Optimizing Stateful Dataflow with Local Rewrites

Autor:	Laddad, Shadaj, Power, Conor, Hou, Tyler, Cheung, Alvin, Hellerstein, Joseph M.
Rok vydání:	2023
Předmět:	Computer Science - Programming Languages Computer Science - Distributed Parallel and Cluster Computing
Druh dokumentu:	Working Paper
Popis:	Optimizing a stateful dataflow language is a challenging task. There are strict correctness constraints for preserving properties expected by downstream consumers, a large space of possible optimizations, and complex analyses that must reason about the behavior of the program over time. Classic compiler techniques with specialized optimization passes yield unpredictable performance and have complex correctness proofs. But with e-graphs, we can dramatically simplify the process of building a correct optimizer while yielding more consistent results! In this short paper, we discuss our early work using e-graphs to develop an optimizer for a the Hydroflow dataflow language. Our prototype demonstrates that composing simple, easy-to-prove rewrite rules is sufficient to match techniques in hand-optimized systems. Comment: EGRAPHS 2023
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2306.10585 Zobrazit plný text záznamu View this record from Arxiv