Yes you are right, it would be limited to dG, all variables of the same
polynomial degree and no local P refinement.
I know that this is very specific but the memory savings are quite
encouraging.

Lorenzo




On Tue, May 28, 2013 at 5:30 PM, Kirk, Benjamin (JSC-EG311) <
[email protected]> wrote:

>
> On May 28, 2013, at 7:55 AM, lorenzo alessio botti <
> [email protected]> wrote:
>
> > I've checked out the master branch to try the blocked implementation and
> it
> > works great.
> > The memory goes down to 1.2GB from 1.7GB, in Gnuid I have a system with a
> > VariableGroup of size 3 and a system with just one variable.
>
> Excellent!
>
> > I have a remark regarding the blocked implementation in the dG framework.
> > In this context It would be great to have a dof based blocked
> > implementation instead of a variable based blocked implementation. Do you
> > think it would be possible?
>
> Well there is one important limitation that I think will complicate this:
>  there can only be one block size per matrix.  So your dg*nvars example
> will work if all your elements are of the same type &/or p-level, but with
> local P refinement this will lead to an inconsistent block size and
> therefore cause a problem…
>
> Or am I missing something?
>
> -Ben
>
>
>
------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service 
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to