Community > Discussion Forum > Cluster job that shows speed-up when using multiple nodes

Discussion Forum

Cluster job that shows speed-up when using multiple nodes

Topics: Solving, 4.2a

Thread index  |  Previous thread  |  Next thread  |  Start a new discussion

RSS FeedRSS feed   |   Email notificationsTurn on email notifications   |   2 Replies   Last post: January 29, 2013 9:59pm UTC
Josh Thomas

Josh Thomas
Certified Consultant
AltaSim Technologies, LLC

May 16, 2012 8:51pm UTC

Cluster job that shows speed-up when using multiple nodes

Hi,

I would be highly interested if someone could post an mph model file that they have shown to see speed-up when distributed across multiple nodes. I am having difficulty seeing any speed up for a number of models and understand from previous threads that speed-up is highly model dependent.

If someone could post an mph file that they know to show speed-up it would be greatly appreciated.

Best regards,
Josh Thomas

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

STS

STS
COMSOL Employee
France

October 4, 2012 5:14pm UTC in response to Josh Thomas

Re: Cluster job that shows speed-up when using multiple nodes

Dear Josh,

unfortunately I have not read your post earlier: hope it's not too late!
The models that you have tested are provided to test proper operations of the cluster, not performance. For performance, you'd better try to reproduce the results of the following paper, based upon models also available in the model library, for the same number of DOFs: www.comsol.fr/papers/10248/

Best regards,
Stephan

--
www.comsol.fr

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Josh Thomas

Josh Thomas
Certified Consultant
AltaSim Technologies, LLC

January 29, 2013 9:59pm UTC in response to STS

Re: Cluster job that shows speed-up when using multiple nodes

Stephan,

Thanks for the reply. I did have a chance to review Freels work and get comparable speed-up results for some of my own large memory models (upwards of 2.5 MDoF). Part of the issue was getting our HPC system interfacing properly with the COMSOL batch commands and PBS scheduler for optimal advantage of the cluster. The other part related to what you brought up about using a good test model. I've seen that until my models go beyond 1 or 2 MDoF that the speedup is minimal across clusters. But, in those cases, my workstation works great for these anyways!

Thanks again for the reply,
Josh

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse


Rules and guidelines