p4x4 actually useful in a given input at the given bitrate.

commit aa40e41abae051191117ae670cadd9cd50f66b6f r1014
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Oct 25 01:50:08 2008 -0700

Optimize CABAC bit cost calculation
Speed up cabac mvd and add new precalculated transition/entropy table.
Add "noup" function for cabac operations to not update the state table when it isn't necessary.
1-3% faster macroblock_size_cabac.
Cosmetics

commit d7df1a477b5e0e851d206e8c25da0b275ae0b7cc r1013
Author: Anders Ossowicki <arkanoid@exherbo.org>
Date: Thu Oct 23 22:36:11 2008 -0700

Replace "git-command" with "git command" in version.sh for git 1.6 support

commit 990274cd5fd276bb26ac0fa13fc9bc1cbcf7acbc r1012
Author: Loren Merritt <pengvado@akuvian.org>
Date: Thu Oct 23 13:45:04 2008 -0700

Add assembly version of CAVLC 8x8dct interleave
Faster CAVLC encoding and RDO with 8x8dct

commit 86a0fe50c6e369d6dacac5b992febb4bd09de85d r1011
Author: Alexander Strange <astrange@ithinksw.com>
Date: Wed Oct 22 15:55:30 2008 -0700

Add support for psy-rd/trellis to encoder_reconfig

commit 4c78f091e625e87b8f82c567af81969c2fd3e671 r1010
Author: Alexander Strange <astrange@ithinksw.com>
Date: Wed Oct 22 15:00:43 2008 -0700

Fix Darwin speed regression

commit 8ca555ef691e20d4eb429b45e178f4c0108b607d r1009
Author: Gabriel Bouvigne <bouvigne@mp3-tech.org>
Date: Wed Oct 22 14:48:47 2008 -0700

Further improve prediction of bitrate and VBV in threaded mode

commit 8fa0bf3d56fa1b02a44f8ee0f673ea998294bd7e r1008
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Oct 22 13:37:09 2008 -0700

Sub-8x8 Qpel-RD in P-frames
Improves quality when using p8x4/p4x8/p4x4 subpartitions
Benefit is proportional to how many sub-8x8 partitions are used; helps most at high bitrates and low resolutions.

commit d0add77f5f084253202747266f85daa65f7fc9cc r1007
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Oct 22 02:20:06 2008 -0700

Faster qpel-RD
3-4% faster qpel-RD; avoid re-checking bmv/pmv during the hex search.

commit f451563f93fad8972c0e9b788b30799e777e913a r1006
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Oct 22 00:37:00 2008 -0700

Some minor optimizations in RD refinement
Don't write b subpartition in CABAC RDO
Calculate nonzero count in i4x4 CAVLC RDO

commit 84ede33bec64332cc4bc5da1106c53f3cffa919b r1005
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Oct 21 20:17:18 2008 -0700

Faster deblocking when p4x4 isn't used
Most of the MV checks can be skipped, resulting in faster strength calculation

commit ce0b11099e5fa920b8d1bc39389ae9373f921358 r1004
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Oct 21 19:38:21 2008 -0700

Print profile and level information upon starting encode
Previously level was only printed as part of autodetect, and only in verbose mode.

commit 296b39dd863ea90a12d8a52848d1135e387a28f3 r1003
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Oct 21 17:10:46 2008 -0700

Fix possible crash in trellis at very low QPs

commit f5da8110606c1bdb8f3a194f11574db28855415e r1002
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Oct 21 14:59:07 2008 -0700

Add assembly versions of decimate_score
3-7x faster decimation, 1-3% faster overall

commit 6d9ef8ad39b0bfa5df0c1305e91ae932aad4997e r1001
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Oct 18 03:40:59 2008 -0700

Fix typo in subme8/9 lossless qpel-RD
Slightly improves compression.

commit 79194caffdc216e338674d88e50adca2f4ea8fa2 r1000
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Oct 16 03:17:53 2008 -0700

Extend trellis to support luma/chroma DC and chroma AC
Small speed loss in trellis 1, slightly larger in trellis 2, but significant quality improvement.

Alles anzeigen

Zitat von LoRd_MuldeR

x264 r1016
http://www.mediafire.com/file/dkn5mzndy…32-fprofiled.7z

MinGW/GCC 4.3.2-tdm-1, yasm 0.7.1.2093, march=pentium2, fprofiled

Zitat von LoRd_MuldeR

libx264 SVN-r1016 + Gruntster's Alignment-Fix:
http://www.mediafire.com/file/lwgmqvyzj…gn-fprofiled.7z

Compiled with MinGW GCC 4.3.2-tdm-1 (fprofiled). Tested with Avidemux 2.4.3 r4494.

Tja, die Revision 1000 Party fand wohl doch nicht in "Pengvado's Villa" statt... sondern im DivX Hauptquartier

http://forum.doom9.org/showthread.php?t=142482

Mal wieder ein Update auf den aktuellen Stand:

Zitat

commit 2652abeae5445ffefded5ee7d0853300d0973b37 r1024
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sun Nov 9 17:39:21 2008 -0800

Faster chroma encoding
9-12% faster chroma encode.
Move all functions for handling chroma DC that don't have assembly versions to macroblock.c and inline them, along with a few other tweaks.

commit f13d4637fe9b2f10b8c103500ac9293bfca3ad1f r1023
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sun Nov 9 17:34:31 2008 -0800

Various cosmetics and minor fixes
Disable hadamard_ac sse2/ssse3 under stack_mod4
Fix one MSVC compilation warning
Fix compilation in debug mode in certain cases on x64
Remove eval.c from MSVC project
Fix crash when VBV is used in CQP mode
Patches by MasterNobody

commit 7cdaf638c3777f2b38fb60181dde7ed4de614cc1 r1022
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Nov 8 20:16:17 2008 -0800

Faster b-adapt + adaptive quantization
Factor out pow to be only called once per macroblock. Speeds up b-adapt, especially b-adapt 2, considerably.
Speed boost is as high as 24% with b-adapt 2 + b-frames 16.

commit 852579be365549db3ccc1c2906c9a1d2f4a92ac9 r1021
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Nov 7 11:39:43 2008 -0800

Faster CABAC residual encoding
6% faster block_residual_write_cabac in RD mode.

commit 418cace8646a6f546a9026da47f79fad7285f577 r1020
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Nov 5 19:51:59 2008 -0800

Fix potential crash in the case that the input statsfile is too short
Also resolve various other potential weirdness (such as multiple copies of the same error message in threaded mode).

commit a5ac6a5b8688915553fe6fccee09f1272f3788ac r1019
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Nov 5 03:11:45 2008 -0800

Initial Nehalem CPU optimizations
movaps/movups are no longer equivalent to their integer equivalents on the Nehalem, so that substitution is removed.
Nehalem has a much lower cacheline split penalty than previous Intel CPUs, so cacheline workarounds are no longer necessary.
Thanks to Intel for providing Avail Media with the pre-release Nehalem CPU needed to prepare these (and other not-yet-committed) optimizations.
Overall speed improvement with Nehalem vs Penryn at the same clock speed is around 40%.

commit 41b8069cb74fa3bc905618225be07ee8d35bbc79 r1018
Author: Gabriel Bouvigne <bouvigne@mp3-tech.org>
Date: Tue Nov 4 09:56:03 2008 -0800

Fix potential infinite loop in VBV under GCC 4.2

commit e4c4568d4f0f234e942b4855391aea7224c41eb6 r1017
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Nov 3 22:59:49 2008 -0800

Encoder_reconfig: esa/tesa can only be enabled if they were on to begin with
Bug report by kemuri-_9.

