[gmx-developers] experimental force calculations

[Andy Somogyi]

On Jun 19, 2012, at 11:15 AM, Berk Hess wrote:

> On 6/19/12 16:46 , Andy Somogyi wrote:
>> Hi
>> 
>> I am doing research in an new way of calculating long range forces, and looking for some advice on how to best implement it in Gromacs.
>> 
>> This method will share many similarities with the fast multipole method (at least in terms of a distributed hierarchal data structure) , and will use a both MPI and GPU.
>> 
>> I noticed the newer GPU code is all based on OpemMM.
> I don't understand what you mean with "newer GPU code".
> The old GPU code is OpenMM, the new one is not.

I was not aware of the new 'native' GPU code until now, so when I said 'newer GPU code', I was referring to the OpenMM. Like Carsten said, I'll take a look at the new new native GPU code in "nbnxn_hybrid_acc".

>> 
>> As a starting template, would you suggest following the style used in the existing PME code, or the new OpenMM code, or something else entirely?
> Recently several electrostatics projects have started. I would suggest to use the PME call framework.
> 
> What method are you talking about?

Its a new method we are developing based on potential splitting, don't have a name for it yet but its going to be very similar to FMM, and should actually be identical to FMM based on a parameter. 

> 
> Cheers,
> 
> Berk
>> 
>> thanks
>> 
>> 
> 
> -- 
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-developers-request at gromacs.org.