[Beowulf] newbie

[Chris Samuel]

----- "Gus Correa" <gus at ldeo.columbia.edu> wrote:

> Hi Chris, list

Hiya,

> Thanks for the info!

Not a problem!

> I have been blindly using "-march=amdfam10" (and -O3)
> out of the gcc 4.1.2 man page, simply because it is
> recommended for Barcelona (and hopefully also the
> Shanghais we have).

Well I did a microbenchmark beforehand (actually one from
Joe Landman's blog) and found that the Barcelona extensions
improved performance for it:

http://www.csamuel.org/2009/03/15/comparing-fortran-compilers

but that doesn't seem to carry over into the code our
users are trying with it (possibly because they're not
doing Fortran).

So I would strongly suggest benchmarking your code both
with and without that option and see what works best for
you!

> What would be the right level of optimization then?

As ever, the answer is "it depends". :-)

Experiment and read the info pages for the optimisation
flags in GCC..

[Intel and Shanghai]
> Is this deliberate?

In the sense that they have no desire to support
competitors hardware, yes. Not really surprising,
if AMD made compilers I doubt they'd try and do
Intel specific optimisations either..

cheers!
Chris
-- 
Christopher Samuel - (03) 9925 4751 - Systems Manager
 The Victorian Partnership for Advanced Computing
 P.O. Box 201, Carlton South, VIC 3053, Australia
VPAC is a not-for-profit Registered Research Agency