GETM memory usage

[Büchmann Engineering]

Hi,

I have problems running getm in simple debug+serial mode for a
moderately big setup (about 850 points square).
I just want to run a few time steps if possible, before I create a
parallel setup.

Does anybody have a current memory estimate for running GETM in serial?

Something like an estimates for the number of 2D and 3D arrays would be
excellent.

Sorry if this is already in the docs somewhere.

Thanks!

-Bjarne

Büchmann Engineering.
http://eng.buchmann.dk/

[kb]

Hello

On 3 Jan., 13:39, Büchmann Engineering <e...@buchmann.dk> wrote:
> Hi,
>
> I have problems running getm in simple debug+serial mode for a
> moderately big setup (about 850 points square).
> I just want to run a few time steps if possible, before I create a

> paallel setup.

>
> Does anybody have a current memory estimate for running GETM in serial?
>
> Something like an estimates for the number of 2D and 3D arrays would be
> excellent.
>

Depending on compilation options - mainly higher order schemes for
velocity advection the number of static 3D arrays varies. Using around
40 would be a good guess. The log-file actually writes an estimate of
the static memory used for 2D and 3D fields.

If you have many layers 850x850xmanyx8x40 will give many MBs

> Sorry if this is already in the docs somewhere.
>

> Tyhanks!
>
> -Bjarne

Karsten