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

divide by zero in dead compset

l_vankampenhout@uu_nl

Leo van Kampenhout
Member
Hi,
I'm currently in the process of porting CESM to Cartesius (https://www.surfsara.nl/systems/cartesius/description)

As the user guide indicate, I would like all validation tests to run successfully before spending time on the actual run. However, during the very first test ERS_D.f19_g16.X I get a divide by zero segmentation fault (see attachment). The testcase was created with the following command:
./create_test -compiler intel -mach cartesius -testname ERS_D.f19_g16.X
As indicated in the log, the error occurs in dead_mod.F90:234I've added a print statement just before this location to verify whether variable NXG becomes zero. It really does.

Index: dead_mod.F90

===================================================================

--- dead_mod.F90        (revision 57780)

+++ dead_mod.F90        (working copy)

@@ -230,6 +230,7 @@
    
   write(logunit,*) subname,' Decomp is ',decomp_type,' lsize = ',lsize

   
n=0

+   print *, "DEBUG", nxg, nyg

    dx = 360.0_R8/nxg * deg2rad

    do n = 1,lsize
 
      ig = mod((gindex(n)-1),nxg) + 1

See attachment for the new log (including debug statement)
I'm on the newest CESM 1.2 (rev 57780 of today)Leo
 

jedwards

CSEG and Liaisons
Staff member
Hi Leo,
We recently discovered this problem ourselves and will have a fix for the 1_2_2 release due out this spring.   This affects only the dead models and hopefully you should have more success withother component sets.   We'll update our known problems page and propose a fix soon.   Thanks,Jim
 
Top