On Tue, 28 Feb 2006, Frans Verster wrote:
> All clusters I (may) have access to have MPICH installed and configured,
> but no LAM/MPI, so my question is: How can I install LAM within a job
> (PBS or JDL) and configure it just as MPICH.
We don't know how MPICH is configured on these clusters, so we can't
suggest how to configure LAM/MPI. The sysadmins of these clusters who
did the MPICH installation could probably give you this information,
f.e. what compiler was used, what extra libraries (like GM), etc.
> Eventually I'd like to submit a job which first installs LAM
This is a strange (although not unreasonable iff you know that the
compute nodes also contain complete compilers and libraries
installations, which is not always the case) way of installing
LAM/MPI. The usual installation happens on (one of the) access node(s)
and the final directory tree is shared via NFS to all nodes.
Another solution is to have LAM/MPI installed as a package (rpm, deb,
etc.) on each node of the cluster by the sysadmin.
> and last but not least does the real work.
This is the only part that is typically done via a queueing system
like PBS.
I would suggest first reading the FAQ on the LAM/MPI website.
--
Bogdan Costescu
IWR - Interdisziplinaeres Zentrum fuer Wissenschaftliches Rechnen
Universitaet Heidelberg, INF 368, D-69120 Heidelberg, GERMANY
Telephone: +49 6221 54 8869, Telefax: +49 6221 54 8868
E-mail: Bogdan.Costescu_at_[hidden]
|
