[Beowulf] Need recommendation for a new 512 core Linux cluster

Steven Truong midair77 at gmail.com
Thu Nov 8 16:42:01 PST 2007


Hi, all.   Thank you very much for your inputs and advices.  I
definitely would need to study the nature of my applications a lot
more before we can make any sensible decisions.

Well, our main application is VASP and from what I have learned the IO
is sporadic and VASP seems to me is not IO intensive but CPU bound
scientific app.  We are building this new cluster so we can perform
more VASP calculations.  From my experiences, we have been able to run
many different VASP jobs simultaneously on our Linux cluster with 144
cores with just one NFS server, which exports the home directory to
the computing nodes.  This cluster has gigabit Ethernet with Intel
e1000 driver and 2 Nortel BayStack 24 port Gigabit switches.  If I
recalled correctly then the developers of VASP hinted that CPU
processing powers are more important than the networking fabrics for
running VASP.  I maybe wrong regarding this comments from the
developers.

I also would like to know if anybody has set up AoE to work with HPC
clusters.  We bought a box from Coraid.com for a file server and like
the ways it works.

For queues and jobs management, we have set up Torque and Maui and we
found them to be pretty OK with our needs.

Please share your thoughts on what I wrote and asked.

Thank you very much for all your helps.
Steven.



On Nov 8, 2007 7:52 AM, Eric Thibodeau <kyron at mekkreations.com> wrote:
> There is another product, though not free, that might be of interest for
> file transfers. It's from ExLudus (www.exludus.com) and might be worth
> looking into.
>
>
> Steven Truong wrote:
> > Hi, all.  I would like to know for that many cores, what kind of file
> > system should we go with?  Currently we have a couple of clusters with
> > around 100 cores and NFS seems to be ok but not great.  We definitely
> > need to put in place a parallel file system for this new cluster and I
> > do not know which one I should go with?  Lustre, GFS, PVFS2 or what
> > else?  Could you share your experiences regarding this aspect?
> >
> > I also would like to know how many head nodes should I need to manage
> > jobs and queues.  And what else should I have to worry about?
> >
> > Thank you very much for sharing any experiences.
>
> > _______________________________________________
> > Beowulf mailing list, Beowulf at beowulf.org
> > To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
> >
>
>



More information about the Beowulf mailing list