ABSTRACT
In this paper, we present a method - parallelization spectroscopy - for analyzing the thread-level parallelism available in production High Performance Computing (HPC) codes.We survey a number of techniques that are commonly used for parallelization and classify all the loops in the case study presented using a sensitivity metric: how likely is a particular technique is successful in parallelizing the loop.
- G. Bikshandi et al. Programming for parallelism and locality with Hierarchically Tiled Arrays. In Proceedings of PPOPP 2006, pages 48--57, New York, NY, 2006. Google ScholarDigital Library
- P. Bose. Workload characterization: A key aspect of microarchitecture design. IEEE Micro, 26(2):5--6, 2006. Google ScholarDigital Library
- Chapel. http://chapel.cs.washington.edu.Google Scholar
- CPMD Consortium page. http://www.cpmd.org.Google Scholar
- Project Fortree Overview. http://projectfortress.sun.com.Google Scholar
- A. Kejariwal, X. Tian, W. Li, M. Girkar, S. Kozhukhov, H. Saito, U. Banerjee, A. Nicolau, A. V. Veidenbaum, and C. D. Poly-chronopoulos. On the performance potential of different types of speculative thread-level parallelism. In Proceedings of the 20th ACM International Conference on Supercomputing, pages 24--35, Cairns, Australia, 2006. Google ScholarDigital Library
- M. Kulkarni, K. Pingali, B. Walter, G. Ramanarayanan, K. Bala, and L. P. Chew. Optimistic parallelism requires abstractions. In Proceedings of PLDI 2007, pages 211--222, San Diego, CA, 2007. Google ScholarDigital Library
- LAM MPI Parallel Computing. http://www.lam-mpi.org/.Google Scholar
- (Linear Algebra PACKage). http://www.netlib.org/lapack/.Google Scholar
- OpenMP Specification, version 2.5. http://www.openmp.org.Google Scholar
- UPC. http://upc.gwu.edu/.Google Scholar
- C. von Praun, R. Bordawekar, and C. Cas¸caval. Modeling optimistic concurrency using quantitative dependence analysis. In Proceedings of PPOPP 2008, pages 185--196, Salt Lake City, UT, 2008. Google ScholarDigital Library
- M. J. Wolfe. More iteration space tiling. In Proceedings of Supercomputing '89, Nov. 1989. Google ScholarDigital Library
- P. Wu, A. Kejariwal, and C. Cas¸caval. Compiler-driven dependence profiling to guide program parallelization. In Proceedings of the 21st International Workshop on Languages and Compilers for Parallel Computing, Alberta, Canada, 2008. Google ScholarDigital Library
- The X10 Programming Language. http://www.x10-lang.org.Google Scholar
Index Terms
- Parallelization spectroscopy: analysis of thread-level parallelism in hpc programs
Recommendations
Parallelization spectroscopy: analysis of thread-level parallelism in hpc programs
PPoPP '09In this paper, we present a method - parallelization spectroscopy - for analyzing the thread-level parallelism available in production High Performance Computing (HPC) codes.We survey a number of techniques that are commonly used for parallelization and ...
The Cilkview scalability analyzer
SPAA '10: Proceedings of the twenty-second annual ACM symposium on Parallelism in algorithms and architecturesThe Cilkview scalability analyzer is a software tool for profiling, estimating scalability, and benchmarking multithreaded Cilk++ applications. Cilkview monitors logical parallelism during an instrumented execution of the Cilk++ application on a single ...
Exploiting Instruction- and Data-Level Parallelism
Historically, there have been two different approaches to high performance computing: instruction-level parallelism (ILP) and data-level parallelism (DLP). The ILP paradigm seeks to execute several instructions each cycle by exploring a sequential ...
Comments