Learning to superoptimize programs - Workshop Version

TL;DR

This paper introduces a reinforcement learning-based approach to improve superoptimization by learning proposal distributions, leading to better program optimization results on classic benchmark programs.

Contribution

It presents a novel method that leverages reinforcement learning to learn proposal distributions, enhancing the efficiency of stochastic superoptimization techniques.

Findings

Improved superoptimization results on 'Hacker's Delight' programs

Reinforcement learning-based proposals outperform uniform proposals

Demonstrates the effectiveness of learned proposals in program optimization

Abstract

Superoptimization requires the estimation of the best program for a given computational task. In order to deal with large programs, superoptimization techniques perform a stochastic search. This involves proposing a modification of the current program, which is accepted or rejected based on the improvement achieved. The state of the art method uses uniform proposal distributions, which fails to exploit the problem structure to the fullest. To alleviate this deficiency, we learn a proposal distribution over possible modifications using Reinforcement Learning. We provide convincing results on the superoptimization of "Hacker's Delight" programs.