WRF only running with one processor

[ephraims]

Hi,
I compiled WRF successfully on the Derecho HPC using option 34 (dmpar), and have been attempting to run an ideal les case using multiple processors. While I have been able to run a test case (100x100 grid points at 100m res), it takes 12 hrs of wall time to complete 40min of simulation time because only one processor is being used (only one rsl.error and rsl.out file is created).

I suspect that a MPI warning I got when running ideal.exe is related to my problem. Attached is a screenshot of that warning. Despite the warning, the last line of the rsl.error file when running ideal.exe is "wrf: SUCCESS COMPLETE IDEAL INIT"

I have also attached a screenshot of my batch script used to run wrf.exe

Any guidance on how to fix this issue is appreciated!

Thanks,
Sam

 

[Ming Chen]

Sam,

Which ideal case did you run?

 

[ephraims]

I ran ideal_les

 

[William.Hatheway]

Hi,
I compiled WRF successfully on the Derecho HPC using option 34 (dmpar), and have been attempting to run an ideal les case using multiple processors. While I have been able to run a test case (100x100 grid points at 100m res), it takes 12 hrs of wall time to complete 40min of simulation time because only one processor is being used (only one rsl.error and rsl.out file is created).

I suspect that a MPI warning I got when running ideal.exe is related to my problem. Attached is a screenshot of that warning. Despite the warning, the last line of the rsl.error file when running ideal.exe is "wrf: SUCCESS COMPLETE IDEAL INIT"

I have also attached a screenshot of my batch script used to run wrf.exe

Any guidance on how to fix this issue is appreciated!

Thanks,
Sam

try mpirun -np (# of cores)

 

[ephraims]

I just tried running wrf.exe using the command you suggested, but it still only used one processor. What is the original warning I got when running ideal.exe? Could that have anything to do with the problem?

 

[ephraims]

Hi, just want to follow up on this thread.

 

[Ming Chen]

Sam,
Can you tell me where your LES case is located? Thanks.

 

[ephraims]

/glade/derecho/scratch/ephraims/WRF_highCCN/test/em_les

 

[Ming Chen]

Hi Sam,
I apologize for the late answer.
I have rerun your case and hopefully I can repeat your issu. However, the error message shows that the model tried to read some data related to radiation. In your namelist.input, i did find that you turn on some physics options like microphysics and radiation. This is not the ideal LES we provide in official WRF release.
Your case looks like a mixture of real-data and ideal case. We don't recommend running WRF in this mode and it is hard to debug what is going on.

 

[ephraims]

Hi Ming,
Yeah, I turned on radiation and microphysics schemes. The radiation isn't critical and is something I mainly did so I can look at cloud top temperature when plotting the cloud field. Changing microphysics scheme is important, as that is something I want to vary in my experiment. I'll try running the case without the radiation scheme and let you know how it goes.