[MITgcm-support] Memory estimates
Christopher L. Wolfe
clwolfe at ucsd.edu
Mon Oct 13 12:54:16 EDT 2008
Also, since the MITgcm uses mostly statically allocated memory,
running size -A mitgcmuv on the compiled executable will tell you the
amount of memory (minus system overhead) the job will require per
process.
Christopher
-----------------------------------------------------------
Dr. Christopher L. Wolfe 858-534-4560
Physical Oceanography Research Division OAR 357
Scripps Institution of Oceanography, UCSD clwolfe at ucsd.edu
-----------------------------------------------------------
On Oct 13, 2008, at 8:15 AM, Gøran Brostrøm wrote:
> Hi David
> An old formula is that memory=nx*ny*nz*8 *50. That is the model uses
> about 50 double prec fields. If you use timeave it is probably
> somewhat higher. Perhaps the newer model versions are also more
> craving on memory. You can always make a relatively small experiment
> and look on how much memory is used (top probably gives you the
> memory usage) to find the no of fields the model uses.
> gøran
>
>
> David Hebert wrote:
>> Hello all,
>>
>> Is there a common rule of thumb for estimating the amount of memory
>> a simulation will use? For example, if there are 9 output fields
>> (U,V,W,T,S,PH,PNH,hFacC,hFacW, perhaps others) * 2 for AB
>> timestepping, could one say there are 18 fields in memory. Then
>> knowing the size of the fields you can get a good estimate on the
>> memory required, and then based on node/processor memory number of
>> processors one should use.
>>
>> Thanks,
>>
>> David
>>
>> --
>> David A. Hebert
>> NRC Postdoc
>> Naval Research Lab
>> Stennis Space Center, MS 39525
>> david.hebert.ctr at nrlssc.navy.mil <mailto:david.hebert.ctr at nrlssc.navy.mil
>> >
>> Phone: (228) 688-5846
>> Fax: (228) 588-5997
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> 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
More information about the MITgcm-support
mailing list