[MPlayer-dev-eng] speed of link strategies

[Uoti Urpala]

On Thu, 2008-07-31 at 03:26 +0200, Michael Niedermayer wrote:
> and after deleting the trash (old raw yuv output from ffmpeg that was still laying
> around) its down to 6gb from which at least half is actually accessed in each
> pass.
> So the really read amount is a little above 3gb + the raw output that is
> written, compared and deleted for each stream, if thats counted too then 6gb
> is likely a better estimate.
> Still this amount flushes the cache very effectively ...

Yes that does flush the cache, but it's also slow enough that I think
it's not a very typical compile/test/change cycle. If you don't care so
much about the speed of the tests but want the next compile to be fast,
what about my suggestion to read the build tree back into cache at the
end of the test script?