[petsc-users] improving efficiency of matrix multiplication by calling BLAS directly

Jed Brown jedbrown at mcs.anl.gov
Thu Nov 8 22:28:49 CST 2012

Only if your matrices are *not sparse*, by which we main that more than,
say, 20% of the entries are nonzero.

On Thu, Nov 8, 2012 at 9:10 PM, Ganesh Hegde <ghegde at purdue.edu> wrote:

> I have an application where I am multiplying several large, complex PETSc
> matrices (that could be sparse) with each other.
> This happens often enough during program execution that I am interested in
> looking for optimal solutions to this.
> The usual algorithm followed is initializing a PETSc matrix with the
> required dimensionality and specifying the sparsity pattern.
> They're then multiplied with each other.
> So my questions are as follows:
> 1) Could specifying a matrix as full instead of sparse improve speed by
> reducing overhead involved in processing sparse matrices?
> 2) If not, should BLAS routines be called directly from the code instead
> of PETSc calling them, to improve performance?
> Regards,
> --
> Ganesh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20121108/0f255be0/attachment.html>

More information about the petsc-users mailing list