Many of the results can probably be replicated through the Intel MPI
benchmarks, available from:
http://www.intel.com/cd/software/products/asmo-na/eng/cluster/clustertoolkit/219848.htm
Huinalu is a pretty old machine, so my guess is that the results for TCP
will be fairly similar, especially against MPICH-1. For GM peformance,
I'd be surprised if the results were similar, as MPICH-GM has had much
more work put into it than has been put into LAM's GM device.
Of course, we always find performance comparisons interesting :).
Brian
On Tue, 2006-08-29 at 08:21 -1000, William Bierman wrote:
> Ahh, that's what I was afraid of.
>
> This means I have to write something like it on my own. Would someone
> with LAM be interested in the results? The report will be made using
> a machine which was only recently declassified by the DoD.
> http://beowulf.org/archive/2001-March/002559.html
>
> Bill
>
> On 8/29/06, Brian Barrett <brbarret_at_[hidden]> wrote:
> On Mon, 2006-08-28 at 21:39 -1000, William Bierman wrote:
> > I'm wondering if the code used to do the stress testing
> documented in
> > the url below is available in the public domain. Thanks!
> >
> > http://www.lam-mpi.org/papers/perf/
>
> I do not believe it is. That report was written approximately
> 10 years
> ago, while LAM was still developed at Ohio Supercomputing
> Center. It
> has since moved to a different set of maintainers and some of
> the code
> like that was lost along the way.
>
> Brian
>
> _______________________________________________
> This list is archived at
> http://www.lam-mpi.org/MailArchives/lam/
>
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
|
