x264 - OpenSource AVC/H.264 Video Codec

Mal wieder ein Release:

r683: Add AltiVec implementation of quant_2x2_dc, fix Altivec implementation of quant_(4x4|8x8)(|_dc) wrt current C implementation
Patch by Noboru Asai % noboru DD asai AA gmail DD com %

Und noch zwei Releases hinterher:

r684: add AltiVec implementation of dequant_4x4 and dequant_8x8, 2.8x faster than C,
1.01x faster than previous revision with default encoding options
Patch by Noboru Asai % noboru DD asai AA gmail DD com %

r685: Add AltiVec implementation of add4x4_idct, add8x8_idct, add16x16_idct, 3.2x faster
on average
1.05x faster overall with default encoding options
Patch by Noboru Asai % noboru DD asai AA gmail DD com %

Hinweis:

AltiVec ist ein System zur Parallelberechnung (wie MMX und SSE) für Motorola / IBM PowerPC-Prozessoren, verwendet in Macintosh G4 / G5. Für PC-Besitzer also völlig uninteressant.

MMX/SSE-Beschleunigung für Quantisierungsfunktionen gibt es für intel- und kompatible CPUs schon länger.

r686: cosmetics: use separate variables for frame width and stride

r687: move hpel_filter cpu detection to a function pointer like everything else
r688: cosmetics: use symbolic constants for frame padding radius

r689: remove sad_pde. it's been unused ever since successive elimination replaced it.

r690: cosmetics in dsp function selection

r691: Add AltiVec implementation ofx264_hpel_filter. Provides a 10-11% overall speed-up
with default encoding options
Patch by Noboru Asai %noboru P asai A gmail P com%

r692: add AltiVec implementation of ssim_4x4x2_core, about 4x faster than C version.
Overall: 0.1-0.2% faster with default encoding settings
Patch by Noboru Asai %noboru P asai A gmail P com%

r693: separate pixel_avg into cases for mc and for bipred

r694: cosmetics: reorder mc_luma/mc_chroma/get_ref arguments for consistency with other functions

und wieder gibt es neue versionen.

die arbeiten ja wie am fliessband da.

r696 | pengvado | 2007-11-20 07:07:17 +0100 (Tue, 20 Nov 2007) | 4 lines

avoid memory loads that span the border between two cachelines.
on core2 this makes x264_pixel_sad an average of 2x faster. other intel cpus gain various amounts. amd are unaffected.
overall speedup: 1-10%, depending on how much time is spent in fullpel motion estimation.

------------------------------------------------------------------------
r695 | pengvado | 2007-11-20 06:57:29 +0100 (Tue, 20 Nov 2007) | 2 lines

add cache info to cpu_detect. also print sse3.

Da hast du doch glatt eine vergessen. (Gut, die kam auch 3 Stunden später)

r697: limit mvs to [-512,511.75] instead of [-512,512]

so nun gibts auch schon die 698

warum geht das dort so flink??!

hubble

Weil das im Grunde nicht viel Arbeit ist, wenn sich jemand damit so richtig auskennt. Ein gewisser "Nachteil" bei SVN ist, dass jede kleine Änderung unter Umständen eine neue Revision erzeugt. Vor allem wenn man sie alle einzeln veröffentlicht (was Vorteile haben kann, wenn man eine fehlerhafte Revision zurückziehen muss).

Wieder ein Altivec-Patch (für PowerPC/Mac, intel-PCs haben das längst als SSE-Variante).

Zitat

Add AltiVec implementation of x264_pixel_ssd_8x8, 3x faster than C version
Overall speed-up: 0.7% with --bframes 3 --ref 5 -m 7 --b-rdo
Patch by Noboru Asai %noboru P asai A gmail P com%

Und schon ist die nächste da:

r699: cover some more options in fprofile. (esa, bime, cqm, nr, no-dct-decimate, trellis2)
previously, esa was slower with fprofile than without, since gcc thought it wasn't
important. now esa benefits like anything else.

Und zur Info der Link zur r698

r700: add support for x86_64 on Darwin9.0 (Mac OS X 10.5, aka Leopard)
Patch by Antoine Gerschenfeld %gerschen A clipper P ens P fr%

r701: fix compilation as a shared library on x86_64 (regression in r696)

r702: fix a memleak in h->mb.mvr

r703: avoid a division in umh. patch by Dark Shikari.
r704: avoid a division in x264_mb_predict_mv_ref16x16. patch by Dark Shikari.
r705: faster removal of duplicate mv predictors
r706: reduce the data type used in some tables. 16KB smaller exe.

r707: check whether ld supports -Bsymbolic before using it
r708: mark asm functions as hidden

r709: revert the x86_32 part of r708. elf shared libraries aren't important enough to be worth the extra lines of code to check for nasm.

r710: adds AliVec? implementation of predict_16x16_p() over 4x faster than C version

r711: Add AltiVec? implementation of <verschiedene Funktionen> - each around 1.3 tp 1.8x faster than C version. Patch by Noboru Asai % noboru P asai A gmail P com%

r712: also test arch-specific x264_zigzag_* implementations in checkasm.c
patch by Patch by Noboru Asai % noboru P asai A gmail P com%

r713: fix a nondeterminism involving 8x8dct, rdo, and threads.
r714: tweak x264_pixel_sad_x4_16x16_sse2 horizontal sum. 168 -> 166 cycles on core2.

