Hi Johan,

Thanks for that pointer. It's (for obvious and good reasons) focussed on
the intra-node performance of assembly, which means that the assembly
benchmarks only go up to 16 cores (I feel horribly hypocritical here
because we do a lot of those sort of comparisons in PyOP2 as well). It
doesn't provide any evidence that Dolfin is capable of employing lots of
processors.

Cheers,

David


On 4 September 2013 10:46, Johan Hake <[email protected]> wrote:

>  It is just a report, but I guess it answers most of your questions.
>
>  https://t.co/1dPJZYZRUa
>
>  Garth and Chris are the obvious goto here.
>
>  Johan
>
>
> On Wed, Sep 4, 2013 at 11:39 AM, David Ham <[email protected]>wrote:
>
>> Hi All,
>>
>>  I'm writing responses to reviewers for the FEniCS on manifolds paper
>> which Marie, Colin, Andrew and I submitted to GMD. One of the reviewers is
>> giving us grief about not including parallel scaling results. The reason we
>> haven't done so is that we were just modifying FFC, which is basically
>> orthogonal to the parallel performance problem, so what I would like to be
>> able to do is point them to the existing Dolfin parallel performance and
>> claim we haven't done anything which would change that. Unfortunately, I
>> haven't so far located a published result which shows Dolfin's parallel
>> performance. The FEniCS book documents the existence of shared and
>> distributed memory parallel in chapter 10, but doesn't give performance
>> results.
>>
>>  Any suggestions?
>>
>>  Many thanks,
>>
>>  David
>>
>>  --
>> Dr David Ham
>> Department of Computing
>> Imperial College London
>>
>> http://www.imperial.ac.uk/people/david.ham
>>
>> _______________________________________________
>> fenics mailing list
>> [email protected]
>> http://fenicsproject.org/mailman/listinfo/fenics
>>
>>
>


-- 
Dr David Ham
Department of Computing
Imperial College London

http://www.imperial.ac.uk/people/david.ham
_______________________________________________
fenics mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics

Reply via email to