Hi,
My application uses MPI_Comm_spawn to periodically spawn a group of
processes to perform some tasks. On one of our linux clusters the
program hung up on the MPI_Comm_spawn call after spawning a certain
number of processes. But the same program works fine on another
cluster.
The cluster that had problem with MPI_Comm_spawn is equipped with dual
Intel Xeon Processors per node. The job scheduling system is SGE (Sun
Grid Engine). I am using lam-7.1.2, gcc 3.4.4 compiler. Could anybody
help me with this or give some hints on how to pin point the problem?
Appreciate it!
Thanks,
Heshan
|