r715: use define _WIN32 instead of __WIN32__ or WIN32 defines.
MSDN reference: http://msdn2.microsoft.com/en-us/library/b0084kay(VS.80).aspx
Patch by BugMaster %BugMaster A narod P ru%
Original thread:
date: Dec 27, 2007 3:18 AM
subject: [x264-devel] VS2008 compilation error (need of replacement __WIN32__ with _WIN32)

lange keine nachricht mehr hier gepostet.

es gibt aber trotzdem eine neue version

r.741

http://mirror01.x264.nl/x264/revision741/x264.exe

hubble

Ich erbarme mich fürn Changelog.

r716: round esa range to a multiple of 4
r717: convert absolute difference of sums from mmx to sse2
convert mv bits cost and ads threshold from C to sse2
convert bytemask-to-list from C to scalar asm
1.6x faster me=esa (x86_64) or 1.3x faster (x86_32). (times consider only motion estimation. overall encode speedup may vary.)
r718: slightly wrong memory allocation in r717, fixes a potential crash with merange>32
r719: shut up some valgrind warnings
r720: add / fix support for FreeBSD, based on a patch by Igor Mozolevsky % igor A hybrid-lab P co P uk %
r721: change the meaning of --ref: it now selects DPB size (including B-frames), rather than L0 size (which B-frames are added to)
r722: configure test for endianness
r723: faster and smaller x264_macroblock_cache_mv etc
r724: shave a couple cycles off cabac functions
r725: fix mvp for b16x8 and b8x16 L1 search
patch by Wei-Yin Chen
r726: fix mapping of mvs to partitions in p4x4_chroma
patch by Noboru Asai
r727: fix cabac context for nonzero delta_qp of the 2nd mb of a frame in interlaced mode
r728: satd exhaustive motion search (--me tesa)
r729: fix a memleak in mkv muxer
patch by saintdev
r730: fix a memleak in cqm
r731: remove a useless bit of me=umh (originally copied from JM, where it was used for something)
r732: misc fixes in checkasm
r733: remove colorspace conversion support, because it has no business in any codec
r734: update date
r735: print average of macroblock QPs instead of frame's nominal QP
r736: intra_rd_refine in B-frames
r737: checkasm: test idct with the range of coefficients what can really be encountered, as opposed to random numbers which might overflow.
r738: checkasm: relax the threshold for floating-point ssim
r739: x86_32 sse2 idct8
ported from ffmpeg by Dark Shikari
r740: faster array_non_zero
r741: some simplifications to mmx intra pred that should have been done way back when we switched to constant fdec_stride.
and remove pic spills in functions that have a free caller-saved reg.
patch partly by Dark Shikari.

und wieder ein update

http://mirror01.x264.nl/x264/revision745/x264.exe

Dann :
1. mach doch keinen direkten Link auf die Datei, da das dem Betreiber gegenüber meist als sehr unhöflich empfunden wird.
2. poste bitte demnächst auch die Einträge aus dem Changelog
-> ideal wäre Changelogeinträge + Link zum Downloadanbieter

hier mal das Changelog:

Zitat

------------------------------------------------------------------------
r745 | pengvado | 2008-03-02 04:04:07 +0100 (Sun, 02 Mar 2008) | 2 lines

pic macros now keep track of which register holds the GOT, so variable access doesn't have to care

------------------------------------------------------------------------
r744 | pengvado | 2008-03-02 03:27:45 +0100 (Sun, 02 Mar 2008) | 2 lines

remove x86_64 predict_8x8_ddl_mmxext because sse2 is faster even on amd

------------------------------------------------------------------------
r743 | pengvado | 2008-03-02 03:26:00 +0100 (Sun, 02 Mar 2008) | 2 lines

cosmetics in dsp init

------------------------------------------------------------------------
r742 | pengvado | 2008-03-02 03:11:12 +0100 (Sun, 02 Mar 2008) | 4 lines

sse2 16x16 intra pred.
port the remaining intra pred functions from x86_64 to x86_32.
patch by Dark Shikari.

Alles anzeigen

ok

Scheinbar gibt es ein Problem mit r745:

Zitat von Dark Shikari

Strongly recommend you get revision 746 [SVN was switched to git, so it might take a while for SVN mirror to update] given that 745 has an alignment bug that results in some SSE2 prediction methods crashing.

Link für 746 ?

Zitat von Eastermeyer

Link für 746 ?

Das Update scheint noch nich im SVN zu sein, da man wohl auf "git" zur Versionsverwaltung umgestiegen is:
http://git.videolan.org/?p=x264.git;a=summary

Weiss jemand, ob es bei git überhaupt Revisionsnummern im klassischen Sinne gibt?

Zitat

Weiss jemand, ob es bei git überhaupt Revisionsnummern im klassischen Sinne gibt?

Gibt es nicht.
lies mal:
http://wincent.com/a/about/wincen…d_numbers_i.php

-> i.d.R. wird anstatt der RevisionsNummer eher das Datum verwendet. (intern immer ein SHA-1 Key, aber die kann man so schlecht lesen ;))