CAREER: Automatically Generating and Processing Program Analyses and Optimizations

University Of California-San Diego, La Jolla CA

Investigators

Abstract

Developing efficient, scalable, correct, and precise program analyzers and optimizers is difficult. There is a long time, often up to a decade, before a new optimizing compiler is mature enough to be widely used. These difficulties hinder the development of new languages and new architectures, and can also discourage end-user programmers from extending compilers with domain-specific checkers or optimizers. Techniques will be investigated for automatically generating efficient, scalable, correct, and precise dataflow analyzers and optimizers from a very high-level specification. The overarching theme is to understand the underlying principles behind designing program analyses and optimizations, and use this understanding to automate as much as possible the analyzer- and optimizer-writing process. Attempting to automate the process of writing analyzers and optimizers enables many new kinds of usage models for compilers, including: allowing end-user programmers to easily extend the compiler with domain-specific checkers or optimizers; allowing end-user programmers to continuously train the compiler, even after it is deployed, based on additional input-output examples; and automatically generating additional analyses when the optimizer discovers the need for new dataflow information, and linking these new analyses into the optimizer while in execution.