[MITgcm-support] Memory usage keeps increasing

Shantong Sun silencesst at hotmail.com
Fri Jul 17 13:59:14 EDT 2015

Thanks for your response, Martin.
The problem is solved. Probably this is related to the version of openmpi. Just in case someone also has this problem, here is my move to have this resolved.
(1) I removed all the packages related to mpi, hdf, and netcdf etc.(2) I reinstall openmpi (previously using 1.8.6, I changed to 1.8.7)(3) reinstall zlib, hdf5, netcdf-c, and netcdf-fortran (no change with the version num).
I am inclined to think it is the version of openmpi issue though it could also be something else since the mpi version is not the only difference before and after.

> Date: Fri, 17 Jul 2015 11:31:47 +0200
> From: Martin Losch <Martin.Losch at awi.de>
> To: MITgcm Support <mitgcm-support at mitgcm.org>
> Subject: Re: [MITgcm-support] Memory usage keeps increasing
> Message-ID: <7186AD18-9958-421A-909B-5E0B2E960670 at awi.de>
> Content-Type: text/plain; charset="utf-8"
> Hi Shantong,
> as the MITgcm is (mostly) written in Fortran 77, there is no dynamic allocation of memory in the code (except for parts of the ptracers code), and as far as I can see, the increase in memory usage with time cannot have anything to do with the MITgcm itself.
> Having said that, from my experience operating systems sometimes don?t allocate all memory in the beginning of the run, but only when it is really used. E.g., if you use the diagnostics package, then the memory associate with this package does not show up in ?top? until it is actually used, so that can lead to an increase in memory some time after starting the model, but this is not a continuous increase, but a one time step up (probably at the time the diagnostic fields are written for the first time or similar. Honestly, I don?t know how that works exactly).
> Martin
> > On 16 Jul 2015, at 19:38, Shantong Sun <silencesst at hotmail.com> wrote:
> > 
> > Hi all,
> > 
> > I am running MITgcm with 2 cores (using openmpi) on a local server. Everything is good except that the memory usage of 'mitgcmuv' keeps increasing. As an example, the mem usage was less than 1% when I started it, but it increases continuously to 17% after 2 hours. It seems to continue to rise and is likely to blow up finally..
> > 
> > Is there anyone having this same problem? How do you resolve it?
> > 
> > Thanks,
> > Shantong
> > UCSD
> > _______________________________________________
> > MITgcm-support mailing list
> > MITgcm-support at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-support
> ------------------------------
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
> End of MITgcm-support Digest, Vol 145, Issue 9
> **********************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20150717/3a28c5c7/attachment.htm>

More information about the MITgcm-support mailing list