I did not add oski.py. Aron, I never wrote SSE matrix kernels, just software prefetch. I don't know how it compares to OSKI, but the inode and block formats should be distinctly faster than scalar OSKI (since I believe they outperform the scalar optimal performance model).
On Apr 8, 2011 3:38 PM, "Matthew Knepley" <knepley at gmail.com> wrote: On Fri, Apr 8, 2011 at 6:50 AM, Barry Smith <bsmith at mcs.anl.gov> wrote: > > > Who put this in and does it do anything? Jed did. As far as I know, this project is dead. Matt > > Barry > > > Begin forwarded message: > > > From: Kadir Akbudak <kadir at cs.bilkent.edu.tr> > ... -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110408/6d656925/attachment.html>
