[mpiwg-hybridpm] moar corez -> endpoints are necessary

[Jeff Hammond]

See the table in
https://computing.llnl.gov/tutorials/mpi_performance/#IBMSpecific
associated with the attached figure.

jeff

On Tue, Dec 10, 2013 at 11:11 AM, Jeff Hammond <jeff.science at gmail.com> wrote:
> BGQ data:
>
> http://dx.doi.org/10.1109/IPDPS.2012.73
>
> On Tue, Dec 10, 2013 at 11:07 AM, Jeff Hammond <jeff.science at gmail.com> wrote:
>> BGP data:
>>
>> https://fs.hlrs.de/projects/eurompi2010/TALKS/MONDAY_MORNING_TRACK1/enabling_concurrent_multithreaded_MPI_communication_on_multiscore_PFLOPS.pdf
>>
>> On Tue, Dec 10, 2013 at 11:04 AM, Jeff Hammond <jeff.science at gmail.com> wrote:
>>> Squyres insists upon real data.  Here is one data point:
>>>
>>> "Evaluating NIC hardware requirements to achieve high message rate
>>> PGAS support on multi-core processors"
>>> http://dx.doi.org/10.1145/1362622.1362671
>>>
>>> Jeff
>>>
>>> --
>>> Jeff Hammond
>>> jeff.science at gmail.com
>>
>>
>>
>> --
>> Jeff Hammond
>> jeff.science at gmail.com
>
>
>
> --
> Jeff Hammond
> jeff.science at gmail.com



-- 
Jeff Hammond
jeff.science at gmail.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: bandwidthTaskPerNode.gif
Type: image/gif
Size: 13365 bytes
Desc: not available
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-hybridpm/attachments/20131210/2f533480/attachment-0001.gif>