[Bioclusters] mrbayes + lam-mpi/mpich + grid engine integration

[Tony Travis]

Nick Brockner wrote:

> Hi Tony,
> 
> We are using the Bioteam iNquiry product for our web-based access.  So 
> far, so good.  What type of COTS cluster have you built (number of 
> nodes, OS, hardware)?  We are running an 8 node Apple Xserve G5 cluster 
> with Gig ethernet.

Hello, Nick.

We've got 2 servers + 32 compute nodes running the openMosix kernel 
under RedHat 8.0 at present. All are Athlon XP2400+ with 1GB DDR266. The 
servers share a 1.6TB RAID and the compute nodes PXE boot 'diskless' 
(they have a 40GB local disk for swap and /var/tmp). We've bought 
another 32 @ Athlon XP 2600+ with 1GB DDR333 and will add these soon. 
I'm sure I'm not the only one on this list to encounter difficulties 
scaling up a COTS cluster because of things like the mains supply and 
air conditioning! I don't think we will go beyond 64 compute nodes.

> I am in the process of implementing mrbayes on our cluster.  I have not 
> gotten into the parallel implementation of mrbayes yet, but may do so in 
> the near future.  If you want to access my nexus files, let me know and 
> I will send you the access info directly.  You may have to tweak the 
> files a little bit for mrbayes, as I have seen some issues with these 
> particular nexus files (which were originally made to be used by PAUP).  
> You may have to add some information about the dimensions before mybayes 
> will happily read them.

Yes, I would like to - I've already run some of Derek Gatherer's Nexus 
files using an MPI version of MrBayes that Shane Sturruck 
(shane@bioss.sari.ac.uk) got working for us. It would be interesting to 
compare yours.

Best wishes,

	Tony.
-- 
Dr. A.J.Travis,                     |  mailto:ajt@rri.sari.ac.uk
Rowett Research Institute,          |    http://www.rri.sari.ac.uk/~ajt
Greenburn Road, Bucksburn,          |   phone:+44 (0)1224 712751
Aberdeen AB21 9SB, Scotland, UK.    |     fax:+44 (0)1224 716687