ASEArch - Algorithms and Software for Emerging Architectures

Study group: Future-proof parallelisation of multi-block structured grid codes

April 25-26, 2013 (near Oxford)

The first CCP ASEArch Study Group “Future-proof parallelisation of multi-block structured grid codes” took place at the Weston Manor Hotel near Oxford on April 25-26, 2013. Meeting notes can be found in this link. The agenda is shown below with some links to the presentations on the day.

If you are interested in this subject or if you need further information about this study group please contact Prof Mike Giles.

Click this link to access Study Group Meeting Notes.

Agenda

Thursday, 25th April

10:00    Welcome (Mike Giles)
10:15 Scene setting (Mike Giles)
  • ASEArch and OPS
  • Opportunities and challenges of manycore computing
  • Specific features (e.g. tiling, checkpointing)
  • Objectives of developing "library" support for structured grid applications
11:00 Coffee
11:30 Code 1 (Chris Allen)
12:30 Lunch
1:30    Code 2 (Sebastian Timme)
2:30 Introduction to OpenACC (Simon McIntosh-Smith)
3:30 Coffee
4:00    More on OpenACC (Mike Giles)
 
5:00    Lattice Boltzmann codes -- some similarities (Mike Giles)
 
5:30    Finish
7:00    Dinner

Friday, 26th April

9:00 Code 3 (Nico De Tullio)
10:00 A summary of code requirements (Mike Giles)
11:00 Coffee
11:30 Outline ideas for an API (Mike Giles)
12:30 Lunch
1:30 Discussion of API (all)
3:30 Coffee
4:00 Next steps for ASEArch and OPS projects
5:00 Finish

   Other Resources

COSA Compressible Finite Volume Parallel Multiblock Multigrid Navier-Stokes Solver
Dr M.Sergio Campobasso, Jernej Drofelnik
University of Lancaster, Lancaster LA1 4YR, UK
University of Glasgow, Glasgow G12 8QQ, UK

The Dally-nVIDIA-Stanford Prescription for Exascale Computing

Result: The End of Historic Scaling - Figure 1: Conventional techniques to boost performance are no longer applicable
(Source: W. Dally, Keynote at IPDPS 2011).

GPUs and the Future of Parallel Computing - Keckler, S.W.; Dally, W.J.; Khailany, B.; Garland, M.; Glasco, D.,

From Here to ExaScale Challenges and Potential Solutions
Bill Dally
Chief Scientist, NVIDIA
Bell Professor of Engineering, Stanford University