[Trilinos-Users] scalability tests with aztec file example

[Heroux, Mike]

Lucie,

This is a very simple example that should not present any scaling problems
except for two possible reasons:

- The linear system size is too small to make effective use of more than
one processor.
- You do not have a properly working MPI environment.

>From the results you show, I would be surprised if you saw no speed up
going from 1 to 2 processors, but perhaps you can try a much larger sized
problem, say 10,000,000, and see if there is any speedup.  If this does
not work, then I am guessing you have perhaps not compiled Trilinos with
MPI enabled, or your MPI environment is not working properly.

Mike

On 3/28/14 10:35 AM, "Lucie Freret" <lfreret at arrow.utias.utoronto.ca>
wrote:

>Hello all,
>
>I am a new user of this list and maybe my question has already been
>addressed - sorry if it is the case.
>
>I am using the file example cxx_main.cpp located at
>trilinos-11.4.2-Source/packages/aztecoo/example/simple_epetra_aztecoo
>to realize some scalability tests.
>
> From this file, I changed the number of equations (default
>value=1000) to get a global matrix size equal to 500,000 (half a
>million). According to the number of cores, I'm changing the number of
>equations per core to keep the global matrix size constant. The
>speed-up that I get is quite bad - see the file below.
>
>Does anyone have already looked at this example or done scalability
>tests with Aztecoo?
>
>Thanks for your help,
>Best regards,
>Lucie Freret
>
>----------------------------------------------------------------
>This message was sent using IMP, the Internet Messaging Program.
>