Failing to build WRF v4.7.1 (and v4.6.1) on Derecho using cmake

[Yongxin]

I downloaded WRF v4.7.1 (and v4.6.1) using the command:
git clone --recurse-submodule --branch v4.7.1 GitHub - wrf-model/WRF: The official repository for the Weather Research and Forecasting (WRF) model

and did the following:
module --force purge
module load ncarenv/24.12
module reset
module load cmake
module list
Currently Loaded Modules:
1) ncarenv/24.12 (S) 5) libfabric/1.15.2.0 9) ncl/6.6.2
2) craype/2.7.31 6) cray-mpich/8.1.29 10) ncview/2.1.9
3) intel/2024.2.1 7) hdf5/1.12.3 11) nco/5.3.1
4) ncarcompilers/1.0.0 8) netcdf/4.9.2 12) cmake/3.26.6

and then issued:
cd WRF
./configure_new

which worked fine and lastly I issued:
./compile_new >& compile01.log &

Somehow the build fails after some time and the place where the build fails is different if I issue "./compile_new" immediately. My compile01.log file is attached here. I am wondering if anyone has any suggestions on how to fix this issue.

Thanks,
Yongxin

 

[Ming Chen]

Hi Yongxin,

When you logon to derecho, please issue the command:

module load gcc

Then issue ./clean -a in top WRF directory, and run ./configure, choose option 34 (gnu dmpar mode), and compile WRF.

Please save the log file for me to take a look if the compiling fails again.

 

[islas]

This issue will most likely not arise with the gcc configuration, regardless of using make or cmake (old ./configure or new ./configure_new, respectively).

I tentatively suspect this is due to the high memory usage of ifx (Intel oneAPI Fortran compiler). See this thread for a running history on the issue:

Compiling WRF-4.5.2 with Intel OneAPI on Derecho

I'm trying to compile WRF-4.5.2 on Derecho using the Intel OneAPI compilers. (I've already gotten it to compile fine using the intel-classic/2023.2.1 compilers, but I'm wondering why OneAPI isn't working.) I'm getting very similar compilation failures whether I use intel-oneapi/2023.2.1 or...

forum.mmm.ucar.edu

If you need to run with the Intel compilers, the recommendation is to compile using a compute job on derecho as those give you more memory than the login nodes.