[petsc-dev] Fwd: update from exascale workshop

[Barry Smith]

Begin forwarded message:

> From: Lois Curfman McInnes <curfman at mcs.anl.gov>
> Date: October 13, 2011 5:14:51 PM CDT
> To: LANS Council <lans-council at mcs.anl.gov>
> Subject: update from exascale workshop
> 
> 
> Hi folks ... I'm writing with a brief update from this week's exascale workshop; meeting info is here:
>     http://exascaleresearch.labworks.org/ascrOct2011/ascr
> including presentations, pointers to existing projects, prior workshops, etc.  
> 
> Sven, Mihai, and I represented LANS; they may have additional comments.  Other attendees from Argonne included reps from CESAR and Radix.
> 
> From my perspective, the most important note is that Bill Harrod announced plans for 2 upcoming exascale FOAs:
> 
> (1) X-Stack FOA: expected in November 
>     - runtimes, programming models, languages, compilers, and tools: emphasis on cross-cutting among these
> 
> (2) Extreme-scale algorithms FOA:  no specific date mentioned
>     - first a workshop for extreme-scale algorithms
> 
> Other notes:
> 
> Sandy Landsberg convened a brief meeting yesterday to discuss exascale math needs.  An outcome is that a group will write a white paper; Mike Heroux is leading the whitepaper group, and Sven is the Argonne rep on the writing team.
> 
> I briefly spoke w. Rajeev Thakur and Rob Ross about the utility of increased dialogue among Radix and LANS folks in general -- they indicated interest in this. I know that some discussions are already happening.
> 
> Another observation is that some other labs have pots of funding available for incubating projects that can then be poised to compete strongly in various FOAs.  I wonder whether ANL has ever considered increasing the pool of money available for LDRD projects that serve this purpose + provide more opportunities for junior researchers to get experience in writing proposals. 
> 
> All for now --
> Lois 
> 
> P.S. - I plan to be at the lab on Nov 3 and 4.
> 
> - - - - - - - 
> 
> Below are brief/incomplete notes from the final presentation of the workshop by Bill Harrod (ASCR) and Thuc Hoang (NNSA):  
> 
> Challenges to achieving exascale computing: see prior presentation
> 
> Real challenges to achieving success: see prior presentation
>  - avoid mediocre solutions, etc ... 
> 
> How will we succeed:
>  - formulating dynamic plan, executing plan
>  - focus on enabling HPC community to achieve extreme scale computing to enable extreme-scale scientific discovery
>  - recognize what we do best and do it
>  - accept change
>  - open competition within DOE community for FOAs
>  - enable open discussions on exascale within DOE community
>  - enable an open collaborative environment
> 
> What's next:
>  - Complete exascale plan: needs to be done by Nov 10
>  - Release ASCR/X-Stack solicitation: expected in November 
>     - runtimes, prog models, languages, compilers, and tools
>  - formulate the software stack plan
>  - workshop for extreme-scale algorithms
>  - electrical vs. optical interconnect workshop
>  - release ASCR/Extreme scale algorithms FOA
>  - more workshops, more FOAs
>  - subject to available funding
>     - start funding software technology, path forward and system design programs
>  - April 16-19, 2012:  Exascale research meeting and roadmapping workshop (might expand to include April 20)
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20111013/13fc6c23/attachment.html>