Stream variation

[Greg Lindahl]

On Thu, Apr 26, 2001 at 09:37:15AM -0700, Patrick Geoffray wrote:

> Running 2 instances of Stream at the same time on several Netfinity x330
> Dual PIII boxes, I see a variation of performance from one node to another
> one, from 185 MB/s to 260 MB/s per run for TRIAD for example.

Running 2 instances of stream? Well, that depends on the details. You
see, stream takes the best time and uses that to give the answer. So
if once the 2 instances begin running at exactly the same time, you
get one answer, and if the 2 instances start at different times so
that each one gets a full cycle without the other one going, you get a
much better answer. If this effect is getting you, then the results
shouldn't be repeatable on an individual machine.

> Have you seen so large variation between identical nodes in a cluster
> before ? Do you have any ideas ? (temperature, cheesy chipset, ghost in
> the machine, ...)

I've never seen variations so large. But I usually run one instance
that runs for a very long time (10x normal), and use the time for a
second normal instance. That's a lot more stable for the problem I
described.

-- g