Alles anzeigen

Zitat

libx264 SVN-r1024 + Gruntster's Alignment-Fix:
http://www.mediafire.com/file/uhgvuylfv…gn-fprofiled.7z

Compiled with MinGW GCC 4.3.2-tdm-1 (fprofiled). Tested with Avidemux 2.4.3 r4494.

Zitat

2^10 GET!

x264.1024.modified.exe - Alternate Download

Patches used:

x264_hrd_pulldown.09_interlace.diff

gcc 3.4.5 fprofiled build with -march=pentium2.

Wird wohl Zeit den in die Jahre gekommenen Q6600 durch einen Core i7 (Nehalem) zu ersetzen

Revision 1029:

Zitat

commit cb3c213850320fb0c1b17ae8bbbbf5d687e43961 r1029
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Nov 20 20:11:14 2008 -0800

A few tweaks to decimate asm
A little bit faster on both 32-bit and 64-bit

commit 83baa7fdd2edf3e2f9522fc8b79e0826bcf190fc r1028
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Nov 12 16:50:31 2008 -0800

Nehalem optimization part 2: SSE2 width-8 SAD
Helps a bit on Phenom as well
~25% faster width8 multiSAD on Nehalem

commit aa14719bf2b78f8fd3da7bbabb0faf142313dae1 r1027
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Nov 10 23:34:02 2008 -0800

Add subme=0 (fullpel motion estimation only)
Only for experimental purposes and ultra-fast encoding. Probably not a good idea for firstpass.

commit 745a48beddc58e2ef121326e9156d3d42590a4b5 r1026
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Nov 10 15:34:48 2008 -0800

Fix minor memory leak in r1022

commit fdb6114d1b456e1438374671ec42d1d77cdd05f8 r1025
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Nov 10 15:32:06 2008 -0800

r1024 borked checkasm
Remove idct/dct2x2 from checkasm as they are no longer in dctf

Alles anzeigen

Und passend dazu:

Zitat von skystrife

x264.1029.modified.exe - Alternate Download

Patches used:

x264_hrd_pulldown.09_interlace.diff
x264_custom_strtok_r.r1024.diff

gcc 3.4.5 fprofiled build with -march=pentium2.

Zitat von LoRd_MuldeR

libx264 SVN-r1029 + Gruntster's Alignment-Fix:
http://www.mediafire.com/file/ijymygioy…gn-fprofiled.7z

Compiled with MinGW GCC 4.3.2-tdm-1 (fprofiled). Tested with Avidemux 2.4.3 r4494.

Zitat von LoRd_MuldeR

Wird wohl Zeit den in die Jahre gekommenen Q6600 durch einen Core i7 (Nehalem) zu ersetzen

Dann doch wohl lieber den hier.:D

Ich sollte das noch erwähnen, da es auch einige größere Einflüsse hat. Sharktooth wird seine Presets ebenfalls anpassen.

Zitat

commit a0a1bfac7f4a09159f6ef2bf13fb69548b6c5a02 r1015
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Oct 28 20:35:15 2008 -0700

Full sub8x8 RD mode decision
Small speed penalty with p4x4 enabled, but significant quality gain at subme >= 6
As before, gain is proportional to the amount of p4x4 actually useful in a given input at the given bitrate.

Jetzt kann man p4x4 wieder problemlos aktivieren. Erste Tests bringen eine 1 %ig bessere Komprimierung. Nur bei der Leistung bin ich nicht so sicher.

Zitat

Erste Tests bringen eine 1 %ig bessere Komprimierung. Nur bei der Leistung bin ich nicht so sicher.

-> Höh?

Mit crf komprimiert. Vorher wars so, dass mit p4x4 die Komprimierung schlechter war (größere Datei). Und mit Leistungs meinte ich eigentlich die Geschwindigkeit. Meine eigene Erfahrung in der Richtung.

Zitat von Deinorius

Mit crf komprimiert. Vorher wars so, dass mit p4x4 die Komprimierung schlechter war (größere Datei). Und mit Leistungs meinte ich eigentlich die Geschwindigkeit. Meine eigene Erfahrung in der Richtung.

Größenvergleiche bei CRF sind sinnlos, denn CRF liefert ja nur "ungefähr" die gleiche Qualität. Und wenn man andere Optionen ändert, kann sich die Bedeutung der CRF Werte ohnehin verändern. Wenn also die Datei bei gleichem CRF Wert größer wird, wie willst du dann feststellen, ob die größere Datei den Qualitätsgewinn wert war oder nicht? Oder umgekehrt: Wenn die Datei kleiner wird, wie willst du feststellen, ob die Qualität nicht stärker gelitten hat, als es die Platzersparnis wert ist? Das ist praktisch nicht möglich! Gegen die Wand bin ich anfänglich übrigens auch gelaufen. Aber um einen sinnvollen Vergleich zu machen, musst du zwei 2-Pass encodes ausführen und dann die beiden Dateien (identischer Größe) visuell vergleichen. Eine andere Möglichkeit ist mir nicht bekannt...

Dazu auch:

Zitat von akupenguin

With CRF, various settings affect bitrate or quality or some mixture thereof; if you see bitrate was reduced, you can't even be sure whether that was an improvement without comparing quality too. [...] Repeat after me: CRF is quality-based, but not constant quality. It's good enough that you can generally get away with picking one CRF value that's transparent to you, but it's not constant enough to evaluate other options.

Ist mir eigentlich schon bekannt, aber ob der Unterschied überhaupt sichtbar ist... naja, aber ich habs ja auch nicht probiert. :\

Update:

Zitat

commit 5df2a7162aecda66b3da8dde501971389b1bbd44 r1042
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Nov 28 14:32:11 2008 -0800

10L in r1041

commit 12724e638e8d28ce97bcb9c77d2bb7336b087af3 r1041
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Nov 27 19:37:56 2008 -0800

Significantly faster CABAC and CAVLC residual coding and bit cost calculation
Early-terminate in residual writing using stored nnz counts
To allow the above, store nnz counts for luma and chroma DC
Add assembly functions to find the last nonzero coefficient in a block
Overall ~1.9% faster at subme9+8x8dct+qp25 with CAVLC, ~0.7% faster with CABAC
Note this changes output slightly with CABAC RDO because it requires always storing correct nnz values during RDO, which wasn't done before in cases it wasn't useful.
CAVLC output should be equivalent.

