Re: realtime altivec fft -- checking & performance
Re: realtime altivec fft -- checking & performance
- Subject: Re: realtime altivec fft -- checking & performance
- From: Brian Whitman <email@hidden>
- Date: Mon, 13 Jan 2003 12:52:35 -0500
On Monday, January 13, 2003, at 12:14 PM, Urs Heckmann wrote:
>
bool is_16byte_aligned ( void* theStuff )
>
{
>
if ( (int) theStuff & 15 != 0 ) return false; // should work.
>
return true;
>
}
Thanks Urs and Shaun for this trick, it works well. And thanks irieblue
for the sample code pointer, that's what I'm basing this off of. It
looks like everything is altivec-enabled (I was using -faltivec, yes)
and I'm pretty happy with the current speedup to 9%-10%. So if anyone
has this problem later on, here's how I aligned my vector:
Float32 __attribute__ ((aligned (16))) *d;
& I use new[]s to malloc it in.
thanks!
--
Brian Whitman - email@hidden
http://web.media.mit.edu/~bwhitman
Music, Mind and Machine Group - MIT Media Lab
_______________________________________________
coreaudio-api mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/coreaudio-api
Do not post admin requests to the list. They will be ignored.