DialEgg: Dialect-Agnostic MLIR Optimizer using Equality Saturation with Egglog
MLIR’s ability to optimize programs at multiple levels of abstraction is key to enabling domain-specific optimizing compilers. However, expressing optimizations remains tedious and error-prone. Moreover, optimizations can interact in unexpected ways, making it hard to unleash full performance. Equality saturation is an optimization technique that promises to solve these challenges. First, it simplifies the expression of optimizations using rewrite rules. Secondly, it considers all possible optimization interactions, through saturation, selecting the best possible program variant. Despite these advantages, equality saturation remains absent from production/research-ready compilers such as MLIR. This paper proposes to integrate Egglog, a recent equality saturation engine, with MLIR, in a dialect-agnostic manner. This paper shows how the main MLIR constructs such as operations, types, or attributes can be modeled in Egglog. It also presents DialEgg, a tool that pre-defined a large set of common MLIR constructs in Egglog, and automatically translates between the MLIR and Egglog program representations. Using a few use-cases, this paper demonstrates the potential for combining equality saturation and MLIR.
Mon 3 MarDisplayed time zone: Pacific Time (US & Canada) change
17:00 - 18:00 | |||
17:00 20mTalk | DialEgg: Dialect-Agnostic MLIR Optimizer using Equality Saturation with Egglog Main Conference Pre-print | ||
17:20 20mTalk | Combining MLIR Dialects with Domain-Specific Architecture for Efficient Regular Expression Matching Main Conference Andrea Somaini Politecnico di Milano, Filippo Carloni Politecnico di Milano, Giovanni Agosta Politecnico di Milano, Italy, Marco D. Santambrogio Politecnico di Milano, Davide Conficconi Politecnico di Milano | ||
17:40 20mTalk | The MLIR Transform Dialect - Your compiler is more powerful than you think Main Conference Martin Lücke University of Edinburgh, Michel Steuwer Technische Universität Berlin, Albert Cohen Google DeepMind, William S. Moses University of Illinois Urbana-Champaign, Alex Zinenko Google DeepMind |