Controlling the Size of Supercompiled Programs using Multi-result Supercompilation

TL;DR

This paper introduces a method combining multi-result supercompilation with a generalization strategy to control program size, achieving significant optimization while preventing code bloat.

Contribution

It presents a novel approach that effectively manages the size of supercompiled programs, addressing unpredictability in existing supercompilation techniques.

Findings

Results show small program sizes with powerful optimizations

Method avoids code duplication effectively

Early experiments are promising

Abstract

Supercompilation is a powerful program transformation technique with numerous interesting applications. Existing methods of supercompilation, however, are often very unpredictable with respect to the size of the resulting programs. We consider an approach for controlling result size, based on a combination of multi-result supercompilation and a specific generalization strategy, which avoids code duplication. The current early experiments with this method show promising results -- we can keep the size of the result small, while still performing powerful optimizations.