[Trilinos-Users] errors build MPI tpetra CrsMatrix (David Hysom)

[David Hysom]

Mark,

Will do. I've found something mysterious: I can make the
error go away by commenting out a method that isn't
even called by my driver. This is extremely mysterious,
and makes no sense at the moment. I'm trying to reproduce
the errors in a smaller toy problem. Will let you know what
I find out.

- David

On 09/04/2014 11:11 AM, Hoemmen, Mark wrote:
> On 9/4/14, 12:00 PM, "trilinos-users-request at software.sandia.gov"
> <trilinos-users-request at software.sandia.gov> wrote:
>
>> ------------------------------
>>
>> Message: 2
>> Date: Wed, 03 Sep 2014 14:23:52 -0700
>> From: David Hysom <hysom1 at llnl.gov>
>> To: "trilinos-users at software.sandia.gov"
>> 	<trilinos-users at software.sandia.gov>
>> Subject: [Trilinos-Users] errors build MPI tpetra CrsMatrix
>> Message-ID: <540786E8.3070807 at llnl.gov>
>> Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"
>>
>> I'm getting an error building a Tpetra::CrsMatrix;
>> The attached is a slightly edited version of the code
>> used to build the matrix.
>>
>> When I invoke our driver with a single processor,
>> everything works fine. When I use two or more
>> processors I get a traceback (using valgrind; see attached)
>> That shows an error in the call to A->fillComplete();
>> this subsequently shows a problem when makeColMap()
>> is invoked.
>>
>> My debug output shows that the correct rows/columns are being
>> inserted on the appropriate processors.
>>
>> Can you spot an error, or suggest ways of debugging
>> to find out what's going on?
> Hm, I've never seen _that_ before.  Would you mind rebuilding in debug
> mode and running it again?  Set the CMake option
> Teuchos_ENABLE_DEBUG:BOOL=ON, and make a DEBUG build.  That should turn on
> some additional error checking.
>
> mfh
>
> _______________________________________________
> Trilinos-Users mailing list
> Trilinos-Users at software.sandia.gov
> https://software.sandia.gov/mailman/listinfo/trilinos-users