[Beowulf] 1 multicore machine cluster

Glen Beane Glen.Beane at jax.org
Fri Apr 24 05:19:36 PDT 2009




On 4/24/09 8:07 AM, "Jonathan Aquilina" <eagles051387 at gmail.com> wrote:

in a way arent multicore processors taking for example 4 single core
machines and merging them into one with 1 for core processor?


I think the term "node" is a loaded term in HPC.  This is what comes to mind when I hear node, and I'm sure a lot of other people think the same:  A node is a physical building block of a cluster.  It has an operating system, it probably has some kind of daemon running to launch jobs submitted to a batch system.  It is managed as a single unit, but may contain many processors & cores.  I don't think anyone in the HPC field would say that a cluster of 128 systems with 8 cores per system is a 1024 node cluster. They would say it is a 128 node cluster with 1024 cores. It has 128 operating system instances running, 128 pbs_moms, etc.  Calling it a 1024 node cluster is misleading.  Much of the management here tend to confuse node with core, but we're a genetics research laboratory and most people don't have a background in this.


On 4/24/09, Glen Beane <Glen.Beane at jax.org> wrote:
>
>
>
> On 4/24/09 3:03 AM, "Jonathan Aquilina" <eagles051387 at gmail.com> wrote:
>
> im impressed with the different views everyone has. i dont know how many of
> you would agree with me a multicore processor lets say a quad is 4 nodes in
> one. could one say it like that?
>
> I would not.  To me a node is a physical thing.  One or more processor(s),
> RAM, running its own OS instance.  I have a cluster of N nodes with M cores
> per node, or N*M total cores.
>
> In your example, N is 1 and M is 4.
>
> One thing that annoys me is when an intern working on an MPI program keeps
> saying "node" when they really mean MPI rank since we are long past the days
> where we have a 1 to 1 mapping between MPI ranks and nodes (we don't do any
> kind of hybrid thing where we have 1 multi-threaded process per node).
>
>
>
> --
> Glen L. Beane
> Software Engineer
> The Jackson Laboratory
> Phone (207) 288-6153
>
>


--
Jonathan Aquilina


--
Glen L. Beane
Software Engineer
The Jackson Laboratory
Phone (207) 288-6153

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20090424/4a54a248/attachment.html>


More information about the Beowulf mailing list