[Trilinos-Users] expected Trilinos on shared memory machine.

[Daniel Wheeler]

Hi,

I currently work on a code called FiPy
(http://www.ctcms.nist.gov/fipy/). It is a python code that uses
Trilinos for back end solvers. Trilinos is used as it significantly
improves functionality and memory usage over other solver libraries
that we have used previously. We are currently evaluating high end
shared memory machines to use with our code. As yet we haven't used
Trilinos in parallel only on one CPU. Before I start going through the
machinations of compiling and testing Trilinos in parallel, I was
wondering if anyone could give me insight into the speedup that I
should expect on a shared memory machine with the usage below?

In our code, for a typical problem the majority of the compute time is
spent in the "AztecOO.AztecOO(A, LHS, RHS)"  function. The python code
for solving the matrix is here.

   <http://matforge.org/fipy/browser/trunk/fipy/solvers/trilinos/trilinosAztecOOSolver.py>

The code for building the matrix is here.

   <http://matforge.org/fipy/browser/trunk/fipy/tools/trilinosMatrix.py>

Given this typical use case what speedups should be expected on a
shared memory machine?

Thanks

-- 
Daniel Wheeler