[FFmpeg-devel] [RELEASE] ffmpeg-0.6.90-rc0

[Mikhail I. Izmestev]

04.04.2011 19:15, Mikhail I. Izmestev пишет:
> 04.04.2011 16:56, Michael Niedermayer wrote:
>> On Mon, Apr 04, 2011 at 02:33:48PM +0400, Mikhail I. Izmestev wrote:
>>> Hi
>>> 03.04.2011 21:13, Michael Niedermayer wrote:
>>>> Hi
>>>>
>>>> Iam proud to announce that after a way too long time we finally are
>>>> making a new release.
>>>>
>>>> See: http://ffmpeg.org/releases/
>>>>
>>>> Please help test it and report regression& security bugs to
>>>> ffmpeg.org/trac so we can fix them in rc1 or the final release.
>>>> (depending on if we will need a rc1)
>>>>
>>>> Also please help testing on not so common platforms if you can. (as you
>>>> know the 3 server admins took all machines they had access to with
>>>> them for their fork)
>>> I'm not sure, but I have tested ffmpeg at
>>> 2cae9809e2d59c7336fc2cccb97b82c7f764868a on win7 and have performance
>>> dropping:
>>>
>>> $ time ./ffmpeg.exe -i BackgroundLoopBlueHorizon.mov -sameq -f rawvideo
>>> -> /dev/null
>>> FFmpeg version UNKNOWN, Copyright (c) 2000-2011 the FFmpeg developers
>>> built on Apr 4 2011 14:14:06 with gcc 4.4.0 20090326 (experimental)
>>> configuration: --enable-shared --disable-static --enable-memalign-hack
>>> --disable-avfilter --arch=x86_64 --cross-prefix=x86_64-pc-mingw32-
>>> --target-os=mingw32
>>> --disable-dxva2 --disable-yasm
>> disable-yasm is a bad idea, you should install yasm
>> also please show us your config.h of both versions
>>
>> and if you can figure out which revission causes the slowdown for you
>> with "git bisect" thats easy, this would help
> There are git bisect results:
>
> 90441276e4f661c6aec5e4d2c5718cde1ff1946d is the first bad commit
> commit 90441276e4f661c6aec5e4d2c5718cde1ff1946d
> Author: Nicolas George <nicolas.george at normalesup.org>
> Date: Fri Feb 4 19:12:37 2011 +0100
>
> Non-blocking protocol: core wrapper functions
>
> Signed-off-by: Nicolas George <nicolas.george at normalesup.org>
> Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
>
> :040000 040000 b4ef1b8023662f6e575f2a7e421aee93142dfc95 2cdaedf8102e66edf97ae2f4c51a2730824f471f M
> libavformat
[...]

Ok, I have tracked down source of performance dropping.
Problem caused interrupt checking and related only to ffmpeg app not to libavformat.

Mikhail.