[x264-devel] Yet another updated threaded slicetype patch (v14)

[Steven Walters]

./x264.exe --lookahead 30 --b-adapt 2 -b10 --crf 18 -o NUL
../foreman_cif_352x288.yuv --progress --pass 1 --threads 4
still seg faults on mingw

it's the exact same seg fault that i reported from v13... so check my
report from there for the details.

at 9 bframes, it's working fine,
but when i raise it to 10 and higher it starts seg faulting on b-adapt 2.
haven't seen problems with b-adapt 1.

On Wed, Apr 8, 2009 at 5:46 PM, Mike Kazmier <DaKaZ at zenbe.com> wrote:
> I am resending this as ithas been 24 hours and my first email did not show up on the list:
>
> Alright - thanks to all those who found yet a few more bugs, and to Dark Shikari who continually insisted it was a bug in my code, and not the pthread_cond_wait library that was causing it to go slower than the old usleep version. Please find attached yet another updated patch - this one is back to getting our 6-10% speed gains - settings and CPU count dependent. I am getting 10%+ on a dual core system with b-adapt 2. I think I have all the bugs (16 frames @ b-adapt 2, and >8 threads == deadlock) out of this one, but again encourage folks to test on various platforms and settings. FWIW: the speed problem was that I was blocking the input queue while in slicetype decide, thereby stalling the encoding threads.
>
> --DaKaZ
>
> _______________________________________________
> x264-devel mailing list
> x264-devel at videolan.org
> http://mailman.videolan.org/listinfo/x264-devel
>
>