Current Patches, Where to get them, How they affect speed/output

[kemuri-_9]

i used my compiles of gcc 4.4.1 to compile both x86 and x64 versions of x264 and it works fine.

compiling from the svn can be a dual edged sword - bugs are likely to appear as other ones are fixed,
this is why i only use official releases now.
(plus it saves time from building from weekly snaps and such)

[Audionut]

Subme10

http://git.videolan.org/?p=x264.git;...d2a394b2a2b713

[Tarutaru]

x264.0.68.1189M.x86.exe

* patch -p1 < x264_hrd_pulldown.14_interlace.diff
* patch -p1 < x264_win_zone_parse_fix_05.diff
* pthread, gpac
* fprofiled, -msse4.2, -march=native(core i7)

[JEEB]

x264 r1189 32bit
download ; release notes

• built on Jul 26 2009, gcc: 4.3.4 20090220 (prerelease) (x32.generic.Komisar)
• fprofiled, -march=i686

x264 r1189 64bit
download ; release notes

• built on Jul 26 2009, gcc: 4.3.4 20090220 (prerelease) (x64.generic.Komisar)
• fprofiled, -march=core2

patched with:

• x264_win_zone_parse_fix_05.diff
• x264_hrd_pulldown.14_interlace.diff

[imk]

x264-r1190M-imk-win.7z
win_build_info.txt

[bob0r]

Need someone who can compile x264 64bit fprofiled from GIT (no patches), gcc prefered.
Techouse seems very busy.

If you can provide this please host it on a http host (i can provide if needed):
http://yourhost.com/whatever/revisionXXXX/x264.exe
Like that.
Also include:
md5sum x264.exe | awk '{print $1}' >x264.md5
please, thats the file i use and check for updates.
( http://yourhost.com/whatever/revisionXXXX/x264.md5 )

If you can provide this please post for me x264.exe --version, so i can update x264.nl

Maybe JEEB?

[juGGaKNot]

Originally Posted by JEEB

x264 r1189 32bit
download ; release notes

• built on Jul 26 2009, gcc: 4.3.4 20090220 (prerelease) (x32.generic.Komisar)
• fprofiled, -march=i686

x264 r1189 64bit
download ; release notes

• built on Jul 26 2009, gcc: 4.3.4 20090220 (prerelease) (x64.generic.Komisar)
• fprofiled, -march=core2

patched with:

• x264_win_zone_parse_fix_05.diff
• x264_hrd_pulldown.14_interlace.diff

Was autovaq commited or still diff ?

A subme 10 build with autovaq ?

[LoRd_MuldeR]

Originally Posted by juGGaKNot

Was autovaq commited or still diff ?

Yes. Look at the GIT

http://git.videolan.org/gitweb.cgi?p...5752da9f2cd032

Originally Posted by juGGaKNot

A subme 10 build with autovaq ?

You don't need a "SubME 10" build. SubME=10 is available in all regular builds (r1187 and later). Note that it requires Trellis=2.

[juGGaKNot]

Originally Posted by LoRd_MuldeR

Yes. Look at the GIT

http://git.videolan.org/gitweb.cgi?p...5752da9f2cd032



You don't need a "SubME 10" build. SubME=10 is available in all regular builds (r1187 and later). Note that it requires Trellis=2.

subme 10 + autovaq ( that is if autovaq was not commited )

anyway tested, guess this means death to deathzones.

will the --preset be edited so that subme 10 is on @ slower and placebo ?

[LoRd_MuldeR]

Originally Posted by juGGaKNot

subme 10 + autovaq ( that is if autovaq was not commited )

That is what you get with "--aq-mode 2" + "--trellis 2" + "--subme 10" using a standard (unpatched) build.

Originally Posted by juGGaKNot

will the --preset be edited so that subme 10 is on @ slower and placebo ?

The "slower" preset uses Trellis=2 + SubMe=9, but "placebo" already uses Trellis=2 + SubMe=10.

I wonder whether SubMe=10 really is a "placebo" setting?

[juGGaKNot]

Originally Posted by LoRd_MuldeR

That is what you get with "--aq-mode 2" + "--trellis 2" + "--subme 10" using a standard (unpatched) build.

Well yes if its not a diff anymore

What does the zone parse fix do ? now that autovaq is in i can move to a unpatched build.

The "slower" preset uses Trellis=2 + SubMe=9, but "placebo" already uses Trellis=2 + SubMe=10.

I wonder whether SubMe=10 really is a "placebo" setting?

We will see when subme 11 is out i'm going to use it.

Originally Posted by Dark Shikari

Does that mean it's any louder?

I do not have amps :|

[Dark Shikari]

Originally Posted by juGGaKNot

We will see when subme 11 is out i'm going to use it.

Does that mean it's any louder?

[LoRd_MuldeR]

Originally Posted by juGGaKNot

Well yes if its not a diff anymore

As said before, AutoVAQ is committed

Originally Posted by juGGaKNot

What does the zone parse fix do ? now that autovaq is in i can move to a unpatched build.

See this post:
http://forum.doom9.org/showpost.php?...92&postcount=7

Originally Posted by juGGaKNot

I do not have amps :|

You don't need to have an amp at home to understand the joke

[Selur]

now that autovaq is in i can move to a unpatched build.

not before hrd_pulldown_interlace get's stable and finds it way into the maintree

[JEEB]

x264 r1190 64bit unpatched:
download ; hash
________________________________________________________________________________

x264 r1190 32bit
download ; release notes

• built on Jul 28 2009, gcc: 4.3.4 20090220 (prerelease) (x32.generic.Komisar)
• fprofiled, -march=i686

x264 r1190 64bit
download ; release notes

• built on Jul 28 2009, gcc: 4.3.4 20090220 (prerelease) (x64.generic.Komisar)
• fprofiled, -march=core2

patched with:

• x264_win_zone_parse_fix_05.diff
• x264_hrd_pulldown.15_interlace.diff

[XhmikosR]

Using JEEB's builds, r1190, r1189 and r1184 I get the following when x264 crashes:

avis [info]: 1280x720 @ 23.98 fps (1442 frames)
Assertion failed: dpb_output_delay < pow( 2, sps->vui.nal_hrd_parameters.i_dpb_o
utput_delay_length ), file encoder/set.c, line 643

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

cmd used:

x264 --profile high --slow-firstpass --pass 1 --quiet --bitrate 5000 --stats "1.stats" --level 4.1 --keyint 24 --min-keyint 1 --ref 1 --no-mixed-refs --bframes 3 --no-weightb --direct auto --chroma-qp-offset 0 --subme 1 --trellis 0 --analyse none --no-8x8dct --ipratio 1.1 --pbratio 1.1 --vbv-bufsize 50000 --vbv-maxrate 50000 --qcomp 0.5 --me dia --threads auto --thread-input --sar 1:1 --output NUL "test.avs" --mvrange 511 --aud --nal-hrd

If I use imk's build r1190 everything is fine.

[Dark Shikari]

Originally Posted by XhmikosR

If I use imk's build r1190 everything is fine.

Note where it terminates. It's a bug in the nal-hrd patch.

[XhmikosR]

Yes, but the odd thing is that the same script, same cmd, same patches used, all the same, only the compiler is different, with imk's build I get no crashes. Anyway, I didn't know where else to report it. I hope it gets fixed.

[Trahald]

thats why i put the assert in there rather the encoder die than a decoder down the line. (because w/out the assert the encode would have completed). will have a patch out shortly.

[Trahald]

Hrd 15.