commit 1591275a92faa3d63186e6de1e9022956113bc1d r1040
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Nov 26 23:42:55 2008 -0800

dequant_4x4_dc assembly
About 3.5x faster DC dequant on Conroe

commit 2338e1301aded556be8b85c6c3b4050e562ed862 r1039
Author: Loren Merritt <pengvado@akuvian.org>
Date: Thu Nov 27 02:37:46 2008 +0000

fix an overflow in dct4x4dc_mmx
(unlikely to have occurred in any real video)

commit a7fd9f5da062de323ae89f9a71ede03bfd6ddb6a r1038
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Nov 25 16:30:39 2008 -0800

Remove nasm support
Nasm won't correctly parse the SSE4 code introduced a few revisions ago, so we're removing support.
Users should upgrade to yasm 0.6.1 or later.

commit 0fe9b85f393e00b79c37d8c81aeae2a2f3d41290 r1037
Author: BugMaster <BugMaster@narod.ru>
Date: Tue Nov 25 15:11:24 2008 -0800

Fix rare warning messages in ratecontrol due to r1020

commit adffb7fa3d3db6d2fa2ad97e0c950afec8889ea5 r1036
Author: BugMaster <BugMaster@narod.ru>
Date: Tue Nov 25 15:10:43 2008 -0800

Fix MSVC compilation and clean up MSVC build file
Remove Release64 which never worked anyways.

commit e1013e8152254614696bbc9d92959bc9705d98b1 r1035
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Nov 25 01:04:26 2008 -0800

Faster width4 SSD+SATD, SSE4 optimizations
Do satd 4x8 by transposing the two blocks' positions and running satd 8x4.
Use pinsrd (SSE4) for faster width4 SSD
Globally replace movlhps with punpcklqdq (it seems to be faster on Conroe)
Move mask_misalign declaration to cpu.h to avoid warning in encoder.c.
These optimizations help on Nehalem, Phenom, and Penryn CPUs.

commit cab23dd7e6bf4ed501f36e1e6f64d4902d0489c9 r1034
Author: Guillaume Poirier <gpoirier@dusk.caps-entreprise.com>
Date: Tue Nov 25 17:27:27 2008 +0100

fix indentation, whitespace cleanup, more consistent indentation of macro backslashes

commit 4bf4109aee5b602f8a124b434e18f93ef539bbe6 r1033
Author: David Wolstencroft <wolstencroft@alum.rpi.edu>
Date: Sat Nov 22 17:54:38 2008 +0100

Change some macros to be more sensitive to memory alignment, thus avoiding
useless loads/stores and calculations of permutation vectors.
Affected functions are all of mc_luma, mc_chroma, 'get_ref', SATD, SA8D and deblock.
Gains globally vary from ~5% - 15% on a depending on settings running on a 1.42 ghz G4.


commit 8215f79c7e19747b3206bcd6245be6cc6e668145 r1032
Author: Loren Merritt <pengvado@akuvian.org>
Date: Fri Nov 7 05:31:24 2008 +0000

refactor satd. 20KB smaller binary.
refactor sa8d. slightly faster.
more checkasm for hadamard.

commit 3a028c8e50238b7799175bd5a172e5517b4baf8d r1031
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Nov 24 21:56:24 2008 -0800

Fix crash with threads and SSEMisalign on Phenom
Misalign mask needed to be set separately for each encoding thread.

commit f9dba8bb274dffb19394db20912823464efcb8e1 r1030
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Nov 21 03:39:11 2008 -0800

Phenom CPU optimizations
Faster hpel_filter by using unaligned loads instead of emulated PALIGNR
Faster hpel_filter on 64-bit by using the 32-bit version (the cost of emulated PALIGNR is high enough that the savings from caching intermediate values is not worth it).
Add support for misaligned_mask on Phenom: ~2% faster hpel_filter, ~4% faster width16 multisad, 7% faster width20 get_ref.
Replace width12 mmx with width16 sse on Phenom and Nehalem: 32% faster width12 get_ref on Phenom.
Merge cpu-32.asm and cpu-64.asm
Thanks to Easy123 for contributing a Phenom box for a weekend so I could write these optimizations.

Alles anzeigen

Zitat

libx264 SVN-r1042 + Gruntster's Alignment-Fix:
http://www.mediafire.com/file/3nzzijdty…gn-fprofiled.7z

Compiled with MinGW GCC 4.3.2-tdm-1 (fprofiled). Tested with Avidemux 2.4.3 r4494.

Zitat

x264.1042M.exe - Alternate Download

Patches used:

x264_hrd_pulldown.09_interlace.diff
x264_win_zone_parse_fix_03.diff
x264_mingw_aligned.diff

gcc 3.4.5 fprofiled build with -march=pentium2.

Achtung: r1041 is kaputt, r1042 geht wieder

Der x264 "lossless" Modus wird nun von libavcodec unterstützt :cool:

Zitat von Dark Shikari

x264's new lossless mode is now supported in libavcodec.

CoreAVC is no longer necessary to play such streams :). And once everyone updates their ffdshow/mplayer we can stop getting the weekly post in #x264 about lossless output being corrupt...

Und mal wieder ein Update auf den aktuellen Stand der Dinge:

Zitat

commit 0e1a92c56f8c89735e2d5044f89832bdc22b6e50 r1065
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Dec 29 22:00:02 2008 -0500

Much faster CAVLC RDO and bitstream writing
Pure asm version of level/run coding. Over 2x faster than C.
Up to 40% faster CAVLC RDO. Overall benefit up to ~7.5% with RDO or ~5% with fast encoding settings.

commit 648e132f7135c7e18625198e3ffe2c6c7d824df6 r1064
Author: Loren Merritt <pengvado@akuvian.org>
Date: Mon Dec 29 21:52:25 2008 -0500

Cosmetics: cleaner syntax for defining temporary registers in asm
Globally define t#[qdwb], so that only t# needs to be locally defined when reorganizing registers

commit 84a1ca6ce70fe7bad4922ddc5a72c2e9cd73703b r1063
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Dec 27 21:36:14 2008 -0500

Much faster CABAC RDO
Since RDO doesn't care about what order bit costs are calculated, merge sigmap and level coding into the same loop in RDO.
This is bit-exact for 4x4dct but slightly incorrect for 8x8dct due to the sigmap containing duplicated contexts.
However, the PSNR penalty of this is extremely small (~0.001db).
Speed benefit is about 15% in 4x4dct and 30% in 8x8dct residual bit cost calculation at QP20.
Overall encoding speed benefit is up to 5%, depending on encoding settings.
Also remove an old unnecessary CABAC table that hasn't been used for years.

commit 40ad1b6d411d1e7d0788d29627804a19977bb6ee r1062
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Dec 26 07:35:49 2008 -0500

VLC table optimizations
Slightly reorganize VLC tables for ~2% faster block_residual_write_cavlc.
Also a small optimization in p8x8 CAVLC.

commit 839cd8cf33492213f7878bf14c26b387b6599abd r1061
Author: Loren Merritt <pengvado@akuvian.org>
Date: Wed Dec 24 22:58:17 2008 -0500

