[Beowulf] Cluster Benchmarks?

Greg Lindahl lindahl at pathscale.com
Tue Jun 15 22:29:14 PDT 2004


On Tue, Jun 15, 2004 at 09:25:21PM -0700, Bill Broadley wrote:

> I'd argue, that yes, the world needs more good microbenchmarks.  Especially
> in the area of performance scaling with array size, number of threads, and
> number of nodes.  

Unfortunately, most of the microbenchmarks that people think are
"good" actually suck. flops.c is one of those. An example that deserves
to be in the good category is Stream.

> Sure, real world application performance is king, but that doesn't mean
> that microbenchmarks are not justified.

Good thing I wasn't arguing that.

-- greg




More information about the Beowulf mailing list