Compiler-R1: Towards Agentic Compiler Auto-tuning with Reinforcement Learning

Yang Liu (CUHK) · Haoran Luo (Nanyang Technological University) · Libo Zhang (University of Auckland) · Yanjun Wu (Institute of Software, Chinese Academy of Sciences) · Haolin Pan (Institute of Software, Chinese Academy of Sciences) · Hongyu Lin (University of Chinese Academy of Sciences) · Kaichun Yao (Institute of Software, Chinese Academy of Sciences) · Mingjie Xing (Institute of Software, Chinese Academy of Sciences)

compilation environmentcompiler auto-tuningcompiler optimizationenvironment explorationextensive experiments high-quality datasetsinstruction count reductionintermediate representationoutcome-based rewardperformance metricsreasoning datasetsreinforcement learningrl-trained llmstwo-stage training pipeline

Compiler auto-tuning optimizes pass sequences to improve performance metrics such as Intermediate Representation (IR) instruction count. Although recent advances leveraging Large Language Models (LLMs) have shown promise in automating compiler tuning, two significant challenges still remain: the absence of high-quality reasoning datasets for agents training, and limited effective interactions with the compilation environment. In this work, we introduce Compiler-R1, the first reinforcement learning (RL)-driven framework specifically augmenting LLM capabilities for compiler auto-tuning. Compiler-R1 features a curated, high-quality reasoning dataset and a novel two-stage end-to-end RL training pipeline, enabling efficient environment exploration and learning through an outcome-based reward. Extensive experiments across seven datasets demonstrate Compiler-R1 achieving an average 8.46\% IR instruction count reduction compared to opt -Oz, showcasing the strong potential of RL-trained LLMs for compiler optimization. Our code and datasets are publicly available at https://github.com/Panhaolin2001/Compiler-R1.