[FFmpeg-devel] [PATCH] force dnxhd encoder to be independent of qsort internals
Sun Sep 20 15:04:50 CEST 2009
On Sun, Sep 20, 2009 at 02:13:03PM +0200, Reimar D?ffinger wrote:
> And obviously I am somehow messing up the whole test since the overall
> speed seems to have doubled according to START_TIME/STOP_TIMER which is
> just impossible.
This is one of the days where I feel really silly.
I forgot that this machine is one of those that can adjust the clock
frequency independently for each core, and I was using my old script
that forced the CPU frequency only for the first CPU.
So the benchmark results depended on which CPU ffmpeg ended up on.
I feel like hammering my head against the wall.
More information about the ffmpeg-devel