From: Roy Stogner <roy@st...> - 2008-08-01 17:49:42
-------- Forwarded Message --------
From: Adam C Powell IV <hazelsct@...>
Subject: Ridiculously slow matrix insertion
Date: Mon, 21 Jul 2008 15:34:12 -0400
I'm working on a problem similar to example 14 in 3-D, and the matrix
insertion process in system assembly is taking ridiculous amounts of
time, on the order of 95+% of the solution time. (The perf_log facility
doesn't seem to work; this is based on printf statements at the same
places as those calls.)
I notice that this takes longer for the later elements than earlier
ones. And the problem is far worse in 3-D than 2-D. With max_r_steps=6
some of the elements take as long as 0.4 second to insert on a 2.2 GHz
T7500. Also the "time" command shows that about 1/3 of the time is
spent in "system" activity, though there's no swapping going on.
It's odd, because it seems that this should be very fast in PETSc. That
is, unless it's computing the matrix fill at the same time -- but the
default solver seems to be PETSc's GMRES which doesn't involve fill...
Where should I start to investigate?
Adam C. Powell, IV http://www.opennovation.com/
P.O. Box 249 apowell@...
Newton, MA 02464-0002 USA (617) 290-8831