[Beowulf] evaluating FLOPS capacity of our cluster

[Greg Lindahl]

On Mon, May 11, 2009 at 06:58:24PM -0400, Gus Correa wrote:

> How far/close I am now to the maximum performance that can be achieved?

In this case you're only asking about measuring memory. Measure the
curve on your current system. You roughly know the shape of the
curve, so that will allow you to extrapolate.

BTW the various HPL parameters do make a difference, I've seen up to
10%. I watched a team run a grid of 1,000 runs for a short time each
to figure out which parameters were best on a particular maachine.

-- greg