Anyone able to run a 'B_RCP4.5_WACCM_CN' case at f09_g16 resolution?

[bdobbins@gmail_com]

Hi everyone,

I'm trying to track down an issue with a B_RCP4.5_WACCM_CN run with the 1-degree (f09_g16) resolution. I see that it's not an officially supported resolution now, but wondered if anyone has gotten it to work? There must be some sort of bounds issue in the code itself since the failures occur in very non-standard way (eg, failures in compiler libraries rather than in parts of the CESM code, or so it seems). I haven't tried with bounds checking yet.

I'll dig a bit more into it and try to trace where the error is happening, but if this is a really foolish idea since it isn't supported, a few words to that effect would be greatly appreciated.

(Note: The 2-degree works fine.)

Cheers,
- Brian

 

[jedwards]

It sounds like you are describing a compiler failure but you didn't tell us what compiler or machine you are trying to use.

Seems like it may be a foolish idea, there is no initial file for that compset and resolution, so even if you complete the build you are going to continue to have problems.

 

[bdobbins@gmail_com]

Hi Jim,

I actually think it might be a code issue - perhaps with array bounds? I leaned towards a compiler issue in the beginning (Intel 12.1) but switched to PGI and got the exact same kind of error - internal to a memcpy, I believe, which is the strange bit. I haven't yet tried 'gfortran', nor investigated more deeply the issue. I was really just curious if anyone else has run this on any other system, especially x86-based ones.

Also, this happened while running - we were testing via setting RUN_STARTDATE to 2000-02-01 which gives you the 'cami_2000-02-01_0.9x1.25_L66_c040928.nc' file for initial conditions, I think. Is this a very bad idea for some reason?

Thanks,
- Brian