[Bioclusters] MPI clustalw

Tim Cutts bioclusters@bioinformatics.org
Sun, 9 May 2004 19:10:45 -0300

On 6 May 2004, at 10:30 pm, James Cuff wrote:

> Guy/Tim - did you ever deploy that HMMer PVM cluster we talked about 
> for the Pfam web site?

That's still a watch this space.  The PFAMers are still talking about 
doing some MPI/PVM stuff.  Like you, I am still a sceptic.

Having said that, some of the EBI chaps are running MPI clustalw -- not 
sure which one -- on the IBM blades, and claim to be getting decent 
performance out of it.  That's GBit interconnect, as you know, and we 
limit them to 10 CPUs (i.e. 5 machines) per job.  Haven't got as far as 
getting it to schedule sensibly with the network topology.  LSF has a 
useful looking sc_topology.so mbschd plugin, but I've not found any 
docs for it yet.


Dr Tim Cutts
Informatics Systems Group
Wellcome Trust Sanger Institute
Hinxton, Cambridge, CB10 1SA, UK