[Trilinos-Users] IFPACK ILUT far slower than AztecOO buitin ILUT

[Davood Ansari]

Hi
I am using AztecOO to solve large linear problems resulting with very high
order (up to 12th) finite elements.
My experiments have lead to some thing which is a little bit annoying to me.
I observe that IFPACK's "ILUT" and "ILUT stand-alone" preconditioner  are
far  more  slower than  those of  AztecOO ( and even slower than Intel MKL).
The difference is very very big (abnormal) in  general and I  wonder if
there is a trick or tweak or at least a fundamental difference in IFPACK's
tresholded
preconditioners. I wonder if I can observe similar performances using IFPACK
by adjusting certain variables etc.

Please comment

Davood
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://software.sandia.gov/mailman/private/trilinos-users/attachments/20080214/6532388e/attachment.html