Abstract
Optimizing communication is a key issue in compiling data-parallel languages for distributed memory architectures. We examine here the cyclic distribution, and we derive symbolic expressions for communication sets under the only assumption that the initial parallel loop is defined by affine expressions of the indices. This technique relys on unimodular changes of basis. Analysis of the properties of communications leads to a tiling of the local memory addresses that provides maximal message vectorization.
Chapter PDF
Keywords
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
References
R. Asenjo and al. Sparse Block and Cyclick data distribution for matrix computation. In High Performance Computing, Technology and Applications, Elsevier, 95.
C. Ancourt and F. Irigoin. Scanning polyhedra with DO loops. In 3rd ACM Symp. on Principles and Practice of Parallel Programming, pages 39–50, 91.
D. Callahan and K. Kennedy. Compiling programs for distributed memory architectures. Jal. Supercomputing, (2):151–169, Oct. 88.
F. Cappello and al. Balanced distributed memory parallel computers. In 22nd Int. Conf. on Parallel Processing, pages I.72–I.76-, August 93.
S. Chatterjee and al. Generating local addresses and communication sets for data-parallel programs. In Symp on Principles and Practice of Programming Languages 93. ACM, 93.
C.Koelbel. Compile-time generation of regular communication patterns. In Supercomputing 91, pages 101–110, 91.
A. Feldmann, T.M. Stricker, and T.E. Warfel. Supporting sets of arbitrary connections on iWarp through communication context switches. In ACM, editor, 5th ACM Symp. on Algorithms and Architectures, pages 203–212, 93.
F.Irigoin and al. A linear algebra framework for static HPF code distribution. In 4th Int. Work. on Compilers for Parallel Computers, pages 117–132, 93.
M. Le Fur. Scanning parameterized polyhedron using Fourier-Motzkin elimination. Technical report, IRISA, Sept. 94. PI 858.
S.K.S. Gupta and al. On compiling array expressions for efficient execution on distributed-memory machines. In 1993 Int. Conf. on Parallel Processing, pages II-301–II-305, 93.
D.E. Maydan, S.P. Amarasinghe, and M.S. Lam. Data dependence and data-flow analysis of arrays. In 5th Work on Languages and Compilers for Parallel Computing, pages 283–292, 92.
W. Pugh. The omega test: a fast and practical integer programming algorithm for dependence analysis. Comm. ACM, (8):102–114, Aug. 92.
A. Schrijver. Theory of Linear and Integer Programming. Wiley, 86.
J.M. Stichnoth, D. O'Hallaron and T.R. Gross. Generating Communications for Array Statements; Design, Implementation and Evaluation. Jal. Parallele and Distributed computing, (21) 150–159, Apr. 94.
C-W. Tseng. An optimizing Fortran D compiler for MIMD Distributed-memory machines. PhD thesis, Rice University, 93.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1995 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Germain, C., Delaplace, F. (1995). Automatic vectorization of communications for data-parallel programs. In: Haridi, S., Ali, K., Magnusson, P. (eds) EURO-PAR '95 Parallel Processing. Euro-Par 1995. Lecture Notes in Computer Science, vol 966. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0020483
Download citation
DOI: https://doi.org/10.1007/BFb0020483
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-60247-7
Online ISBN: 978-3-540-44769-6
eBook Packages: Springer Book Archive