Scheduled Downtime
On Tuesday 24 October 2023 @ 5pm MT the forums will be in read only mode in preparation for the downtime. On Wednesday 25 October 2023 @ 5am MT, this website will be down for maintenance and expected to return online later in the morning.
Normal Operations
The forums are back online with normal operations. If you notice any issues or errors related to the forums, please reach out to help@ucar.edu

pnetcdf on edison.nersc.gov

Using cesm1_2_2 on edison.nersc.gov

> When I set these variables in env_run.xml
> env_run.xml:
> env_run.xml:
> env_run.xml:
> env_run.xml:
>
> I get a segmentation fault. Has anybody tried parallel i.o in cesm with edison? Are these the correct settings?

Without changing these the code runs fine, albeit slow.
Thanks
Michael
 

jedwards

CSEG and Liaisons
Staff member
Hi Michael,Something was lost in your post and the variables you set are missing, but for most compsets on edison PIO_TYPENAME=pnetcdf this is a default value which can be overwritten by the component specific values.   Look for the settings actually used near the top of the cpl.log file.     Please include your cesm.log and cpl.log file along with the compset and grid you are trying to use and I'll see if I can be of more help. 
 
sorry about the lost part of the posthere are the settings that fail entry id="ATM_PIO_STRIDE"   value="24" 
entry id="ATM_PIO_ROOT"   value="0"    
entry id="ATM_PIO_NUMTASKS"   value="128" 
entry id="ATM_PIO_TYPENAME"   value="pnetcdf"


 

jedwards

CSEG and Liaisons
Staff member
Hi Michael,I'm still waiting to see the output logs from your run.   Aliso did you try just setting PIO_TYPENAME=pnetcdf without changing any of the other parameters?   
 
The log file is  /scratch1/scratchdirs/mwehner/seCAM5_test7200x2/run/cesm.log.140818-131355Shows a seg faultPermissions should be readable for that run
 

jedwards

CSEG and Liaisons
Staff member
I'm not sure if this is an IO problem or not - did you also change the pe-layout for this case?    Can you send me yourcase setup including any user_nl or source modifications?   
 

jedwards

CSEG and Liaisons
Staff member
I don't think that it's an IO problem.  It looks like a compiler optimization issue.   Add the following to a file called Depends.edison in your case directory and rebuild: cube_mod.o: cube_mod.F90
    $(FC) -c $(INCLDIR) $(INCS) $(FFLAGS) $(FREEFLAGS) -O0 $<

 
JimI am doing something wrong with the Depends file.I get the error/scratch1/scratchdirs/mwehner/seCAM5_test7200x2_new/Depends.edison:1: *** multiple target patterns.  Stop.

In the log file /scratch1/scratchdirs/mwehner/seCAM5_test7200x2_new/bld/intel/mpt/nodebug/threads/mct.bldlog.140826-172244

this is what I put in the file. it is just one line.more /scratch1/scratchdirs/mwehner/seCAM5_test7200x2_new/Depends.edison
cube_mod.o: cube_mod.F90     $(FC) -c $(INCLDIR) $(INCS) $(FFLAGS) $(FREEFLAGS) -O0 $<

  
 

jedwards

CSEG and Liaisons
Staff member
It's Makefile format, you need a new line which begins with a tab:Look in one of the other Depends files in scripts/ccsm_utils/Machines for an example.
 
Top