[petsc-users] Enquiry regarding log summary results

TAY wee-beng zonexo at gmail.com
Sat Oct 6 16:19:55 CDT 2012


On 5/10/2012 5:17 PM, Jed Brown wrote:
> Now most of the time is spent in PCApply, though an inordinate amount 
> is still in VecScatterEnd, indicating that either your partition is 
> very bad or the machine has contention (either with other jobs, zombie 
> processes, incorrect affinity, over-subscribed).
>
> In any case, you can try changing the smoother. For example, try 
> skipping the Chebyshev iterations and just use one iteration of SOR...

May I know where can I find all the options for GAMG? I can't find it in 
the manual.

I also used -info for the GAMG and what does these mean:

1. [9] MatAssemblyBegin_MPIAIJ(): Stash has 14396 entries, uses 1 mallocs

How serious is this? Does it mean there's a lot of mallocs going on? Do 
I correct this by using higher d_nz and o_nz?

2. [4] MatAssemblyEnd_SeqAIJ(): Matrix size: 112785 X 112785; storage 
space: 0 unneeded,3628127 used
[4] MatAssemblyEnd_SeqAIJ(): Number of mallocs during MatSetValues() is 0
[4] MatAssemblyEnd_SeqAIJ(): Maximum nonzeros in any row is 59

When I use MatCreateAIJ, my d_nz and o_nz are both 7. If the row is 59, 
does it mean that my d_nz and o_nz are too small. Must I set it to the 
maximum nonzeros given by -info?

3. [0] MatSetOption_SeqAIJ(): Option STRUCTURALLY_SYMMETRIC ignored

My Poisson matrix is symmetrical, so why is it ignored?
>
> On Fri, Oct 5, 2012 at 10:08 AM, TAY wee-beng <zonexo at gmail.com 
> <mailto:zonexo at gmail.com>> wrote:
>
>     On 4/10/2012 11:44 PM, Satish Balay wrote:
>
>         On Thu, 4 Oct 2012, Matthew Knepley wrote:
>
>             the Mom-Z solve is bad because the assembly of the operator is
>             screwed up. You are communicating too many values. So,
>             jsut go into
>             your code and count how many off process entries you set.
>
>         running with '-info' should give debug info regarding this
>         communication [stack size used for messages, and the number of
>         mallocs
>         needed etc..]
>
>         Satish
>
>
>     Ok, I found my problem. I was using:
>
>     PetscLogStage  stage
>
>     and the time for the poisson eqn got combined into that of the
>     momentum_z solve. Hence the confusion in results.
>
>     I have attached the new results. The time is taken up mainly by
>     GAMG. I am using:
>
>     -poisson_pc_gamg_agg_nsmooths 1
>     -poisson_pc_type gamg
>
>     What are the possible ways to improve the efficiency or reduce the
>     time?
>
>     If I use HYPRE's geometric multigrid, performance is somewhat
>     better, such that momentum_x/y/z accounts for a total of 15%
>     computing time.
>
>     From the log_summary, for the momentum eqns, it seems that
>     MatAssembly is the main problem, is that so?
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20121006/305b205c/attachment.html>


More information about the petsc-users mailing list