Compiling issues for supercomputers
Milind Girkar, C. Polychronoupoulos · Conference on High Performance Computing (Supercomputing) · 1988
Accurate and fast methods for computing data dependencies are vital to the efficiency of vectorizing and parallelizing compilers. Program transformations employed by these compilers are effective only if dependencies are computed as accurately as possible. In this paper we consider two issues for supercomputer compilers. First, we discuss in detail exact and inexact tests for computing data dependencies and focus on the linearization scheme. We then discuss a program transformation call loop spreading which allows us to execute in parallel adjacent loops with interloop dependencies.