[petsc-users] log_summary MPI

[Matthew Knepley]

On Fri, Mar 6, 2015 at 12:56 PM, Justin Chang <jchang27 at uh.edu> wrote:

> Matt, thank you for the response. One more question, is there a way to
> output only the MPI messages and MPI message lengths? I don't want to print
> everything when I do -log_summary, so are there any command line options
> that do this? I can't seem to find anything in the manual that discusses
> this.
>

These are just global counters:


https://bitbucket.org/petsc/petsc/src/1ddf9febf6a13a173c13b6b0dde02ed9ccfd590e/src/sys/logging/plog.c?at=master#cl-51

You can output them whenever you want. If you want global numbers, you will
need to call reductions yourself.

  Thanks,

    Matt


> Thanks,
>
> On Fri, Mar 6, 2015 at 9:10 AM, Matthew Knepley <knepley at gmail.com> wrote:
>
>> On Thu, Mar 5, 2015 at 6:12 PM, Justin Chang <jchang27 at uh.edu> wrote:
>>
>>> Hi all,
>>>
>>> I apologize if these may be simple questions that could be found
>>> somewhere. In the log_summary profiling, I have a few questions about the
>>> MPI messages/MPI message lengths metrics:
>>>
>>> 1) For MPI messages, does the value under Max correspond to the maximum
>>> number of messages a single processor passes?
>>>
>>
>> Yes.
>>
>>
>>> 2) What are the units of MPI message lengths? Is it in bytes?
>>>
>>
>> Yes, returned by PetscMPITypeSize().
>>
>>
>>> 3) What does MPI message lengths refer to exactly? Like, does it refer
>>> to the max/min/average size per sent/received message or does it refer to
>>> the cumulative size that a single processor sends/receives?
>>>
>>
>> Cumulative, see
>>
>>
>> https://bitbucket.org/petsc/petsc/src/d8151eeaff97562eb317e17b7b0cecab16831f69/include/petsclog.h?at=master#cl-343
>>
>>   Thanks,
>>
>>      Matt
>>
>>
>>> Thanks
>>>
>>> --
>>> Justin Chang
>>> PhD Candidate, Civil Engineering - Computational Sciences
>>> University of Houston, Department of Civil and Environmental Engineering
>>> Houston, TX 77004
>>> (512) 963-3262
>>>
>>
>>
>>
>> --
>> What most experimenters take for granted before they begin their
>> experiments is infinitely more interesting than any results to which their
>> experiments lead.
>> -- Norbert Wiener
>>
>
>
>
> --
> Justin Chang
> PhD Candidate, Civil Engineering - Computational Sciences
> University of Houston, Department of Civil and Environmental Engineering
> Houston, TX 77004
> (512) 963-3262
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20150306/8fd884f0/attachment.html>