Overview Technical Papers Tutorials Panels Posters  
  Masterworks BOFs GRID 2004 Workshops    
  SC Global HPC Software Challenge MSI Outreach
 


HPC Software Challenge

Submission deadline: July 26, 2004
Submission website: http://www.sc-submissions.org

Call for Proposals
As part of the SC2004 theme of "Bridging Communities", the HPC Challenge has been adapted to bridge the communities of HPC and Software Engineering. This HPC Software Challenge will honor participants working to improve the productivity of HPC software developers and the quality of HPC software. The simulations and computations that drive HPC are becoming more complex, as are the HPC systems that run them. Software development can become as restrictive a bottleneck as computational intensity. "Coding" must evolve into software engineering to enable software that is not only computationally efficient but also maintainable, extensible, and verifiable.

Submissions. HPC software developers are invited to submit case studies of development processes, strategies, or practices that they have applied to improve developer productivity and/or software quality. Submissions may cover anything from full life-cycle processes to simple, high-impact techniques, in areas including but not limited to the following:

Project management
Requirements definition, analysis, and management
Configuration management
Software specification, architecture, and design
Implementation (coding standards, language subsets)
Verification (code inspection, testing, formal proof)
Productivity and quality metrics (collection and analysis)

Submissions are made online at http://www.sc-submissions.org/. The submission deadline is July 26, 2004. Please include discussion of the following in the submission.

1. The HPC software being developed. Application or library? What does it do? How big is the project? How big is the user base? What is the level of parallelism and scalability?
2. The process, strategy, or practice you employed to improve productivity and software quality. How was it implemented? How much effort did it take? What were the expected improvements?
3. The resulting improvements. Did productivity and/or software quality improve? If so, by how much? Did other problems arise?

Presentations. The submissions will be reviewed by the HPC Software Challenge Committee. Selected submitters will be asked to make a presentation in the HPC Software Challenge session of the Technical Program at SC2004. These presentations will be judged by experts in the fields of software engineering and software development for HPC. Up to two awards of $500 each will be given for presentations considered to have the largest positive impact on HPC software productivity and/or quality. Note that the largest positive impact may come by disseminating negative results for a given method. The following factors will be considered.

Improvement Achieved. The documented extent of the effect on productivity and/or software quality.
Innovation. The documented effect from the use of innovative methods.
Return on Investment. Documented benefits achieved relative to the cost of implementing the improvement. (This factor will be ignored when negative results are being reported.)
Credibility. The extent to which results are convincing because they are based on a sound and quantitative analytical design.

HPC Software Challenge Committee
John Goodenough, Software Engineering Institute
Jeremy Kepner, Massachusetts Institute of Technology
Jesse Poore, University of Tennessee
Douglass Post, Los Alamos National Laboratory
Tom Swain, University of Tennessee
James White III (Trey), Oak Ridge National Laboratory, Chair

Questions? Please send email to hpc-challenge04@sc-conference.org.


Problems with this page? Please contact our webmaster.