Shoaib Kamil
Armando Fox
ABSTRACT
Domain-expert productivity programmers desire scalable application performance, but usually must rely on efficiency programmers who are experts in explicit parallel programming to achieve it. Since such programmers are rare, to maximize reuse of their work we propose encapsulating their strategies in mini-compilers for domain-specific embedded languages (DSELs) glued together by a common high-level host language familiar to productivity programmers. The nontrivial applications that use these DSELs perform up to 98% of peak attainable performance, and comparable to or better than existing hand-coded implementations. Our approach is unique in that each mini-compiler not only performs conventional compiler transformations and optimizations, but includes imperative procedural code that captures an efficiency expert's strategy for mapping a narrow domain onto a specific type of hardware. The result is source- and performance-portability for productivity programmers and parallel performance that rivals that of hand-coded efficiency-language implementations of the same applications. We describe a framework that supports our methodology and five implemented DSELs supporting common computation kernels.
Our results demonstrate that for several interesting classes of problems, efficiency-level parallel performance can be achieved by packaging efficiency programmers' expertise in a reusable framework that is easy to use for both productivity programmers and efficiency programmers.
- A. Buluc and J. R. Gilbert. The combinatorial BLAS: Design, implementation, and applications. Technical Report UCSB-CS-2010-18, University of California, Santa Barbara, 2010.Google Scholar
- H. Cook, E. Gonina, S. Kamil, G. Friedland, and D. P. A. Fox. Cuda-level performance with python-level productivity for gaussian mixture model applications. In 3rd USENIX conference on Hot topics in parallelism (HotPar'11), Berkeley, CA, USA, 2011. Google ScholarDigital Library
- P. Hudak. Building domain-specific embedded languages. ACM Comput. Surv., 28:196, December 1996. ISSN 0360-0300. doi: http://doi.acm.org/10.1145/242224.242477. Google ScholarDigital Library
- G. Malewicz, M. Austern, A. Bik, J. Dehnert, I. Horn, N. Leiser, and G. Czajkowski. Pregel: A system for large-scale graph processing. SIGMOD, Jun 2010. Google ScholarDigital Library
- M. Mohiyuddin, M. Hoemmen, J. Demmel, and K. Yelick. Minimizing communication in sparse matrix solvers. In Supercomputing 2009, Portland, OR, Nov 2009. Google ScholarDigital Library
- S. Williams, A. Waterman, and D. A. Patterson. Roofline: an insightful visual performance model for multicore architectures. Commun. ACM, pages 65--76, 2009. Google ScholarDigital Library
Index Terms
- Portable parallel performance from sequential, productive, embedded domain-specific languages
Recommendations
Portable parallel performance from sequential, productive, embedded domain-specific languages
PPOPP '12Domain-expert productivity programmers desire scalable application performance, but usually must rely on efficiency programmers who are experts in explicit parallel programming to achieve it. Since such programmers are rare, to maximize reuse of their ...
Development of internal domain-specific languages: design principles and design patterns
PLoP '11: Proceedings of the 18th Conference on Pattern Languages of ProgramsA great part of software development challenges can be solved by one universal tool: Abstraction. Developers solve development challenges by using expressions and concepts that abstract from too technical details. One especially supportive tool for ...
Comments