SC14 WACCPD Workshop on Accelerator Programming Using Directives

Call For Papers for the SC14 Workshop on Accelerator Programming Using Directives (WACCPD), which  brings together leading researchers and software designers at the forefront of the application of high-level directives to program accelerator-based architectures. Using directives improve productivity, and program portability with minimal changes to the applications while achieving good power-efficiency and performance. HPC researchers and programmers (including Energy, Climate, Oil & Gas, Computational Chemistry, and Machine Learning) wishing to tap into the performance benefits of commodity accelerators are starting to use directives to program accelerators including OpenACC, OpenMP  and other similar programming interfaces vigorously to tap into commodity accelerators and speed up applications.

The goal of this workshop is to bring together users, vendors, and tools providers to share their knowledge and experiences to program heterogeneous systems using directive-based programming interfaces such as OpenACC, OpenMP, etc to achieve good performance.  This year’s workshop will  also emphasize in the future direction of accelerator programming using directives and how we can address better the user and tools-ecosystem needs. Although, this is a workshop for accelerator programming using directives, we welcome ideas used for other languages/APIs that may be used to interoperate with directives, used to translate the directives, or target the directives.

CALL FOR PAPERS

Paper Submission Deadline August 22nd, 2014

Website: http://www.openacc.org/WACCPD14

Submission topics

Topics of interest for workshop submissions include (but are not limited to):

• Application experiences in any scientific domain using directives
• Compiler and runtime implementations of accelerator directives
• Extensions to and shortcomings of current accelerator directives APIs
• Experiences with hybrid heterogeneous or many-core programming using accelerator directives and other  models  (i.e.OpenMP, MPI, OpenSHMEM)
• Interoperability of Scientific libraries with accelerator directives
• Implementation experiences of accelerator directives on new architectures
• Performance evaluation and studies
• Extending the HPC tool chain (such as profiling, debugging, etc) to support directives
• Modeling and performance analysis tools
• Auto-tuning or optimization strategies
• Benchmarks and validation suites

Papers Submission Guidelines:

Papers should present original research and should provide sufficient background material to make them accessible to the broader community.

Format: Submissions are limited to 10 pages in the IEEE format (see link). The 10-page limit includes figures, tables, and appendices, but does not include references, for which there is no page limit.

Submissions can be made through EasyChair: (see link).

Proceedings will be published by ACM, and will be available to SIGHPC members through the ACM Digital Library. Authors of accepted papers and posters will be required to sign the ACM copyright form. Instructions for preparing papers for the proceedings will be emailed to authors of accepted papers.

Program Chair and Co-Chairs

Sunita Chandrasekaran, University of Houston

Fernanda Foertter, ORNL

Oscar Hernandez, ORNL

Steering Committee:

Barbara Chapman (University of Houston, cOMPunity)

Satoshi Matsuoka (Titech)

Duncan Poole (OpenACC)

Thomas Schulthess (CSCS)

Jeffrey Vetter (ORNL)

Program Committee

Important Deadlines:

Submission deadline: August 22nd, 2014 (Midnight 12:00 Pacific TimeZone)

Author notification: September 22th, 2014

Panel Session   

A panel session dedicated to discussion of current open problems, drawbacks and future evolutions of high-level directive approaches. A roadmap for accelerator directives will also be discussed.

Best Paper Award   

The papers will be peer-reviewed where invited submissions will be candidates for a best paper award. Criterion for selection will be based on originality, scientific impact, utility, and quality of presentation.