PEEPS: Programming Environments for Emerging Parallel Systems



A programming model is a bridge between a developers natural model of an application and an implementation of that application on available hardware. With the growing complexity of parallel computing systems there is an increased emphasis upon developing programming models that achieve high performance and portability, while simultaneously enabling programr productivity. These goals will become particularly challenging for future many-core based architectures, where applications are increasingly likely to rely on fine-grained parallelism and strong scaling while supporting fault resilience — to accommodate the massive growth of explicit on-chip parallelism and constrained bandwidth.

The PEEPS workshop will provide opportunities for researchers and practitioners in all fields related to programming frameworks and models for high-performance computing to exchange ideas and experiences.


(CITRIS, Room 250 - June 22, 2010 )

08:30-08:40 Welcome Remarks
keynote talk
Analyzing Large-Scale Heterogeneous Systems through the Ages
Darren Kerbyson (Pacific Northwest National Laboratory)
09:30-10:30 Exploiting on-chip Parallelism with Hybrid MPI/OpenMP Programming [slides]
Gabriele Jost (Texas Advanced Computing Center)
Lattice Boltzmann Hybrid Auto-Tuning on High-End Computational Platforms
Sam Williams (Lawrence Berkeley National Laboratory)
10:30-11:00 Coffee Break
11:00-12:00 Compiler Analysis and Optimization of Habanero-Java Program [slides]
Rajkishore Barik (Rice University)
Experiences with UPC at Scale [slides]
Yili Zheng (Lawrence Berkeley National Laboratory)
12:00-13:30 Lunch
13:30-15:00 GPU Acceleration of Numerical Weather Prediction [slides]
John Michalakes (UCAR)
Thrust: A Parallel Template Library for CUDA [slides]
Jared Hoberock (NVIDIA)
Multi-GPU Computing and GPU MapReduce [slides]
John Owens (UC Davis)
15:00-15:30 Coffee Break
15:30-17:00 An Overview of the Morfeus Project: Object-oriented Multiphysics Framework Development in Fortran 2003 [slides]
Damian Rouson (Sandia National Laboratory)
Parallel Scripting with Swift for Applications at the Petascale and Beyond [slides]
Michael Wilde (Argonne National Laboratory)
Charm++: on the road to Exascale [slides]
Isaac Dooley (University of Illinois at Urbana-Champaign)