AFAIK we haven't yet done this. It's a bit off the short term critical path which probably explains why nobody has looked at it, although we really should since we want to claim performance portability in upcoming publications, and this is a potential bottleneck.

David

On 22 October 2013 11:59, Gorman, Gerard J <g.gorman@imperial.ac.uk> wrote:

On 22/10/13 05:20, Jed Brown wrote:
> Have you set up a test case showing matrix insertion to be
> performance-critical, so that we can show Matt?

Lawrence, Florian et al - where are we with this? Anything interesting
for the PETSc folks yet?

Cheers
Gerard

_______________________________________________
firedrake mailing list
firedrake@imperial.ac.uk
https://mailman.ic.ac.uk/mailman/listinfo/firedrake

Dr David Ham
Departments of Mathematics and Computing
Imperial College London

http://www.imperial.ac.uk/people/david.ham