ssfate@gmail_com
New Member
I have successfully compiled CLM3.5 in a Linux cluster with both PGI and Intel compilers. I have turned on the SPMD flag during the compliation. The sample case runs smooth in a single processor.
But when I submit the job to a MPI run with bsub and mpirun, the computational speed seems doesn't increase with increasing the number of processor. It does output all the hiistory file. But it always reported "1 pes participating in computation" in the output message no matter I use any number of processor. The involved processors are actually working but the domain seems not decomposited. It seems all the processors were working on the same domian.
Any ideas on this probelm?
How can I check CLM3.5 is really running in a parallel mode but not all the processors working on the same grids?
But when I submit the job to a MPI run with bsub and mpirun, the computational speed seems doesn't increase with increasing the number of processor. It does output all the hiistory file. But it always reported "1 pes participating in computation" in the output message no matter I use any number of processor. The involved processors are actually working but the domain seems not decomposited. It seems all the processors were working on the same domian.
Any ideas on this probelm?
How can I check CLM3.5 is really running in a parallel mode but not all the processors working on the same grids?