Fix crash in --me esa/tesa introduced in r1058
Also suppress the last mingw warning message

commit 42070dff1bc3019a6f56773fce3dd6e328e3a61b r1060
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Dec 23 22:33:28 2008 -0500

Optimize variance asm + minor changes
Remove SAD argument from var, not needed anymore.
Speed up var asm a bit by eliminating psadbw and instead HADDWing at end.
Eliminate all remaining warnings on gcc 3.4 on cygwin
Port another minor optimization from lavc (pskip)

commit 265d61bcdc5e6ce6688e377af9f6c0136724ed59 r1059
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Dec 23 18:31:48 2008 -0500

Minor CABAC cleanups and related optimizations
Merge the two list tables to allow cleaner MC/CABAC/CAVLC code
Remove lots of unnecessary {s
Port some very minor opts from lavc

commit a4ec1020efb1a2a6757f8f891d78c2dd9344bb91 r1058
Author: Loren Merritt <pengvado@akuvian.org>
Date: Thu Dec 11 19:47:17 2008 +0000

faster ESA init
reduce memory if using ESA and not p4x4

commit 5f8a1490eb0bc2a934c34bc8307bfdc1ade6a92d r1057
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Dec 15 23:02:49 2008 -0800

More macroblock_cache optimizations
Patch partially by Loren Merritt

commit e59ee249829049de338bebc3a2a00f9e471b40f3 r1056
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Dec 15 13:15:29 2008 -0800

Faster macroblock_cache_rect
Explicit loop unrolling

commit 2b8d6a6f957be623186ea2a20bcb13c3637440b8 r1055
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sun Dec 14 18:30:51 2008 -0800

Optimizations in predict_mv_direct
Add some early terminations and minor optimizations
This change may also fix the extremely rare direct+threading MV bug.

commit 918ff3c6a33e170655b61af85f8955ec5590fff8 r1054
Author: David Wolstencroft <wolstencroft@alum.rpi.edu>
Date: Sun Dec 14 10:47:28 2008 +0000

Fix visual corruption when picture width was not mod 32.
The previous Altivec implemention of mc_chroma assumed that i_src_stride was always mod 16.

commit 9089d217078450fad075b5eb61f372572d094a5f r1053
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Mon Dec 8 21:11:45 2008 +0100

Add support for FSF GCC version >= 4.3 on OSX.
So far, only Apple GCC version was supported.

commit ad2c84f76f9fbb4f360caeb87df824beab023bbf r1052
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Dec 11 17:31:52 2008 -0800

More accurate refcost for p8x8 CAVLC
Slightly better quality, especially in non-RD mode, with CAVLC.

commit 549cc55b50df76d5167c0ace75c62595feb753ca r1051
Author: Loren Merritt <pengvado@akuvian.org>
Date: Wed Dec 10 20:54:17 2008 -0800

use lookup tables instead of actual exp/pow for AQ
Significant speed boost, especially on CPUs with atrociously slow floating point units (e.g. Pentium 4 saves 800 clocks per MB with this change).
Add x264_clz function as part of the LUT system: this may be useful later.
Note this changes output somewhat as the numbers from the lookup table are not exact.

commit 877d22e071f1b73fb33c628cc273e2819d10c3de r1050
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Dec 10 20:53:13 2008 -0800

Suppress saveptr warnings on Windows GCC

commit 0fdd0403cc9cc95637e287ddd1b257d6b65b7ddb r1049
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Dec 10 20:52:06 2008 -0800

More small speed tweaks to macroblock.c

commit 77028cd3671de855affb02ffefe6bbd99ac7816e r1048
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Dec 8 13:44:23 2008 -0800

Much faster CAVLC residual coding
Use a VLC table for common levelcodes instead of constructing them on-the-spot
Branchless version of i_trailing calculation (2x faster on Nehalem)
Completely remove array_non_zero_count and instead use the count calculated in level/run coding. Note: this slightly changes output with subme > 7 due to different nonzero counts being stored during qpel RD.

commit f773bf06256a467f6b18418d97ce2c7ddbe5728c r1047
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Fri Dec 5 22:26:55 2008 +0100

fix compilation with GCC-4.3+

commit 71d34b4eb454027cd742e6d96e2a70cce8cd163c r1046
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Nov 29 23:13:58 2008 -0800

High Profile allows 25% higher maxbitrate/cpb
Correct level detection to take this into account.

commit cc40e308c919010ef9ae6ff376cda56a58a79f3a r1045
Author: BugMaster <BugMaster@narod.ru>
Date: Sat Nov 29 14:04:29 2008 -0800

s/nasm/yasm in VS project file

commit 6c3e0258776c1117929aec7c12d51b88c214467c r1044
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Nov 29 04:49:18 2008 -0800

Cosmetic: update various file headers.

commit 2f031b0e31b799072383792358eee376baeb2ba7 r1043
Author: Loren Merritt <pengvado@akuvian.org>
Date: Sat Nov 29 11:54:02 2008 +0000

add date and compiler to `x264 --version`

Alles anzeigen

Zitat

libx264 SVN-r1065 + Gruntster's Alignment-Fix:
http://www.mediafire.com/file/oziknwyjz…gn-fprofiled.7z

Compiled with MinGW GCC 4.3.2-tdm-2 (fprofiled). Tested with Avidemux 2.4.3 r4502.

Zitat

x264.1065M.exe - Alternate Download

Patches used:

x264_hrd_pulldown.09_interlace.diff
x264_win_zone_parse_fix.04.diff

gcc 3.4.5 fprofiled build with -march=pentium2.

-----------------------------------------------

x264.1065M.x64.exe - Alternate Download

Patches used:
x264_hrd_pulldown.09_interlace.diff
x264_win_zone_parse_fix_04.diff
x264_win64_support.01.r1065.diff <-- fixed patch to work with latest commits, but I know nothing as far as asm code goes, so it's highly possible I completely screwed something up. Use at your own risk.

gcc 4.4.0 fprofiled build.

Alles anzeigen

Und mal wieder Zeit fürn Update:

Zitat

commit 9d6cc8e28319b3935698d52a6711414435444029 r1074
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Jan 8 15:07:16 2009 -0500

Fix regression in r1066
With some combinations of video width and other settings, the scratch buffer was slightly too small.
This caused heap corruption on some systems.

commit 6a4a9beae060d69bbeaeb8c1c3056fb6ae6765f6 r1073
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Jan 6 16:55:44 2009 -0500

Disable B-frames in lossless mode
They hurt compression anyways, and direct auto was bugged with lossless.

commit f586ba52b87ebb6b1d7689603680ef7d3219e09a r1072
Author: Brad Smith <brad@comstyle.com>
Date: Mon Jan 5 22:53:11 2009 +0000

Factorize in ppccommon.h the conditional inclusion of altivec.h on Linux systems.

commit 87b6d55ebda8a11186f7b09b5866b05a4584d13d r1071
Author: Brad Smith <brad@comstyle.com>
Date: Mon Jan 5 15:58:32 2009 -0500

