论文信息 - A New Algorithm for Global Optimization for Parallelism and Locality

A New Algorithm for Global Optimization for Parallelism and Locality

Converting sequential programs to execute on parallel computers is difficult because of the need to globally optimize for both parallelism and data locality. The choice of which loop nests to parallelize, and how, drastically affects data locality. Similarly, data distribution directives, such as DISTRIBUTE in High Performance Fortran (HPF), affects available parallelism and locality. What is needed is a systematic approach to converting programs to parallel form, based upon analysis that identifies opportunities for both parallelism and locality in one representation.

[1] Monica S. Lam,et al. A Loop Transformation Theory and an Algorithm to Maximize Parallelism , 1991, IEEE Trans. Parallel Distributed Syst..

[2] Kevin Smith,et al. Determining Transformation Sequences for Loop Parallelization , 1992, LCPC.

[3] Monica S. Lam,et al. A data locality optimizing algorithm , 1991, PLDI '91.

[4] Ken Kennedy,et al. Optimizing for parallelism and data locality , 1992, ICS '92.

[5] Constantine D. Polychronopoulos,et al. Symbolic analysis for parallelizing compilers , 1996, TOPL.

[6] Barbara M. Chapman,et al. Supercompilers for parallel and vector computers , 1990, ACM Press frontier series.

[7] Ii C. D. Callahan. A global approach to detection of parallelism , 1987 .

[8] Monica S. Lam,et al. Global optimizations for parallelism and locality on scalable parallel machines , 1993, PLDI '93.

[9] J. Ramanujam,et al. Tiling multidimensional iteration spaces for nonshared memory machines , 1991, Proceedings of the 1991 ACM/IEEE Conference on Supercomputing (Supercomputing '91).

[10] John R. Gilbert,et al. The Alignment-Distribution Graph , 1993, LCPC.

[11] Manish Gupta,et al. PARADIGM: a compiler for automatic data distribution on multicomputers , 1993, ICS '93.