[MITgcm-support] building with MPI on a dual-core mac

Klymak Jody jklymak at uvic.ca
Sun Jul 19 13:47:19 EDT 2009


Hi Constantinos,

On 19-Jul-09, at 10:24 AM, Constantinos Evangelinos wrote:

> Quite obviously you
> don't expect it to double performance unless very specific  
> conditions can be
> met but it can be a boost, especially if you're not FP pipeline  
> bound and your
> memory bandwidth is not already consumed by a single thread.

Thanks for the more-knowledgeable explanation.  OS X allows you to  
turn the multithreading off if you desire.  Any idea if that makes  
things go faster if you do not oversubscribe the cores?

> Just to help everyone with quick and dirty benchmarking: if you add  
> the "-ts"
> flag to genmake2 (or even testreport) you get output of user, system  
> and
> wallclock time per timestep as the code runs so you can have a  
> pretty good
> idea of the speed achieved without waiting for the run to finish.

I'm afraid I have to run genmake2 with -ignoretime because of a  
missing library.   I suppose I should track that down. I'm using  
gfortran, gcc, and openMPI, so I'm surprised the timing code isn't in  
there.  I will also have to admit that I'm using a 4 year old version  
of the gcm so maybe the timing calls have changed to be more flexible.

Thanks,  Jody



More information about the MITgcm-support mailing list