Disable __builtin_clz() intrinsic on gcc versions prior to 3.4.
The function did not exist before that version.

commit 6f7c9be698848e8d9fd116b728af7d718ea43a2f r1070
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Jan 1 21:44:00 2009 -0500

Small tweaks to coeff asm
Factor out a few redundant pxors
Related cosmetics

commit ed32ad20d6914d1781c7711574f10a8df49e3e20 r1069
Author: Steven Walters <kemuri9@gmail.com>
Date: Tue Dec 30 22:20:37 2008 -0500

Use the correct strtok under MSVC
Also change one malloc -> x264_malloc

commit 390d26ad2ca72b420448c36a747d3ee49b79e75b r1068
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Dec 30 22:14:45 2008 -0500

Add stack alignment for lookahead functions
Should allow libx264 to be called from non-gcc-compiled applications without adding force_align_arg_pointer.

commit 00cef64dd3fff5d4b5b9b0e63314c11bfb7d33e0 r1067
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Dec 30 20:47:45 2008 -0500

Add support for SSE4a (Phenom) LZCNT instruction
Significantly speeds up coeff_last and coeff_level_run on Phenom CPUs for faster CAVLC and CABAC.
Also a small tweak to coeff_level_run asm.

commit 045ae4045a1827555b3eaab4fbf3c9809e98c58f r1066
Author: Steven Walters <kemuri9@gmail.com>
Date: Mon Dec 29 05:14:26 2008 +0000

factor mallocs out of hpel, ssim, and esa.
there should now be no memory allocation outside of init-time.

Alles anzeigen

Zitat

x264.1074M.exe - Alternate Download

Patches used:

x264_hrd_pulldown.09_interlace.diff
x264_win_zone_parse_fix_05.diff

gcc 3.4.5 fprofiled build with -march=pentium2.
-----------------------------------------------
x264.1074M.x64.exe - Alternate Download

Patches used:
x264_hrd_pulldown.09_interlace.diff
x264_win_zone_parse_fix_05.diff
x264_win64_support.01.r1065.diff

gcc 4.4.0 fprofiled build.

Alles anzeigen

Zitat

libx264 SVN-r1074:
http://www.mediafire.com/file/nm1nwjwyz…32-fprofiled.7z

Compiled with MinGW GCC 4.3.2-tdm-2, march=pentium2, fprofiled. Tested with Avidemux 2.4.3 r4508.

Weitere Updates:

Zitat

commit 1df50b9287c83d5443d19482345b6842b78081c3 r1093
Author: Loren Merritt <pengvado@akuvian.org>
Date: Tue Jan 27 05:42:51 2009 +0000

remove $ECHON kludge, which broke on SunOS. bring back `gcc -MT`.
remove auto-reconfigure on svn update, which has done nothing since we stopped using svn.
fix $AS on sparc (was disabled by mmx check).
fix --extra-asflags (was ignored).
mark bash scripts as bash, not sh

patch partly by Greg Robinson and Jugdish.

commit 60f4cd8936af4997cfbd9650ea27152df00c5669 r1092
Author: Loren Merritt <pengvado@akuvian.org>
Date: Mon Jan 26 14:28:48 2009 +0000

1.6x faster satd_c (and sa8d and hadamard_ac) with pseudo-simd.
60KB smaller binary.

commit 71c5a8dca6e5f7bf2330d028989eeeab27701151 r1091
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Jan 27 23:27:56 2009 -0800

Hack around a potential failure point in VBV
pred_b_from_p can become absurdly large in static scenes, leading to rare collapses of quality with VBV+B-frames+threads.
This isn't a final fix, but should resolve the problem in most cases in the meantime.

commit 8e2c4b76dd49c44cedf46343e46b292d5a0ca39e r1090
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Jan 26 23:43:25 2009 -0800

Much faster chroma encoding and other opts
~15% faster chroma encode by reorganizing CBP calculation and adding special-case idct_dc function, since most coded chroma blocks are DC-only.
Small optimization in cache_save (skip_bp)
Fix array_non_zero to not violate strict aliasing (should eliminate miscompilation issues in the future)
Add in automatic substitutions for some asm instructions that have an equivalent smaller representation.

commit 355c445f222eae4f953f1be9f8e1d44a2a07d639 r1089
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Mon Jan 26 06:28:23 2009 -0800

add AltiVec implementation of x264_mc_copy_w16_aligned

commit 71ac0a34bc0460bf67da68f300e4150bc50d9aae r1088
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Fri Jan 23 13:53:06 2009 -0800

add AltiVec implementation of x264_pixel_var_16x16 and x264_pixel_var_8x8

commit 1959672cf6f18da888940261916dbf81248e0598 r1087
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Fri Jan 23 01:11:20 2009 -0800

add AltiVec 16 <-> 32 bits conversions macros

commit 39a279613d10fa4dbe608f1a2af1eb86686033af r1086
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Mon Jan 19 21:29:27 2009 +0100

Replace 16x16=>32 mul + pack + add by a simple 16x16=>16 multiply-add.
Suggested by Loren.

commit 5fb6417309febda6f73a98494cb935379129e15d r1085
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Jan 19 15:17:53 2009 -0800

Eliminate support for direct_8x8_inference=0
The benefit in the most extreme contrived situation was at most 0.001db PSNR, at the cost of slower decoding.
As this option was basically useless, it was a waste of code and prevented some other useful optimizations.
Remove some unused mc code related to sub-8x8 partitions.
Small deblocking speedup when p4x4 is used.
Also remove unused x264_nal_decode prototype from x264.h.

commit a48d1d0a2ad590d041b79bb152ed47d00451ba8d r1084
Author: Brad <brad@comstyle.com>
Date: Mon Jan 19 05:14:53 2009 -0800

Add AltiVec and CPU numbers detection on OpenBSD.

commit 09e76c903d3419619ed326a4dd114369a55bdd6e r1083
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Sun Jan 18 22:44:14 2009 +0100

Add AltiVec implementation of predict_8x8c_p. 2.6x faster than scalar C.

commit bde164d50f1936b28a9cd66f8be8d9995cc5a01b r1082
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Jan 17 15:16:37 2009 -0500

Warn if direct auto wasn't set on the first pass
And, if it wasn't, run direct auto as if it was the first pass, rather than simply forcing temporal direct mode on all frames.
Also a small tweak to coeff_level_run asm.

commit 201f7ad8ad50ff460f79cb44e0bee6aebbf039ca r1081
Author: Brad <brad@comstyle.com>
Date: Sat Jan 17 12:52:28 2009 +0000

Changes the PowerPC ppccommon.h header so it no longer checks for a particular
OS such as Linux but instead looks for HAVE_ALTIVEC_H being set.
Fixes all *BSD/PowerPC builds.

commit 56e91836a12f1b119fb4aae43182f2fc012f1eca r1080
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Wed Jan 14 21:56:31 2009 +0100

update x264_hpel_filter_altivec's prototype to match the one of the C version.
It changed in commit 045ae4045a1827555b3eaab4fbf3c9809e98c58f (factorization of mallocs)
(NB: Altivec implementation wasn't allocating and writing to any scratch memory.)

