Re: [OMPI users] Excessive Use of CPU System Resources with OpenMPI 1.2.4 using TCP only ..

On Tue, 2008-01-22 at 20:19 +0100, Pignot Geoffroy wrote: > You could try the following MCA setting in your mpirun command > --mca mpi_yield_when_idle 1 Yes, but to repeat what was said above, it is first essential that you read:

Re: [OMPI users] Excessive Use of CPU System Resources with OpenMPI 1.2.4 using TCP only ..

You could try the following MCA setting in your mpirun command --mca mpi_yield_when_idle 1

Re: [OMPI users] Excessive Use of CPU System Resources with OpenMPI 1.2.4 using TCP only ..

See this FAQ: http://icl.cs.utk.edu/open-mpi/faq/?category=running#oversubscribing OMPI by default assumes that each process has its own cpu, thus aggressively looks for data to receive. I think there is a way to yield the processor if you want, but I think what your seeing is behavior yo

Re: [OMPI users] Excessive Use of CPU System Resources with OpenMPI 1.2.4 using TCP only ..

Well, I have noticed that when a process is waiting for communication from another process the reported CPU usage remains around 100%. Is that what you mean? I haven't explored whether these processes give way to other active processes under the linux scheduler, nor whether I should expect anythi

[OMPI users] Excessive Use of CPU System Resources with OpenMPI 1.2.4 using TCP only ..

We've observed an excessive use of CPU system resources with OpenMPI 1.2.4 using TCP connections only on our SL5 x86_64 Cluster. Typically, for a simple Canonical Ring Program, we're seeing between 30 and 70% system usage. Has anybody else noticed this sort of behaviour? And does anybody have some