[Trilinos-Users] [EXTERNAL] Anasazi timers

[David Hysom]

Many thanks; that's exactly what we need!

On 01/05/2015 11:11 AM, Thornquist, Heidi K wrote:
> Hi David,
>
> The global-participation requirement comes from the 
> Teuchos::TimeMonitor class, if summarize() is called without a 
> communicator, and Anasazi is agnostic of the underlying communicator.
> You can enable the timers for Anasazi, but you cannot ask the 
> eigensolvers' verbosity to include "Anasazi::TimingDetails".  Instead, 
> after the solve() method has returned, you can call 
> Teuchos::TimeMonitor::summarize with the sub-communicator that you are 
> using for that solve.  This is explained in the Teuchos documentation 
> here:
>
> http://trilinos.org/docs/dev/packages/teuchos/doc/html/classTeuchos_1_1TimeMonitor.html#a943add952f1c75ed13ce6a95a092e942
>
> Please let me know if you have any further questions.
>
> Thanks,
> Heidi
>
> -- 
>
> Heidi K. Thornquist
> Electrical Models & Simulation
> Sandia National Laboratories
> Albuquerque, NM  87185-1323
>
>
> From: <Hysom>, "David A." <hysom1 at llnl.gov <mailto:hysom1 at llnl.gov>>
> Date: Sunday, January 4, 2015 3:29 PM
> To: "trilinos-users at software.sandia.gov 
> <mailto:trilinos-users at software.sandia.gov>" 
> <trilinos-users at software.sandia.gov 
> <mailto:trilinos-users at software.sandia.gov>>
> Cc: "ayoo at llnl.gov <mailto:ayoo at llnl.gov>" <ayoo at llnl.gov 
> <mailto:ayoo at llnl.gov>>
> Subject: [EXTERNAL] [Trilinos-Users] Anasazi timers
>
> Happy new year, all!
>
> We are writing an asynchronous eigensolver-based code.
> By "asynchronous" I mean: we spin off a group of processors that
> runs Anasazi on one matrix; a second group that simultaneously
> runs Anasazi on a second matrix; etc.
>
> In our serial prototype
> we my analyzing runtime performance using the Anasazi/Teuchos
> timers. However, these apparently assume that all processors in
> MPI_COMM_WORLD participate in every solve. Since this was not
> our case, our code hung.
>
> We got around this by compiling Trilinos without the timers.
> So our code now works -- but we can't get the timing breakdown
> for the various solve stages that we'd like.
>
> So my question is: is there any reasonably easy way to modify Anasazi
> to get rid of this global-participation requirement?
>
> thanks, David

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://software.sandia.gov/pipermail/trilinos-users/attachments/20150105/3d39e63b/attachment.html>