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

WRF Compilation: High Performance Computting

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.

DanielaMonterde

New member
Dear all,

I am trying to run WRF in a High Performance Computting. I installed WRF using the method on the next website: http://www2.mmm.ucar.edu/wrf/OnLineTutorial/compilation_tutorial.php
with the difference that I'm using MVAPICH2 instead of MPICH. But I am experimenting some problems when initializing the model, in the communication of the nodes. The computer communities using InfiniBand.

The message on my rsl.error is the next one:

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
taskid: 30 hostname: CNS-NODE031
module_io_quilt_old.F 2931 F
Quilting with 1 groups of 0 I/O tasks.
~
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

The characteristics of the cluster are the following:

+ 257.54 TFLOPS double presition
+ Nodes: 86 INTEL XEON X86 BITS SKYLAKE (Purley Platform)
+ Storage: Lustre system with 1.8 Tb available

I only have access to 6 nodes with 32 cores per node and 198 Gb of RAM

I think that the problems may be asociated with my compilation form and I would appreciate if you could help me to make a better compilation of WRF in the cluster.

Thank you very much for your help.
 
Please turn off the quilt option in your namelist, i.e.,

nio_tasks_per_group = 0,
nio_groups = 1,

Then try again. I hope this can fix the problem you have, If not, please let us know.

Note that the quilt option doesn't work well in Lustre system.
 
Note that the quilt option doesn't work well in Lustre system.

Is this still true? I am using WRF 4.1.2 on a lustre file system and I have not been able to get quilting working properly. Sometimes the wrf.exe run will produce the SUCCESS message, but wrf.exe crashes after this and the wrfrst files at the end time will not be written properly. Other times, depending on the quilting configuration chosen, wrf.exe crashes when writing wrfrst files at intermediate times within the run.
 
Top