[Pw_forum] Memory allocation in linux using mpirun

[Axel Kohlmeyer]

On Tue, 19 Jun 2007, Vivian McPhail wrote:


vivian,

VM> Hi,
VM> 
VM> I can successfully run all the examples provided with espresso using mpirun,

even the big water CP-MD examples with cp.x?

VM> however, when I attempt to run my own calculations the program crashes with
VM> a memory allocation failure.  This occurs when real memory is exhausted.

well, your application should _never_ exhaust the physical memory 
and/or go significantly into swap, or else performance will be horrible.

this means, that you either have some serious error in your input
(which we cannot tell, since you didn't post it) or you may simply 
need to use more nodes (whether that will work, we cannot tell, since
you did not tell us which machine you are running on). 

VM> >From examining "top" output, I can see that no virtual memory gets allocated
VM> (swap space is on).

if your application is already heavily swapping, you may not see it
with top in default mode. try 'shift-M' to sort by memory.

cheers,
   axel.

VM> Has anyone had a similar problem?
VM> 
VM> Cheers,
VM> 
VM> Vivian

p.s.: please see stefano's e-mail about adding a signature.
that (and providing more details about your calculations)
will increase your chances of getting help significantly.


-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.