Abstract
We present a cache locality optimization technique that can opitimize a loop nest even if the arrays referenced have different layouts in memory. Such a capability is required for a global locality optimization framework that applies both loop and data transformation to a sequence of loop nests for optimizing locality. Our method finds a non-singular iteratio-space transformation matrix such that in a given loop nest spatial locality is exploited in the innermost loops where it is most useful. The method builds inverse of a non-singular transformation matrix column-by-column starting from the rightmost column. In addition, our approach can work in those eases where the data layouts of a subset of the referenced arrays is unknown. Experimental results on an 8-processor SGI Origin 2000 show that our technic reduces execution times by up to 72%.
M. Kandemir and A. Choudhary were supported by NSF Young Investigator Award CCR-9357840, NSF grant CCR-9509143 and Air Force contract F30602-97-C-0026. J. Ramanujam was supported by NSF Young Investigator Award CCR-9457768. P. Banerjee was supported by NSF grant CCR-9526325 and by DARPA contract DABT-63-97-C-0035.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
A. Abu-Sufah, D. J. Kuck, and D. H. Lawrie. On the performance enhancement of paging systems through program analysis and transformations. IEEE Trans. Comp., C-30(5):341–356, 1981.
M. Cierniak and W. Li. Unifying data and control transformations for distributed shared memory machines. Proc. SIGPLAN Conf. Prog. Lang. Des. & Imp., June 1995.
M. Kandemir, A. Choudhary, J. Ramanujam, and P. Banerjee. A Matrix-Based Approach to the Global Locality Optimization Problem In Proc. 1998 Int. Conf. Parallel Architectures & Compilation Techniques (PACT 98), October 1998.
M. Kandemir, A. Choudhary, N. Shenoy, P. Banerjee, and J. Ramanujam. A hyperplane based approach for optimizing spatial locality in loop nests. In Proc. 12th ACM Int. Conf. Supercomputing, July 1998.
M. Kandemir, J. Ramanujam, and A. Choudhary. A compiler algorithm for optimizing locality in loop nests. In Proc. 11th ACM Int. Conf. Supercomputing, pp. 269–276, July 1997.
I. Kodukula, N. Ahmed, and K. Pingali. Data-centric multi-level blocking. In Proc. SIGPLAN Conf. Prog. Lang. Des. & Imp., June 1997.
S.-T. Leungand J. Zahorjan. Optimizing data locality by array restructuring. Technical Report TR 95-09-01, CSE Dept., University of Washington, Sep. 1995.
W. Li. Compilingfor NUMA parallel machines. Ph.D. Thesis, Cornell University, 1993.
K. McKinley, S. Carr, and C. W. Tseng. Improving data locality with loop transformations. ACM Transactions on Programming Languages and Systems, 1996.
M. O’Boyle and P. Knijnenburg. Non-singular data transformations: Definition, validity, applications. In Proc. 6th Workshop on Compilers for Par. Comp., pp. 287–297, Germany, 1996.
J. Ramanujam and P. Sadayappan. Compile-time techniques for data distribution in distributed memory machines. In IEEE Trans. Par. & Dist. Sys., 2(4):472–482, Oct. 1991.
A. Schrijver. Theory of linear and integer programming, John Wiley, 1986.
M. Wolf and M. Lam. A data locality optimizing algorithm. In Proc. ACM SIGPLAN 91 Conf. Programming Language Design and Implementation, pp. 30–44, June 1991.
M. Wolfe. High performance compilers for parallel computing, Addison Wesley, 1996.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Kandemir, M., Ramanujam, J., Choudhary, A., Banerjee, P. (1999). A Loop Transformation Algorithm Based on Explicit Data Layout Representation for Optimizing Locality. In: Chatterjee, S., et al. Languages and Compilers for Parallel Computing. LCPC 1998. Lecture Notes in Computer Science, vol 1656. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-48319-5_3
Download citation
DOI: https://doi.org/10.1007/3-540-48319-5_3
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-66426-0
Online ISBN: 978-3-540-48319-9
eBook Packages: Springer Book Archive