Affine Transformations of Loop Nests for Parallel Execution and Distribution of Data over Processors

TL;DR

This paper presents a method for transforming affine loop nests to optimize parallel execution and data distribution on distributed memory systems, reducing communication and enhancing data locality.

Contribution

It introduces a novel approach for affine transformations of loop nests to improve parallelism and data distribution with minimized communication overhead.

Findings

Method reduces inter-processor communication.

Enhances data locality within processors.

Provides conditions for effective broadcast arrangements.

Abstract

The paper is devoted to the problem of mapping affine loop nests onto distributed memory parallel computers. A method to find affine transformations of loop nests for parallel execution and distribution of data over processors is presented. The method tends to minimize the number of communications between processors and to improve locality of data within one processor. A problem of determination of data exchange sequence is investigated. Conditions to determine the ability to arrange broadcast is presented.