A Function-Composition Approach to Synthesize Fortran 90 Array Operations

Abstract

An increasing number of programming languages, such as Fortran 90 and APL, are providing a rich set of intrinsic array functions and array expressions. These constructs which constitute an important part of data parallel languages provide excellent opportunities for compiler optimizations. In this paper, we present a new approach to combine consecutive array operations or array expressions into a composite access function of the source arrays. Our scheme is based on the composition of access functions, which is analogous to a composition of mathematic functions. Our new scheme can handle not only data movements of arrays with different numbers of dimensions and with multiple-clause array operations but also masked array expressions and multiple-source array operations. As a result, our proposed scheme is the first synthesis scheme which can collectively synthesize Fortran 90 RESHAPE, EOSHIFT, MERGE, array reduction operations, and WHERE constructs. In addition, we also discuss the case that the synthesis scheme may result in a performance anomaly in the presence of common subexpressions and one-to-many array operations. A solution is proposed to avoid such a performance anomaly. Experimental results show speedups from 1.21 to 2.95 over the base code for code fragments from real applications on a Sequent multiprocessor machine and also show comparable performance improvements on an 8-node SGI Power Challenge by incorporating our proposed optimizations