Op dinsdag 24-03-2009 om 08:01 uur [tijdzone -0400], schreef Sean M.
Pappalardo:
> 
> Richard Rasker wrote:
> > The AMD64 "example" is just an empty line, so perhaps there's nothing to
> > optimize there? Or are there any specific options/flags like the above
> > that might be appropriate for the AMD64 CPU? 
> 
> Here are the optimization flags Mixxx uses. I'm sure they'll help:
> -O3 -march=k8-sse3 -mmmx -msse2 -msse3 -m3dnow -mfpmath=sse
> -fomit-frame-pointer -ffast-math -funroll-loops
> 
> If your CPU doesn't support SSE3, use:
> -O3 -march=k8 -mmmx -msse2 -m3dnow -mfpmath=sse -fomit-frame-pointer
> -ffast-math -funroll-loops


OK, it doesn't support SSE3, so I tried the second set of flags -- and
it appears there's an improvement of some 7% in CPU load and render time
(short benchmark project). Not bad, although it's slightly less than I
hoped for.

Then again, I guess I should compile everything from the ground up,
including all libraries, if I want to squeeze the most out of this
beastie. Which means I'd better have a look at Gentoo :-/

Anyway, thanks for the good advice,

Best regards,

Richard Rasker


_______________________________________________
Cinelerra mailing list
[email protected]
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Reply via email to