A data transformations based approach for optimizing memory and cache locality on distributed memory multiprocessors

Authors:
Xia Jun;Xue-Jun Yang
Affiliations:
School of Computer Science, National University of Defense Technology, Hunan, China;School of Computer Science, National University of Defense Technology, Hunan, China
Venue:
APPT'05 Proceedings of the 6th international conference on Advanced Parallel Processing Technologies
Year:
2005

Citing 9
Cited 0

A data locality optimizing algorithm

PLDI '91 Proceedings of the ACM SIGPLAN 1991 conference on Programming language design and implementation
Unifying data and control transformations for distributed shared-memory machines

PLDI '95 Proceedings of the ACM SIGPLAN 1995 conference on Programming language design and implementation
Improving data locality with loop transformations

ACM Transactions on Programming Languages and Systems (TOPLAS)
A hyperplane based approach for optimizing spatial locality in loop nests

ICS '98 Proceedings of the 12th international conference on Supercomputing
A matrix-based approach to global locality optimization

Journal of Parallel and Distributed Computing - Special issue on compilation and architectural support for parallel applications
Automatic memory layout transformations to optimize spatial locality in parameterized loop nests

ACM SIGARCH Computer Architecture News - Special issue on interaction between compilers and computer architectures
Static and Dynamic Locality Optimizations Using Integer Linear Programming

IEEE Transactions on Parallel and Distributed Systems
Skewed Data Partition and Alignment Techniques for Compiling Programs on Distributed Memory Multicomputers

The Journal of Supercomputing
Communication-Free Alignment for Array References with Linear Subscripts in Three Loop Index Variables or Quadratic Subscripts

The Journal of Supercomputing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Data locality is one of the key factors in affecting the performance of parallel programs running on distributed memory multiprocessors. This paper presents an approach for optimizing memory locality and cache locality of perfect or non-perfect loop nests using linear data transformations on distributed memory multiprocessors. The approach optimizes memory locality with the data space fusion technique and cache locality with the projection-delamination technique, and combines the both techniques effectively to make the overheads of remote memory accesses and local memory accesses as low as possible. We conduct experiments with nine programs and the results show the approach is effective in optimizing memory locality and cache locality simultaneously.