[Nek5000-users] how to choose lx1

[nek5000-users at lists.mcs.anl.gov]

Hi Mehmet,

We've seen nek hang on machines when the number of procs was increased, as
you've just encountered. Sometimes, if you change the parameters in SIZE to
give you some extra buffer in memory, you can get past that sticking point.
So, say you need lelt >= 100. Then, if you have enough memory left on the
proc, compile for lelt=150 (or whatever you can manage). This doesn't
always work, but it's a quick thing to try.

Best,
Justin

On Mon, Sep 24, 2012 at 5:30 PM, <nek5000-users at lists.mcs.anl.gov> wrote:

>
> Hi Justin,
>
> Thanks for your quick email.
>
> I killed the simulation after half an hour later, because it was still on
> the same line.
> I don't think my setup is that big. So I am not sure if I should wait more
> or not.
>
> Best,
> Mehmet
>
>
>
>  >
>
> Hi Mehmet,
>
> How long is nek at the gs_setup line? It can take several minutes for large
> cases.
>
> The last error you're having is because you've run out of memory. Remember
> that the memory footprint on each processor scales as lx1^3 * lelt. With
> certain compilers you can add the following option in makenek:
>
> G="-mcmodel=medium"
>
> This allows you compile larger arrays, and generally use more memory.
>
> Best,
> Justin
>
>
>
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20120924/ba90f41d/attachment.html>