Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Problem with parallel execution wrf4.2.1

suso

New member
Hi, I am running WRF4.2.1 through the wrf4g v2 framework (Home).

After many tests and tuning of the parallelization environment, I could run in parallel within a specific domain. Nevertheless, when changing the domain to a different area (though the dimensions are comparable), the parallel execution does not work: the simulation is run in series, and the very first lines of the rsl.out are as follows:

taskid: 0 hostname: node040
Quilting with 1 groups of 0 I/O tasks.
Ntasks in X 1 , ntasks in Y 1

This happens for every value of cores and the number of cores per node I specify within the experiment.wrf4g file, which is attached together with the rsl files. In this case, it was for example 4 cores in the same node of our Neptuno cluster infrastructure, in the MPIRUN environment.

I am getting crazy about this, as I suspect it could be something related to the specific shape of the domain, but I don't see any rule that I could be possibly infringing. Any help would be very much appreciated.

Saludos

S.S.
 

Attachments

  • parallelError.zip
    7.9 KB · Views: 0
How did you compile WRF? I would suggest you recompile WRF in damper mode, then run wrf.exe using its automatic decomposition capability, i..e.,
you don't need to specify nproc_x, nproc_y . You can run the code by, for example,

mpirun -np 16 ./wrf.exe

Here 16 indicates the number of processors you intend to use to run the case.

Note that the command might be different depending on the machine you have.
 
Top