On Tue, 13 Mar 2007, Glen Beane wrote:

> I agree that we should add a '-l cpus=x'  option to torque

[catching up again]

Thinking further what we could really do with is a way to tell Maui & Moab to 
do the right thing, rather than extending the RM (and then extending the 
scheduler to cope).

Perhaps add another JOBNODEMATCHPOLICY option, perhaps ALMOSTEXACTNODE ?

Then a job that asks for nodes=N get N cpus and a job that asks nodes=N:ppn=x 
gets what it asked for.

cheers,
Chris
-- 
 Christopher Samuel - (03)9925 4751 - VPAC Deputy Systems Manager
 Victorian Partnership for Advanced Computing http://www.vpac.org/
 Bldg 91, 110 Victoria Street, Carlton South, VIC 3053, Australia

Attachment: pgpVAFQc7i52X.pgp
Description: PGP signature

_______________________________________________
mauiusers mailing list
[email protected]
http://www.supercluster.org/mailman/listinfo/mauiusers

Reply via email to