On Wed, Jun 1, 2011 at 3:43 PM, John Fettig <span dir="ltr"><<a href="mailto:john.fettig@gmail.com">john.fettig@gmail.com</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Hi Matt,<br>
<div class="im"><br>
On Wed, Jun 1, 2011 at 4:06 PM, Matthew Knepley <<a href="mailto:knepley@gmail.com">knepley@gmail.com</a>> wrote:<br>
> It looks like the calls are taking 0.5 microseconds, but there are 89G<br>
> insertions<br>
> against only 13K MatMults that take about 1s apiece.<br>
> This seems like an awful lot of insertions into a matrix that can be applied<br>
> in 1s.<br>
<br>
</div>Unless I'm reading it wrong, there are 89M insertions. This is from<br>
10 timesteps with nonlinear iteration on each timestep, with ~1.5M<br>
elements. The matrices are reconstructed at virtually every substep.<br>
I don't quite understand what you mean by "can be applied in 1s",<br>
doesn't it take ~1ms per call? Anyways, I agree that this is a poor<br>
ratio of work to assemble/use of the matrices.<br></blockquote><div><br></div><div>Yes, both units were wrong.</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
We had thought about constructing the matrix outside of PETSc and then<br>
passing PETSc pointers to the matrix, but maybe matrix-free would be<br>
better for some of the equations where jacobi preconditioning is<br>
sufficient.<br></blockquote><div><br></div><div>I think it would be hard to speedup the insertion.</div><div><br></div><div>Why are you reassembling? Just for Newton? You could use a lagged version</div><div>of the matrix as a preconditioner, and run matrix-free for the matvecs.</div>
<div><br></div><div> Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Thanks,<br>
<font color="#888888">John<br>
</font></blockquote></div><br><br clear="all"><br>-- <br>What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener<br>