Another quick (I hope) question:
I have a relatively new MCMC code that shows good scaling only when I
use -lamd. I see nearly 100% load per cpu using lamd and only 10% per
cpu using c2c. MCMC is embarassing parallel so I would expect nearly
linear scaling for quite a large N.
Over the last 3 years or so on other codes, I've only seen
improvements using c2c over lamd. The machines are dual PIII running
Linux 2.4.16 kernel and Lam 6.5.6 (same behavior with Lam 6.5.3).
This is not a problem but I'd like to understand what is going on.
Any comments? Is it a Linux SMP issue with kernel 2.4.x does anybody
know (e.g. a spin-lock contention)?
Thanks!
_______________________________________________
This list is archived at http://www.lam-mpi.org/MailArchives/lam/
|
