On Mon, Aug 6, 2012 at 6:32 PM, Anthony Scopatz wrote:
> On Mon, Aug 6, 2012 at 5:25 PM, Daniel Wheeler
> wrote:
>
>>
>>
> Hmm interesting. Was your HDF5 itself configured & compiled with MPI
> support? Or are you using a serial version?
>
Yes, compiled with MPI support.
--
Daniel Wheeler
Yeah, that is probably it. There is probably some weird overlap
of resources for fipy and hdf5. Sorry, I don't know what we can do about
this, but I would bring it up with the fipy people. They should know what
do do and how to fix it.
Be Well
Anthony
On Tue, Aug 7, 2012 at 8:22 AM, Daniel Whe
On Tue, Aug 7, 2012 at 11:49 AM, Anthony Scopatz wrote:
> Yeah, that is probably it. There is probably some weird overlap
> of resources for fipy and hdf5. Sorry, I don't know what we can do about
> this, but I would bring it up with the fipy people.
I am the fipy people.
> They should know
On Tue, Aug 7, 2012 at 11:43 AM, Daniel Wheeler
wrote:
>
>
> On Tue, Aug 7, 2012 at 11:49 AM, Anthony Scopatz wrote:
>
>> Yeah, that is probably it. There is probably some weird overlap
>> of resources for fipy and hdf5. Sorry, I don't know what we can do about
>> this, but I would bring it up w
On Tue, Aug 7, 2012 at 12:46 PM, Anthony Scopatz wrote:
>
> On Tue, Aug 7, 2012 at 11:43 AM, Daniel Wheeler > wrote:
>>
>>
>>
>>> They should know what do do and how to fix it.
>>
>>
>> Maybe mpi init issues with either pytrilinos or mpi4py as a wild guess.
>> Both are imported by fipy.
>>
>
> Y
On Tue, Aug 7, 2012 at 11:50 AM, Daniel Wheeler
wrote:
>
>
> On Tue, Aug 7, 2012 at 12:46 PM, Anthony Scopatz wrote:
>
>>
>> On Tue, Aug 7, 2012 at 11:43 AM, Daniel Wheeler <
>> daniel.wheel...@gmail.com> wrote:
>>
>>>
>>>
They should know what do do and how to fix it.
>>>
>>>
>>> Maybe mpi i