commit d2b6423db8310e5238fd7c0e517b3344578cc08a r1079
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Wed Jan 14 21:49:42 2009 +0100

rename vector+array unions to closer match the vector typedefs names.

commit 264e447cc4ce267d7e4d078b080716093a78a2c8 r1078
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Wed Jan 14 21:13:58 2009 +0100

Add Altivec implementation of all the remaining 16x16 predict routines.

commit e46f64824a718ca146724d6cfa104aaba16eb169 r1077
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Jan 13 21:11:50 2009 -0500

Cache ref costs and use more accurate MV costs
New MV costs should improve quality slightly by improving the smoothness of the field of MV costs (and they're closer to CABAC's actual costs).
Despite being optimized for CABAC, they still help under CAVLC, albeit less.
MV cost change by Loren Merritt

commit 79bfb039de253c986986fbd99935c0d4a95ad503 r1076
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Jan 13 20:22:36 2009 -0500

Support forced frametypes with scenecut/b-adapt
This allows an input qpfile to be used to force I-frames, for example.
The same can be done through the library interface.
Document the format of the qpfile in --longhelp and the forcing of frametypes in x264.h
Note that forcing B-frames and B-refs may not always have the intended result.
Patch partially by Steven Walters <kemuri9@gmail.com>.

commit 7c4f8297e057a439cf1f1d7cf95d05b9f27063c7 r1075
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Jan 13 19:58:44 2009 -0500

Remove an IDIV from i8x8 analysis
Only one IDIV is left in macroblock level code (transform_rd)

Alles anzeigen

Zitat

x264.1093M.exe - Alternate Download

Patches used:
x264_hrd_pulldown.09_interlace.diff
x264_win_zone_parse_fix_05.diff

gcc 3.4.5 fprofiled build with -march=pentium2.

-----------------------------------------------

x264.1093M.x64.exe - Alternate Download

Patches used:
x264_hrd_pulldown.09_interlace.diff
x264_win_zone_parse_fix_05.diff
x264_win64_support.06.r1093.diff

gcc 4.4.0 fprofiled build.

Alles anzeigen

Zitat

libx264 SVN-r1093:
http://www.mediafire.com/file/0ytuq1lnd…iled-patched.7z

Compiled with MinGW GCC 4.3.2-tdm-2, march=pentium2, fprofiled. Patched back to core version 65 in order to make it work with Avidemux 2.4.3 r4532.

64-Bit Support für Windows ist nun offiziell da! :cool:

Zitat

commit f43e22a7873ea3811bbc15e30d67681f23249087 r1112
Author: Anton Mitrofanov <BugMaster@narod.ru>
Date: Wed, 11 Feb 2009 18:38:56 +0000 (10:38 -0800)

Windows 64-bit support
A "make distclean" is probably required after updating to this revision.

commit d56e13f9016b898a3bd4043b26c2e70c2bb9f6c4 r1111
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed, 11 Feb 2009 18:35:56 +0000 (10:35 -0800)

Minor fixes and cosmetics
Suppress a GCC warning, fix a non-problematic array overflow, one REP->REP_RET.

commit 05afd8e02bd83c6c9eba4d41b8d829a383689117 r1110
Author: Manuel <maaanuuu@gmx.net>
Date: Tue Feb 10 12:06:47 2009 -0800

fix 10l in 75b495f2723fcb77f
Original thread:
date: Mon, Feb 9, 2009 at 9:37 PM
subject: [x264-devel] commit: Spare a vec_perm and a vec_mergeh though using a LUT of permutation vectors . (Guillaume Poirier )

commit 75b495f2723fcb77fe7d5c92511136d3fea4cf13 r1109
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Mon Feb 9 21:17:33 2009 +0100

Spare a vec_perm and a vec_mergeh though using a LUT of permutation vectors.

commit 37f98cb85024d288eab5508a3b04ca1324335693 r1108
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Mon Feb 9 21:12:23 2009 +0100

Promote chroma planes to 16 byte alignment.
This will allow simplifying vectors loads that can only load 16-bytes
aligned data (such as AltiVec).

commit b5b9728b9b1cf9e4e54092515fd9fa86cd9023a4 r1107
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Feb 9 11:30:54 2009 -0800

Fix 10L in intra pred
Forgetting a %define resulted in SIGILL on 32-bit systems without SSE (e.g. Athlon XP).

commit 0ee50db35db3e5af0d40936dbc5ff7e2478b1a2c r1106
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sun Feb 8 23:36:40 2009 -0800

Add decimation in i16x16 blocks
Up to +0.04db with CAVLC, generally a lot less with CABAC.

commit 9bf45f6d397559486b5fe038c3847b0d35c61728 r1105
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Feb 7 02:27:16 2009 -0800

Much faster CABAC residual context selection
Up to ~17% faster CABAC RDO, ~36% faster intra-only CABAC RDO.
Up to 7% faster overall in extreme cases.

commit 3927938121ef63d72d9fd429c25202ebd65dd208 r1104
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Feb 7 01:57:43 2009 -0800

Faster coeff_last64 on 32-bit

commit 32615747d3ab5648b666b4f55531f47b3c075521 r1103
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Feb 6 02:59:36 2009 -0800

More intra pred asm optimizations
SSSE3 version of predict_8x8_hu
SSE2 version of predict_8x8c_p
SSSE3 versions of both planar prediction functions
Optimizations to predict_16x16_p_sse2
Some unnecessary REP_RETs -> RETs.
SSE2 version of predict_8x8_vr by Holger.
SSE2 version of predict_8x8_hd.
Don't compile MMX versions of some of the pred functions on x86_64.
Remove now-useless x86_64 C versions of 4x4 pred functions.
Rewrite some of the x86_64-only C functions in asm.

commit 711e6e87967aa3813a894fdfcd1e2b7eb48328a6 r1102
Author: Manuel <manuelro@gmx.de>
Date: Sun Feb 8 21:35:51 2009 +0100

Speed-up mc_chroma_altivec by using vec_mladd cleverly, and unrolling.
Also put width == 2 variant in its own scalar function because it's faster
than a vectorized one.

commit b69548aa3a0218ba1d4f934edcf8942f2b1682f5 r1101
Author: Holger Lubitz <holger@lubitz.org>
Date: Wed Feb 4 12:46:17 2009 -0800

Merging Holger's GSOC branch part 2: intra prediction
Assembly versions of most remaining 4x4 and 8x8 intra pred functions.
Assembly version of predict_8x8_filter.
A few other optimizations.
Primarily Core 2-optimized.

commit 122a54a0dbbfd5a8b649a6ca7eb0b7d3c42f89aa r1100
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Wed Feb 4 10:04:55 2009 +0000

10l: fix compilation with GCC 4.3+

commit c0be8106d40b2ccbfec37229afaecf236b03762c r1099
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Jan 31 05:00:39 2009 -0800

Faster 8x8dct+CAVLC interleave
Integrate array_non_zero with the CAVLC 8x8dct interleave function.
Roughly 1.5-2x faster than the original separate array_non_zero method.

