[petsc-dev] preallocation

Barry Smith bsmith at mcs.anl.gov
Tue Sep 20 12:02:24 CDT 2011


On Sep 20, 2011, at 11:17 AM, Jed Brown wrote:

> On Tue, Sep 20, 2011 at 18:12, Barry Smith <bsmith at mcs.anl.gov> wrote:
> We should move all the additional functionality that AIJ provides (inodes, whatever else) into BAIJ and then remove all of AIJ.   It is a big extra source code duplication maintenance nightmare it would be nice to get rid of.
> 
> Yes, but the bs=1 still needs MatSetValuesBlocked() to be able to use a different block size (so we can assembly a scalar matrix format with the same code).

   Correct, and that is one of the features that needs to be ported over from AIJ. Making getting rid of AIJ a big pain 

   Barry





More information about the petsc-dev mailing list