[MITgcm-support] bluesky build

[Constantinos Evangelinos]

On Monday 13 June 2005 09:21, Patrick Heimbach wrote:

> Mike,
>
> I comiled one of our standard verif. experiments global_ocean.90x40x15
> on bluesky using 4 processors last week.
> You may find the build at
> ~pheimbac/ecco/MITgcm_latest/verification/global_ocean.90x40x15/build
> I used optfile 'sp4', not 'sp4+mpi'
> (both are MPI and this needs some cleanup from our side).
>
> I ran the program in batch using the script
> ~pheimbac/ecco/MITgcm_latest/jobs/global_ocean.csh
> and it went through OK.
>
> Might be worth trying whether you can reproduce this.

I just used sp4+mpi and exp2 and it also worked fine. I've just started a 
complete testreport as well to see if any problem exists elsewhere.

sp4+mpi is using 64-bit compilation and the 64-bit MPI libraries and could be 
faster even for problem sizes that do not need the 64-bit addressing if the 
mroe heavily optimized version of the MPI libraries is indeed the 64-bit one 
as hinted to by IBM docs.

Constantinos
-- 
Dr. Constantinos Evangelinos
Department of Earth, Atmospheric and Planetary Sciences
Massachusetts Institute of Technology