commit 91bffdcc24e59620cc09aa4288f7e0c0b74c8891 r1098
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Jan 31 01:00:26 2009 -0800

Measure CBP cost in i8x8 RD refinement
~0.02-0.05db PSNR gain at high quants in intra-only encoding, pretty small otherwise.
Allows a small optimization in i8x8 encoding.

commit e404f350afc3c52a99d13d22f862cd2ede1438b0 r1097
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Sun Feb 1 20:58:00 2009 +0100

Take advantage of saturated signed horizontal sum instructions in
the variance computation epilogue since there won't be any overflow
triggering an overflow.
Suggested by Loren Merritt

commit 4c171c3b1c803a173ace823275882b3c9a2ecd24 r1096
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Jan 30 03:40:54 2009 -0800

Massive overhaul of nnz/cbp calculation
Modify quantization to also calculate array_non_zero.
PPC assembly changes by gpoirior.
New quant asm includes some small tweaks to quant and SSE4 versions using ptest for the array_non_zero.
Use this new feature of quant to merge nnz/cbp calculation directly with encoding and avoid many unnecessary calls to dequant/zigzag/decimate/etc.
Also add new i16x16 DC-only iDCT with asm.
Since intra encoding now directly calculates nnz, skip_intra now backs up nnz/cbp as well.
Output should be equivalent except when using p4x4+RDO because of a subtlety involving old nnz values lying around.
Performance increase in macroblock_encode: ~18% with dct-decimate, 30% without at CRF 25.
Overall performance increase 0-6% depending on encoding settings.

commit 8ae672fe2e0f779e57abe560bafdf41ec1fea533 r1095
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Thu Jan 29 01:28:12 2009 -0800

Add PowerPC support for "checkasm --bench", reading the time base register.
This isn't ideal since the `time base' register is running at a fraction
of the processor cycle speed, so the measurement isn't as precise as x86's
rdtsc.
It's better than nothing though...

commit c9095eb2792b18a264569d46a3d923eafe333f45 r1094
Author: Brad <brad@comstyle.com>
Date: Thu Jan 29 04:35:34 2009 +0000

fix detection of pthread and isfinite on OpenBSD

commit 1df50b9287c83d5443d19482345b6842b78081c3 r1093
Author: Loren Merritt <pengvado@akuvian.org>
Date: Tue Jan 27 05:42:51 2009 +0000

remove $ECHON kludge, which broke on SunOS. bring back `gcc -MT`.
remove auto-reconfigure on svn update, which has done nothing since we stopped using svn.
fix $AS on sparc (was disabled by mmx check).
fix --extra-asflags (was ignored).
mark bash scripts as bash, not sh

patch partly by Greg Robinson and Jugdish.

Alles anzeigen

Zitat

libx264 SVN-r1112:
http://www.mediafire.com/file/4otmlmzmn…32-fprofiled.7z

Compiled with MinGW GCC 4.3.2-tdm-2, march=pentium2, fprofiled. Tested with Avidemux 2.4.4 Final.

Update: Pre-Scenecut ist jetzt immer an, die entsprechende Option wurde entfernt!

Zitat

commit 8544346a43456720f07e6a438cfbb0d84b39779a r1120
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Mar 4 03:16:06 2009 -0800

Remove pre-scenecut from fprofile commands as well
Also add psy-trellis to fprofile

commit 6f0b2a9b18f3af3fd7e495640756e1d5e43343e1 r1119
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Mar 3 16:21:52 2009 -0800

Slightly faster 8x16 SAD on Penryn Core 2
Same as MMX 8x16 cacheline SAD, but calls SSE2 8x16 SAD in non-cacheline case.
Only Nehalem benefits from sizes smaller than 8x16, and Nehalem doesn't use cacheline functions, so no smaller versions are included.

commit 1cc16dcee61496a5fb4da80d7605c1c88e2d371d r1118
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Feb 26 19:50:09 2009 -0800

Fix scenecut and VBV with videos of width/height <= 32
Also remove an unused variable

commit 3e4946f305317856ed79e0898f25f10859df22ed r1117
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Feb 26 14:29:50 2009 -0800

Remove non-pre scenecut
Add support for no-b-adapt + pre-scenecut (patch by BugMaster)
Pre-scenecut was generally better than regular scenecut in terms of accuracy and regular scenecut didn't work in threaded mode anyways.
Add no-scenecut option (scenecut=0 is now no scenecut; previously it was -1)
Fix an incorrect bias towards P-frames near scenecuts with B-adapt 2.
Simplify pre-scenecut code.

commit 7ddb2c7da0621bb853b6702e6f59619c2d1c6a08 r1116
Author: Guillaume Poirier <gpoirier@mplayerhq.hu>
Date: Tue Mar 3 07:44:18 2009 -0800

Add AltiVec version of hadamard_ac. 2.4x faster than the C version.
Note this this implementation is pretty naive and should be improved
by implementing what's discussed in this ML thread:
date: Mon, Feb 2, 2009 at 6:58 PM
subject: Re: [x264-devel] [PATCH] AltiVec implementation of hadamard_ac routines

commit 11863ace50e918ec75f7c8e22907ebf1000820e1 r1115
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Feb 26 12:07:56 2009 -0800

Fix regression in r1085
Deblocking was very slightly incorrect with partitions=all.
Bug found by BugMaster.

commit a933a3e6a6be918a2ae56e3d94ecea29143b9ea5 r1114
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Feb 16 05:56:12 2009 -0800

Optimize neighbor CBP calculation and fix related regression
r1105 introduced array overflow in cbp handling

commit cc4f807796220a042e386da259ddc38e6ca8e43b r1113
Author: Tal Aloni <tal.aloni.il@gmail.com>
Date: Fri Feb 13 16:30:14 2009 -0800

Show FPS when importing a raw YUV file

Alles anzeigen

Zitat

libx264 SVN-r1119:

MinGW GCC 4.3.3-tdm-1:
http://www.mediafire.com/file/ynfaxm2hj…filed-core66.7z

MinGW GCC 3.4.5-r3:
http://www.mediafire.com/file/murxn2z40…filed-core66.7z

Both builds: march=pentium2, fprofiled. Tested with Avidemux 2.5 SVN-r4648.

