Skip to Content.
Sympa Menu

h5part - [H5part] Re: H5Block and ghost zones

h5part AT lists.psi.ch

Subject: H5Part development and discussion

List archive

[H5part] Re: H5Block and ghost zones


Chronological Thread 
  • From: Mark Howison <MHowison AT lbl.gov>
  • To: Achim Gsell <achim.gsell AT psi.ch>
  • Cc: h5part AT lists.psi.ch
  • Subject: [H5part] Re: H5Block and ghost zones
  • Date: Thu, 8 Jul 2010 08:58:05 -0700
  • List-archive: <https://lists.web.psi.ch/pipermail/h5part/>
  • List-id: H5Part development and discussion <h5part.lists.psi.ch>

I should have added: the extra information I am thinking of is the 3D
index of each proc's block within the larger field, so for example
with 8 procs we might have

0 -> (0,0,0)
1 -> (1,0,0)
2 -> (0,1,0)
3 -> (1,1,0)
etc.

and this could be specified with a call like

H5Block3dSetIndex(f, i, j, k);

Mark

On Thu, Jul 8, 2010 at 8:54 AM, Mark Howison <MHowison AT lbl.gov> wrote:
> Hi Achim,
>
> I'm still working through the H5Block code, and I wanted to find out
> more about the original design decisions for handling arbitrary ghost
> zones, which introduces alot of complexity in "dissolving" them. I
> don't see an easy way to get around the O(n) space required in the
> H5Block file struct to store every proc's field layout for use in the
> ghost zone routines. Do we need to support arbitrary ghost zones? Is
> there a simpler way to handle them? I think if we had some additional
> information, like which procs are neighbors, we could drastically
> reduce the calculations/space required, but this would mean asking the
> user to provide more information.
>
> Mark
>




Archive powered by MHonArc 2.6.19.

Top of Page