[mpiwg-persistence] Notes from meeting today 25th Oct 2017
skjellum at auburn.edu
Wed Oct 25 10:55:26 CDT 2017
Can we do next meeting on nov 2 please
I'll respond in detail on these minutes tonight !
Anthony Skjellum, PhD
On Oct 25, 2017, at 9:44 AM, HOLMES Daniel <d.holmes at epcc.ed.ac.uk<mailto:d.holmes at epcc.ed.ac.uk>> wrote:
Notes from today's meeting:
1) extended version of EuroMPI 2017 "Performance Planning" paper for journal special issue
[Dan] add state-machine diagram and various sentences throughout section 3 - emphasise life-cycle difference between blocking/nonblocking cache (must be tied to comm) and persistent cache (can be tied to request).
[Puri] change format of paper to Elsevier journal layout
[Puri] update CG example to one that gave good results
[Srinivas] integrate the reference implementation of persistent collectives into MLSL and run a machine learning application use-case - add plot (if possible) and text to paper in results section - focus on correctness/more-natural programming style, i.e. lack of need to duplicate comm per collective to avoid ordering issues
2) further implementation/optimisation work, aiming at new paper (possible target SC18)
[Dan] scope out and begin coding work on infrastructure improvement, initially for MPI_Allreduce_init
[Puri] code review and testing of PNBC new code
Next meeting: Wed 1st Nov 2017 8am Central US via GoToMeeting.
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
mpiwg-persistence mailing list
mpiwg-persistence at lists.mpi-forum.org<mailto:mpiwg-persistence at lists.mpi-forum.org>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mpiwg-persistence