[CSC 435] Re: Matrix Multiplication

[Andrew J. Pounds]

No.  Your walltime is what it is. If you did it on a per-thread or 
per-processor basis you would be approaching the cputime (not the 
walltime).   The rule of thumb is that once you move to multiple threads 
or multiple processors, you start using walltime as you basis for all 
measurements.




> Dr. Pounds,
> Whenever we do our testing for our multi-threaded Baby BLAS, should we 
> treat our walltimes as exactly as they are, or should we divide that 
> number by the number of threads? Also, is this also the same with the 
> MFLOPS? Should we scale them up or down?

-- 
Andrew J. Pounds, Ph.D.  (pounds at theochem.mercer.edu)
Associate Professor of Chemistry and Computer Science
Mercer University,  Macon, GA 31207   (478) 301-5627