MLGOPerf: An ML Guided Inliner to Optimize Performance

TL;DR

MLGOPerf introduces an ML-guided inliner for LLVM that optimizes performance, achieving up to 2.2% speedup on benchmarks and significantly increasing autotuning opportunities.

Contribution

It presents the first end-to-end framework using ML to optimize performance in LLVM's inliner, employing a secondary model for reward prediction to enhance training efficiency.

Findings

Achieves up to 2.2% speedup on SPEC CPU2006 and Cbench benchmarks.

Provides up to 26% more autotuning opportunities, leading to an additional 3.7% speedup.

Demonstrates practical integration of ML-based optimization in LLVM.

Abstract

For the past 25 years, we have witnessed an extensive application of Machine Learning to the Compiler space; the selection and the phase-ordering problem. However, limited works have been upstreamed into the state-of-the-art compilers, i.e., LLVM, to seamlessly integrate the former into the optimization pipeline of a compiler to be readily deployed by the user. MLGO was among the first of such projects and it only strives to reduce the code size of a binary with an ML-based Inliner using Reinforcement Learning. This paper presents MLGOPerf; the first end-to-end framework capable of optimizing performance using LLVM's ML-Inliner. It employs a secondary ML model to generate rewards used for training a retargeted Reinforcement learning agent, previously used as the primary model by MLGO. It does so by predicting the post-inlining speedup of a function under analysis and it enables a fast training framework for the primary model which otherwise wouldn't be practical. The experimental results show MLGOPerf is able to gain up to 1.8% and 2.2% with respect to LLVM's optimization at O3 when trained for performance on SPEC CPU2006 and Cbench benchmarks, respectively. Furthermore, the proposed approach provides up to 26% increased opportunities to autotune code regions for our benchmarks which can be translated into an additional 3.7% speedup value.