Re: ATLAS and octave

[chenu-tournier marc]

Hello.

I'm constantly looking for more performance in floating point operations but
I'm stuck to matlab.

I have compiled on my dual pII 300 under Mandrake 7.0 octave 2.0.15 linked
to atlas as jwe proposed it and for the operation
a=rand(1000);b=rand(1000);
t=cputime;a*b;cputime-t

there is indeed a factor 10 for ATLAS

but for

R=0;x=rand(512,1)+j*rand(512,1);
t=cputime;R=R+x*x';;cputime-t

I have lost some time!

normal octave gives .3 sec
Altas ocatve  gives .4 sec

Can you confirm this and do you have an idea why  this happens?

Thanks

______________________________________________
FREE Personalized Email at Mail.com
Sign up at http://www.mail.com?sr=mc.mk.mcm.tag001



-----------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.che.wisc.edu/octave/octave.html
How to fund new projects:  http://www.che.wisc.edu/octave/funding.html
Subscription information:  http://www.che.wisc.edu/octave/archive.html
-----------------------------------------------------------------------