Hello,
I am a novice when it comes to the workings of MPI and lam.
I am currently working on a high performance cluster which nodes are
managed by the LSF queuing system for which I submit jobs through the
batch mode. This cluster does not have LAM on it and I have installed
LAM in my working directory.
My doubts are:
Can LAM operate in such a fashion, where I submit MPI programs through
the batch mode and the LSF determines on which nodes the programs execute?
If so, what do I specify as the nodes in the hostfile for lamboot? Do I
need to know the names of individual nodes/hosts which make up the cluster?
I could not get a satisfactory answer to my doubt in the documentation.
I am hoping someone from the mailing list can help me.
Thanks a lot in advance,
Nandini Kappiah
|
