This sounds like memory badness is occurring somewhere in your
application which eventually corrupts things to make them stop working
(e.g., writing beyond the end of arrays, etc.). Have you run your app
through a memory-checking debugger, perchance?
On May 14, 2009, at 1:00 AM, Valmor de Almeida wrote:
> I am wondering whether light oversubscription could lead to a
> program. The particular case is a fortran 77 (for the most part)
> code I
> am working with that can only run on powers of 2 processes (starting
> with power 1). When I run the program on my single-processor laptop,
> shows some erratic behavior, namely the fortran system calls that
> directories and rename files, etc. do not work after some point in the
> program. I only use -np 2 and the hostfile is
> 'hostname' slots=1 max_slots=1
> Changing the dimension of some larger arrays in the program do affect
> the point where the system calls do not work. The status of the mpi
> calls is normal; no errors reported. I am using gfortran 4.3.2 and
> The program runs normally on my workstation with 4 processors with the
> same compiler/ompi versions.
> Thanks for any inputs.
> users mailing list