International Workshop on OpenMP (IWOMP) 2011

Robert Harrison, Joint Institute of Computational Sciences UT/ORNL

Many are now questioning whether our current approaches to developing software for science and engineering are sustainable. In particular, can we deliver to society and the nation the full benefits expected from high-performance simulation at the peta- and exascales? Or is innovative science being stifled by the increasing complexities of all aspects of our problem space (rapidly changing hardware, software, multidisciplinary physics, etc.)?

Focusing on applications in chemistry and materials science, and motivated co-design of exascale hardware and software, I will discuss many of these issues including how chemistry has already been forced to adopt solutions that differ quite sharply to those in the mainstream, and how these solutions position us well for the technology transitions now underway.

Ralph Altenfeld, Markus Apel, An Mey Dieter, Bernd Boettger, Stefan Benke and Christian Bischof

Dinesh Kaushik, David Keyes, Satish Balay and Barry Smith

Mark Woodyard

Patrick Carribault, Marc Perache and Herve Jourdren

Leo White

James Beyer, Eric Stotzer, Alistair Hart and Bronis De Supinski

Jun Shirako, Kamal Sharma and Vivek Sarkar

Bronis R. de Supinski, Lawrence Livermore National Laboratory

An OpenMP 3.1 Specification draft was released for comment earlier this year. Concurrent with that, the committee has reviewed all sections in detail. Based on feedback from the community and that review, the final version will soon be voted on by the OpenMP Architecture Review Board (ARB). We expect the final version to be adopted in July. In addition, the Language Committee continues its efforts to develop a more comprehensive 4.0 update that will position OpenMP for current and future systems in which locality is increasingly important and accelerators are common. This talk will detail the changes in OpenMP 3.1 as well as directions for 4.0.

Arch Robison, Intel

Intel® Threading Building Blocks (Intel® TBB) is a library approach to parallel programming—no compiler extensions are required. In this talk, I'll explore the motivation for the taking that approach and how some of the consequent challenges were solved. I'll trace TBB's lineage to earlier systems, how its work-stealing scheduler evolved over the last five years, and where it might be headed. I'll also compare TBB with OpenMP and other Intel paradigms for parallel programming such as Cilk Plus.

Vamshi Basupal li, Tomofumi Yuki, Sanjay Rajopadhye, Antoine Morvan, Steven Derrien, Patrice Quinton, and D. Wonnacott

Besar Wicaksono, Ramachandra C. Nanjegowda and Barbara Chapman

Dirk Schmidl, Christian Terboven and Dieter An Mey

Jie Tao, Karl Furlinger and Holger Marten

Dheya Mustafa, Aurangzeb and Rudolf Eigenmann

James Lagrone, Ayodunni Aribuki, Cody Addison and Barbara Chapman

James Beyer and Eric Stotzer
Co-chairs, OpenMP Language committee accelerator sub-committee

The two dominate programming models for GPUs, the most common accelerator, are Compute Unified Device Architecture (CUDA) and Open Computing Language (OpenCL), both of these programming models give the programmer the power to extract performance from the accelerator. However, they also require the programmer to significantly rewrite their codes. Rather than require the programmer to rewrite the code, it would be preferable to use a standard set of high level language directives that would help the compiler to generate optimized code to run on the heterogeneous system. The OpenMP language committee accelerator sub-committee is working on a set of OpenMP extensions to support non-cache coherent and cache coherent accelerators. This talk will present some of the work already accomplished by the sub-committee, as well as present some of the major task that remain to be completed.