November 2006 A
High Productivity Computing Systems and the Path Towards Usable Petascale Computing
Andrew Funk, MIT Lincoln Laboratory
Victor Basili, University of Maryland, College Park
Lorin Hochstein, University of Nebraska, Lincoln
Jeremy Kepner, MIT Lincoln Laboratory

Conclusions and Future Work

We have previously introduced a common metric for measuring relative development time productivity of HPC software development. The RDTP metric has been applied to data from benchmark codes and classroom experiments, with consistent trends for various programming models.

In general the results support the theory that traditional HPC programming models such as MPI yield good speedup but require more relative effort than other implementations (Figure 5). OpenMP generally provides speedup comparable to MPI, but requires less effort. This leads to higher values of the RDTP metric. There are questions of scalability with regard to OpenMP that are not addressed by this study.

The pMatlab implementations of HPC Challenge provide an example of a language that can yield good speedup for some problems, while requiring less relative effort, again leading to higher values of the RDTP metric.

Further classroom experiments are planned, and as more data is collected it will be analyzed in the same manner, to see if other trends emerge. Our current work is focused on designing a standard framework for automating the collection, storage, and analysis of workflow data (see “Modeling HPC workflows with timed Markov models” in this issue).

Figure 5

Figure 5. Speedup and Relative Effort are both important indicators of time to solution.
We wish to thank all of the professors whose students participated in this study, including Jeff Hollingsworth, Alan Sussman, and Uzi Vishkin of the University of Maryland, Alan Edelman of MIT, John Gilbert of UCSB, Mary Hall of USC, and Allan Snavely of UCSD.
This work is sponsored by Defense Advanced Research Projects Administration, under Air Force Contract F19628-00-C-0002. Opinions, interpretations, conclusions and recommendations are those of the author and are not necessarily endorsed by the United States Government.
1High Productivity Computer Systems - www.HighProductivity.org/
2Kepner, J. “HPC Productivity Model Synthesis.” IJHPCA Special Issue on HPC Productivity, Vol. 18, No. 4, SAGE 2004.
3Funk, A., Basili, V., Hochstein, L., Kepner, J. “Application of a Development Time Productivity Metric to Parallel Software Development.” Second International Workshop on Software Engineering for High Performance Computing Systems Applications. St. Louis, Missouri. May 15, 2005.
4Funk, A., Kepner, J., Basili, V., Hochstein, L. "A Relative Development Time Productivity Metric for HPC Systems." Ninth Annual Workshop on High Performance Embedded Computing. 20 – 22 September 2005, Lexington, MA.
5Wheeler, D. SLOCcount. www.dwheeler.com/sloccount/
6Choy, R., Edelman, A. MATLAB*P 2.0: A unified parallel MATLAB. MIT DSpace, Computer Science collection, Jan. 2003. hdl.handle.net/1721.1/3687
7NAS Parallel Benchmarks - www.nas.nasa.gov/Software/NPB/
8ZPL - www.cs.washington.edu/research/zpl/home/
9Haney, R. et. al. “pMatlab Takes the HPC Challenge.” Poster presented at High Performance Embedded Computing (HPEC) workshop, Lexington, MA. 28-30 Sept. 2004.
10HPC Challenge - icl.cs.utk.edu/hpcc/, www.hpcchallenge.org/

Pages: 1 2 3 4 5 6

Reference this article
Funk, A., Basili, V., Hochstein, L., Kepner, J. "Analysis of Parallel Software Development using the Relative Development Time Productivity Metric," CTWatch Quarterly, Volume 2, Number 4A, November 2006 A. http://www.ctwatch.org/quarterly/articles/2006/11/analysis-of-parallel-software-development-using-the-relative-development-time-productivity-metric/

Any opinions expressed on this site belong to their respective authors and are not necessarily shared by the sponsoring institutions or the National Science Foundation (NSF).

Any trademarks or trade names, registered or otherwise, that appear on this site are the property of their respective owners and, unless noted, do not represent endorsement by the editors, publishers, sponsoring institutions, the National Science Foundation, or any other member of the CTWatch team.

No guarantee is granted by CTWatch that information appearing in articles published by the Quarterly or appearing in the Blog is complete or accurate. Information on this site is not intended for commercial purposes.