Main menu

Navigation

FGEOS_C5SMAM_L40CN compilation error

9 posts / 0 new
Last post
yaoxian.huang@...
FGEOS_C5SMAM_L40CN compilation error

Hi,

 

I would like to run a compset FGEOS_C5SMAM_L40CN using CESM 1.2.2 CAM5.3-Chem. There is no problem for me to configure, setup and compile this compset at a resolution of f19_f19, however, when I changed the resolution to f09_f09, compilation errors occured during the cesm_setup. Is it because the input data,e.g. GEOS5 and other dataset, has not been uploaded to the CESM server? Or other settings that I need to revise? Thank you for providing a hint. I am attaching the compilation error messages as follows.


Creating Macros file for bulldog

/lustre/home/client/apps/geo/Sources/Applications/CESM/1.2.2/scripts/ccsm_utils/Machines/config_compilers.xml intel bulldog

Setting MPI_PATH from Environment

Creating batch script CAM_CRL.run 

Locking file env_mach_pes.xml 

Creating user_nl_xxx files for components and cpl

Running preview_namelist script 

 infile is /home/geo/unger/yh463/CAM_CRL/Buildconf/cplconf/cesm_namelist 

CAM build-namelist - ERROR: No default value found for ncdata

user defined attributes:

key=ic_md  val=20050101

Died at /lustre/home/client/apps/geo/Sources/Applications/CESM/1.2.2/models/atm/cam/bld/build-namelist line 3384.

ERROR: cam.buildnml.csh failed

ERROR: /lustre/home/client/geo/unger/yh463/CAM_CRL/preview_namelists failed: 25344

Y.X.

yaoxian.huang@...

Update from this post:

 

I found a way to solve this problem by adding the ncdata path in user_nl_cam as 

&cam_inparm

ncdata = "FILE LOCATION/FILENAME"

/

Then use ./preview_namelists to inform the ncdata routine. By doing this, ./cesm_setup can be executed without errors. 

Y.X.

krc139@...

Hi,

I am running into a similar problem/same compile error with compset F1850C5. I compiled resolution T31_g37 no problem but I can't compile a higher resolution T85_g16. I don't follow how you set ncdata (i.e. where exactly was that file located?). Can you please explain how you found this to set in user_nl_cam?

Thanks,

Kyle

Kyle Clem
Postdoc
Institute of Earth, Ocean, and Atmospheric Sciences
Rutgers University

jedwards

Unless you have a compeling reason to use the Eulerian dycore you should try f19_f19 or f09_f09 instead of T85.

CESM Software Engineer

krc139@...

Thank you for this helpful suggestion. The f19_f19 resolution is exactly what I want for the atmosphere/land grid. However, I am encountering a run problem for f19_f19, I am attaching the cesm.log for a run with 64 processors/cores. I have two questions if you could please help: 1) for the ocean, what is the difference between f19 and e.g. g16? I am running F compsets, and I am only interested in modifying the prescribed SSTs-could g16 work instead of f19 (i.e. f19_g16)? I did not encounter problems running T31_g37, so I am wondering if, for my purposes, I could just run f19_g16 to increase the atmosphere/land grid resolution and maybe this will avoid my current problems. 2) for the f19_f19, could my errors be related to the number of processors? I tried first with 28 processors and the log contained no information, I then tried with 64 processors and the model seemed to make it further but still crashed. Is there a minimum number or a divisible/combination number needed for an f19_f19 run?

Thank you for your help,

Kyle

Kyle Clem
Postdoc
Institute of Earth, Ocean, and Atmospheric Sciences
Rutgers University

jedwards

You didn't attach any logs.  The model at f19_f19 should run fine on 64 processors and probably aslo on 28 processors. 

The difference between f19_f19 and f19_g16 for an F compset is just when the data is interpolated from g16 to f19.  The f19_f19 case is a little more effiecient and should be fine for your case. 

CESM Software Engineer

krc139@...

Thank you for this. Attached is the log-I'm sorry this didn't get "Displayed"

Kyle Clem
Postdoc
Institute of Earth, Ocean, and Atmospheric Sciences
Rutgers University

jedwards
Here is the error:
FLDLST: DTH in fexcl( 34 ) not found

Do you have a user_nl_cam with an fexcl variable? If so remove that variable and try again.

CESM Software Engineer

krc139@...

Brilliant! I did have many fexcl variables named in the user_nl_cam. I removed those, and it runs-but ONLY on 64 processors. I previously tried a run with no fexcl variables but with 28 processors and it crashed immediately so I did not realize the fexcl was the issue. Any idea on how to find out what the minimum number of processors required is, or better yet, how to set the number of processors? Looks like this may be done in the run script, but I'm not sure...

Thanks,

Kyle

Kyle Clem
Postdoc
Institute of Earth, Ocean, and Atmospheric Sciences
Rutgers University

Log in or register to post comments

Who's new

  • liaogh@...
  • liquanxin2014@...
  • marro.michele89@...
  • sowon@...
  • kiranjadhav1209@...