[petsc-dev] Lower memory DMDA

[Barry Smith]

On May 8, 2013, at 7:54 AM, Peter Brune <prbrune at gmail.com> wrote:

> DMGetCoordinateDM is, however, collective.  Looking deeper, it is created only when used, but setting uniform coordinates is using it.  We might want to be more upfront about this.  It's probably not their problem.

    You guys are speculating too much. Create a DMDA characteristic for their code with reasonably large array dimensions and then -malloc_dump it (or something similar) to find out where the memory actually is. Then given this table of sizes propose changes, not before. 

   Barry

> 
> 
> On Wed, May 8, 2013 at 7:50 AM, Jed Brown <jedbrown at mcs.anl.gov> wrote:
> Peter Brune <prbrune at gmail.com> writes:
> 
> > The coordinate DM also is automatically created and has dim fields in its
> > vectors, making it far more heavyweight than a scalar problem.  It should
> > definitely be something that's lazily evaluated.
> 
> DMGetCoordinates() is currently labeled "Not Collective".
>