Athlon memory speed asymmetry

Joe Landman landman at scalableinformatics.com
Tue Feb 25 14:27:49 PST 2003


Hi Robert:

I am sure someone has pointed it out to you, but ...

On Tue, 2003-02-25 at 17:02, Robert G. Brown wrote:

>  for (i=0; i<size; i+=stride){
>    aindex = ai[i];
>    aitmp = ai[aindex];
>    ai[aindex] = aitmp;
>  }

This loop (the above) is not likely to be optimized away ....

> less the time required for (separately evaluated):
> 
>  for (i=0; i<size; i+=stride){
>    aindex = ai[i];
>  }

But this one is.  

What compiler and optimization level do you use, and did you look at the
disassembled binary to verify that these loops are represented as you
wrote them?

Joe

-- 
Joseph Landman, Ph.D
Scalable Informatics LLC,
email: landman at scalableinformatics.com
web  : http://scalableinformatics.com
phone: +1 734 612 4615




More information about the Beowulf mailing list