Klingt nach einem beachtlichen Speed-Up und ist daher wohl eine News wert:

Zitat

commit 10d6ef07409ebe38b5f1e8e4516155a2fe66d4c6 r1125
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Mar 7 16:43:09 2009 -0800

Faster SSE2 pixel_var
Optimized using the DEINTB method from r1122. ~32% faster var_16x16 on Conroe.

commit e1c7f12cb828072b2b5b096195ff3004b83c7785 r1124
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sat Mar 7 00:27:27 2009 -0800

SSSE3 hpel_filter_v
Optimized using the same method as in r1122. Patch partially by Holger.
~8% faster hpel filter on 64-bit Nehalem

commit 3d780622c1ff19bc3d6a522a65879779a7ddb3dd r1123
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Mar 6 18:57:15 2009 -0800

Update some asm copyright headers

commit 2dca5f5413051a26cbba4e20f3c77ff69b694ba3 r1122
Author: Holger Lubitz <holger@lubitz.org>
Date: Fri Mar 6 18:16:30 2009 -0800

Vastly faster SATD/SA8D/Hadamard_AC/SSD/DCT/IDCT
Heavily optimized for Core 2 and Nehalem, but performance should improve on all modern x86 CPUs.
16x16 SATD: +18% speed on K8(64bit), +22% on K10(32bit), +42% on Penryn(64bit), +44% on Nehalem(64bit), +50% on P4(32bit), +98% on Conroe(64bit)
Similar performance boosts in SATD-like functions (SA8D, hadamard_ac) and somewhat less in DCT/IDCT/SSD.
Overall performance boost is up to ~15% on 64-bit Conroe.

commit f3872178768cca2973f759c479e26f3ac35e55fe r1121
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Fri Mar 6 15:28:47 2009 -0800

Update x264 copyright date

Alles anzeigen

Zitat

libx264 SVN-r1125:

MinGW GCC 4.3.3-tdm-1:
http://www.mediafire.com/file/y2i5nwiyy…filed-core66.7z

MinGW GCC 3.4.5-r3:
http://www.mediafire.com/file/jth0mymrm…filed-core66.7z

Both builds: march=pentium2, fprofiled. Tested with Avidemux 2.5 SVN-r4658 -and- 2.4.4 Final.

Zitat

x264.1125M.x86.exe - Alternate Download / x264.1125M.x64.exe - Alternate Download
gcc 3.4.5 fprofiled build with -march=pentium2. / gcc 4.3.4 fprofiled build.

Patches used:

x264_hrd_pulldown.10_interlace.diff
x264_win_zone_parse_fix_05.diff

Es ist mal wieder Zeit, den News Thread auf den aktuellen Stand zu bringen:

Zitat

commit 37424f87a43a2dcecba61bc24b46506ced32307c r1140
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Apr 14 14:47:02 2009 -0700

Slightly faster SSE4 SA8D, SSE4 Hadamard_AC, SSE2 SSIM
shufps is the most underrated SSE instruction on x86.

commit 1024283b0321e53a3e08fddb1411429330bf1731 r1139
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Apr 9 02:14:41 2009 -0700

Various CABAC optimizations
Move calculation of b_intra out of the core residual loop and hardcode it where applicable.
Inlining cabac_mb_mvd was unnecessary and wasted tremendous amounts of code size. Inlining only cache_mvd is faster and significantly smaller.

commit fe11a6f39e4e8235d685591ec9c0ec86eca4fee9 r1138
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Wed Apr 8 05:45:03 2009 -0700

CAVLC optimizations
faster bs_write_te, port CABAC context selection optimization to CAVLC.

commit 1fda88277f6b2eda27a0f741d58b31532ad0664d r1137
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sun Apr 5 13:01:42 2009 -0700

Faster CABAC RDO
Since the bypass case is quite unlikely, especially when doing merged sigmap/level coding,
it's faster to use a branch than a cmov.

commit 3f9ba82b9788231c879d31d9c4c8ebf4518d07fe r1136
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Mar 31 10:36:57 2009 -0700

Activate intra_sad_x3_8x8c in lookahead

commit 9d0c378b235182341edd2e95f01d4fd25132ad50 r1135
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Mar 31 10:34:35 2009 -0700

MBAFF interlaced coding is not allowed in baseline profile

commit b8808bf0a35b83c2fac05e07a504ad164c931960 r1134
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Mar 30 19:30:59 2009 -0700

intra_sad_x3_8x8 assembly

commit 8df40a4c41eb5e79d0055f2e5af4b214285b9c8c r1133
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Mar 30 16:37:46 2009 -0700

intra_sad_x3_4x4 assembly

commit d39f8ae1730b07286f1bb281a22d8cd57d0f90b9 r1132
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Mar 30 04:07:50 2009 -0700

intra_sad_x3_8x8c assembly
Also fix intra_sad_x3_16x16's use of "n" as a loop variable (broke SWAP)

commit d3ca4647a247186c2df7760be2a9c649efe34815 r1131
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Sun Mar 29 18:27:32 2009 -0700

Shave one instruction off CABAC encode_decision
range_lps>>6 ranges from 4-7, so (range_lps>>6)-4 == (range_lps>>6) & 3

commit 847597773aff56a5612a9edbebb40b350c637edf r1130
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Mar 26 22:22:23 2009 -0700

Faster probe_skip
Add a second chroma threshold after the DC transform.

commit c109c8e7db67df7194f2f913a7a4d65217caee26 r1129
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Thu Mar 19 12:28:21 2009 -0700

Add missing "static" qualifier to two arrays
Should slightly improve performance.

commit 682b54d6175f98dfa14fec4d951f4b3b6e686b95 r1128
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Tue Mar 17 11:01:57 2009 -0700

SSE2 zigzag_interleave
Replace PHADD with FastShuffle (more accurate naming).
This flag represents asm functions that rely on fast SSE2 shuffle units, and thus are only faster on Phenom, Nehalem, and Penryn CPUs.

commit 8d82fecc3377b3052279f038f2273ade3a5b65cc r1127
Author: Jason Garrett-Glaser <darkshikari@gmail.com>
Date: Mon Mar 9 23:37:53 2009 -0700

Faster integral_init
palignr to avoid unaligned loads is worth it in inith, but not initv.

commit 96733ab692b4a268685d65070d6977964a466c91 r1126
Author: Holger Lubitz <holger@lubitz.org>
Date: Mon Mar 9 14:05:16 2009 -0700

Faster SSSE3 hpel_filter_v
~10% faster hpel_filter on 64-bit Penryn.
32-bit version by Jason Garrett-Glaser.

Alles anzeigen

Zitat

x264_x64_r1140_unpatched | MD5
GCC 4.3.4 20090408 (prerelease) (x64.core2.Komisar), unpatched, generic, fprofiled

________________________________________________________________________________

x264_x86_r1140_techouse | INFO
GCC 4.3.3, fprofiled, -march=core2

x264_x64_r1140_techouse | INFO
GCC 4.3.4 20090408 (prerelease) (x64.core2.Komisar), fprofiled, -march=core2

Patches used:

x264_hrd_pulldown.11_interlace.diff
x264_win_zone_parse_fix_05.diff

Alles anzeigen

Zitat

libx264 SVN-r1140:

libx264-r1140, MinGW GCC 4.3.3, optimized for Core 2
libx264-r1140, MinGW GCC 4.3.3, optimized for K8
libx264-r1140, MinGW GCC 4.3.3, optimized for Pentium 2 and later
libx264-r1140, MinGW GCC 4.2.1, optimized for Pentium 2 and later
libx264-r1140, MinGW GCC 3.4.5, optimized for Pentium 2 and later

All builds are fprofiled. Tested with Avidemux 2.5 SVN-r4685 -and- 2.4.4 Final.
The patches used for my builds can be found at this location.