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

Slow running WRF

This post was from a previous version of the WRF&MPAS-A Support Forum. New replies have been disabled and if you have follow up questions related to this post, then please start a new thread from the forum home page.

D_Fora

New member
Hi,
I find it slow to run WRF real-case in my workstation. Here're details:
WRFV3.8.1 is used to test for a 37-hours real case, which congfigured with 2 one-way nested domains with horizontal grid spacing of 27, 9 km respectively.
Both WPS and WRF are installed in the Linux-subsystem in Window10(workstation with 28cores & 56processors). WRF model is configured with "35(dm+sm) GNU(gfortran/gcc)".
"mpirun -np 16 ./wrf.exe" was used to run WRF. However, it's too slow to run WRF. It there any common or suggestion? Thanks a lot!
View attachment rsl.out.0000.txtView attachment rsl.error.0000.txtView attachment namelist.input
Best regards,
HUANG
 
Yes this case ran very slow. I wonder whether there is special reason you build WRF in dm+sm mode? If possible, can you rebuild WRF in dumper mode, then rerun this case?
 
Ming Chen said:
Yes this case ran very slow. I wonder whether there is special reason you build WRF in dm+sm mode? If possible, can you rebuild WRF in dumper mode, then rerun this case?

Thanks for your suggestion. I have tried dumper mode but it was slow as well as dm+sm mode.
It is worth noting that it run very faster by serial mode than either dm+sm mode or dm mode. So I wonder whether there is any wrong with the Linux subsystem in Windows-10?
Here're the details in serial mode with the same namelist files.
View attachment rsl.error.0000.serial.txtView attachment rsl.out.0000.serial.txt
 
Top