[petsc-users] Can I use TAO for embarrassingly parallel problems with multiple threads?

[Matthew Knepley]

On Thu, Dec 20, 2018 at 10:46 AM Krzysztof Kamieniecki <krys at kamieniecki.com>
wrote:

> Hi Matt,
>
> What does "COMPLETELY independent" mean? Should I call PetscInitialize in
> each thread?
>

No, it means that separate threads do not access the same objects. Your
error message indicated
that this was happening.


> We have an "approximate" model where we treat different regions as
> independent. It's a "realtime" system where measurements from the different
> regions arrives a different times and we want to process regions ASAP.
>

Yes. None of that necessitates threads, or is even really easier with
threads. Tons of realtime systems use
processes. The only real difference between processes and threads is that
threads share memory, which
is inherently unsafe.

  Thanks,

    Matt



> Best Regards,
> Krys
>
> On Thu, Dec 20, 2018 at 10:35 AM Matthew Knepley <knepley at gmail.com>
> wrote:
>
>> On Thu, Dec 20, 2018 at 10:30 AM Krzysztof Kamieniecki via petsc-users <
>> petsc-users at mcs.anl.gov> wrote:
>>
>>> Hello All,
>>>
>>> I have an embarrassingly parallel problem that I would like to use TAO
>>> on, is there some way to do this with threads as opposed to multiple
>>> processes?
>>>
>>
>> You should be able to run COMPLETELY independent object on different
>> threads. However, notice that
>> there is absolutely no advantage to doing this. Why do you want to run
>> independent instances on multiple
>> threads?
>>
>>   Thanks,
>>
>>      Matt
>>
>>
>>>  I compiled PETSc with the following flags
>>> ./configure \
>>> --prefix=${DEP_INSTALL_DIR} \
>>> --with-threadsafety --with-log=0 --download-concurrencykit \
>>> --with-openblas=1 \
>>> --with-openblas-dir=${DEP_INSTALL_DIR} \
>>> --with-mpi=0 \
>>> --with-shared=0 \
>>> --with-debugging=0 COPTFLAGS='-O3' CXXOPTFLAGS='-O3' FOPTFLAGS='-O3'
>>>
>>> When I run TAO in multiple threads I get the error "Called VecxxxEnd()
>>> in a different order or with a different vector than VecxxxBegin()"
>>>
>>> Thanks,
>>> Krys
>>>
>>>
>>
>> --
>> 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
>>
>> https://www.cse.buffalo.edu/~knepley/
>> <http://www.cse.buffalo.edu/~knepley/>
>>
>

-- 
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

https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20181220/d666a35e/attachment-0001.html>