Hi all - I'm attempting to set up a solar forecast instance using WRF 4.5.1 and the Deng shallow cumulus parameterization (shcu_physics=5), and I've been getting all kinds of errors. The latest is the "not allocated" error in the title of this posting. Since I'm not using the actual WRF-Solar fork, but the main WRF distribution, I figure the posting probably belongs in this subforum.
I've had some success running the model with WRF 2.2.2, but even with that I still had occasional crashes. So I decided to try the most recent version. I don't know if my problem is that I've got incompatible physics, issues with the compilers or libraries, or something else. In this case, the job is crashing very shortly after it starts up, but I'm pretty confident that I'm not short on memory. I'm using 100 threads, on 4 nodes (25 threads on each), with each node having 128 GB. When monitoring the job (before the crash), no more than 15% of the total memory was used.
I compiled WRF with the debug option, so I have a full traceback. I've attached my namelist.input, rsl.error.000, and rsl.error.0099 (one of the rsl.error files that contains the traceback).
Is it maybe a problem with my physics selections in the namelist? Is there a standard "solar" namelist someplace that I could begin with? Suggestions as to what I can try to get past this?
Thanks,
Mike
I've had some success running the model with WRF 2.2.2, but even with that I still had occasional crashes. So I decided to try the most recent version. I don't know if my problem is that I've got incompatible physics, issues with the compilers or libraries, or something else. In this case, the job is crashing very shortly after it starts up, but I'm pretty confident that I'm not short on memory. I'm using 100 threads, on 4 nodes (25 threads on each), with each node having 128 GB. When monitoring the job (before the crash), no more than 15% of the total memory was used.
I compiled WRF with the debug option, so I have a full traceback. I've attached my namelist.input, rsl.error.000, and rsl.error.0099 (one of the rsl.error files that contains the traceback).
Is it maybe a problem with my physics selections in the namelist? Is there a standard "solar" namelist someplace that I could begin with? Suggestions as to what I can try to get past this?
Thanks,
Mike