[MITgcm-support] Runtime for non-hydrostatic model

Martin Losch Martin.Losch at awi.de
Mon Sep 23 04:14:21 EDT 2013


This is a hack:

You can reduce the cost of the non-hydrostatic part of the code by not requiring full convergence of the cg3d solver. For example by setting the parameter cg3dMaxIters to a "small" number like 100 or even 40, the non-hydrostatic pressure contribution to the pressure gradients will not be exact (because not really converged), but since this contribution is usually small compared to the hydrostatic pressure, the overall error you make with this choice can be acceptable. (you need to checked that for your application)

Martin

On Sep 22, 2013, at 10:05 PM, yang.zhang <yang.zhang at mun.ca> wrote:

> non-hydrostatic is very expensive for numerical stimulation, of course, it is reasonable
> 
> 
> On Sun, Sep 22, 2013 at 6:01 AM, yulinghui <sysuylh at 163.com> wrote:
>     When I use the non-hydrostatic model to simulate the internal wave, the runtime for non-hydrostatic model  is much larger(about ten times) than that for hydrostatic model. I don't know whether it is right or not ?  And is there any set to improve it?
> 
> Thanks
> 
> 
> 
> 
> 
> _______________________________________________
> 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