Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] Mimicking timeout for MPI_Wait
From: Eugene Loh (Eugene.Loh_at_[hidden])
Date: 2009-12-06 21:42:55

Douglas Guptill wrote:
On Sun, Dec 06, 2009 at 02:29:01PM +0200, Katz, Jacob wrote:
Yes, I meant in the question that I was looking for something creative, both fast responding and not using 100% CPU all the time.
I guess I’m not the first one to face this question. Have anyone done anything “better” than the simple solution?
I got tired of seeing both cpus at 100% load, and based on suggestions
from Jeff Squyres and Eugene Loh, wrote MPI_Recv.c and MPI_Send.c.  I
load these with my application, and bingo!  Only one cpu busy at any
given time.

They use a graduated sleep; the first sleep is short, the second is
twice as long, and so on, up to a maximum sleep time.
Another option is to use a condition variable -- e.g., pthread_cond_wait() and the like.  So, there is a certain wake-up time, but you don't need to tune sleep times or wait for some long sleep time, etc.

There is no magic bullet here.  It's a matter of tuning various trade-offs.  Some bullets are better than others, but none of them is magic.  :^)