WRF RUN ERROR: Program received signal SIGSEGV: Segmentation fault - invalid memory reference

[lixi202404]

I'm running WRF452 on the Linux subsystem of my own Windows 11 system in my pc to simulate a typhoon. When I use ONE processor, wrf.exe runs successfully but slowly. However, when I use 16 processors for fast running speed, wrf.exe runs several hours and output 4 hours wrf_out* data, then it collapses with the following error:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Namelist.input was attached. Any help or thoughts would be appreciated.

 

[William.Hatheway]

I'm running WRF452 on the Linux subsystem of my own Windows 11 system in my pc to simulate a typhoon. When I use ONE processor, wrf.exe runs successfully but slowly. However, when I use 16 processors for fast running speed, wrf.exe runs several hours and output 4 hours wrf_out* data, then it collapses with the following error:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Namelist.input was attached. Any help or thoughts would be appreciated.

Can you attach all the rsl.out and rsl.error files located in the /run folder as a zip file?

 

[Ming Chen]

Your namelist.input options related to WRF-ARW physics look fine to me except that I would suggest you set radt = 5. which is consistent with your dx = 5000.

I am not sure of the chemistry options used in this case. Hope someone in the wrf-chem community might be able to confirm they are right.

Since this case can run successfully with a single processor but failed when using 4 processors, I am suspicious there might have some issues of communication between processors. This sounds more like a machine problem.

 

[William.Hatheway]

I'm running WRF452 on the Linux subsystem of my own Windows 11 system in my pc to simulate a typhoon. When I use ONE processor, wrf.exe runs successfully but slowly. However, when I use 16 processors for fast running speed, wrf.exe runs several hours and output 4 hours wrf_out* data, then it collapses with the following error:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Namelist.input was attached. Any help or thoughts would be appreciated.

try setting this command before running the cases

ulimit -s unlimited

 

[lixi202404]

Can you attach all the rsl.out and rsl.error files located in the /run folder as a zip file?

 

[lixi202404]

try setting this command before running the cases

ulimit -s unlimited

ulimit -s unlimited has been set and have the same problem. Timestep has been set 30, 25, 20, 15 respedtively and have the same problem.

 

[lixi202404]

I am used my own PC to do the simulatation. When I compile WRF I use smpar mode, GCC compile.

 

[lixi202404]

Your namelist.input options related to WRF-ARW physics look fine to me except that I would suggest you set radt = 5. which is consistent with your dx = 5000.

I am not sure of the chemistry options used in this case. Hope someone in the wrf-chem community might be able to confirm they are right.

Since this case can run successfully with a single processor but failed when using 4 processors, I am suspicious there might have some issues of communication between processors. This sounds more like a machine problem.

By setting radt=5 there is still an error as the pictures attached.

 

[lixi202404]

The log file is 11.8G, which is too large to upload. I attached the fiirst part and another part of the log file.

 

[Ming Chen]

I never see messages like below:

****
   8********************************
   9********************************
  10********************************

And your log file is filled with such kind of message. I am not sure whether they are related to chemistry.

Sorry that I don't have an immediate answer to your question....