REASOR <dareas0_at_[hidden]> wrote:
> I currently run lam-mpi on small commodity clusters. I use lam-mpi with ifort 9.1.3 for compiling and running fortran 90 fluid dynamics codes.
> I have been running on AMD 2500+ (32bit), P4 (32bit), and AMD 3200+ (64bit) nodes single core without any problems, but when I upgraded to running on AMD 4600+ dual core nodes and Core2Duo e6400 dual core nodes I get the error below after an arbitrary number of iterations. Has anyone had any similar problems using lam-mpi with ifort on these processors? If so have they had any luck addressing the problem?
With Intel compilers you need to worry about the stack size. Use:
ulimit -s unlimited # bash shell
limit stacksize unlimited # tcsh shell
You may need to replace "unlimited" by a number such as 65000.
--
Ole Holm Nielsen
Department of Physics, Technical University of Denmark
|
