The field of compiler optimization and testing is experiencing significant advancements with the integration of Large Language Models (LLMs). Researchers are leveraging LLMs to detect missed peephole optimizations, improve fuzzing techniques for FPGA logic synthesis compilers, and enhance compiler testing frameworks. These innovations have the potential to greatly improve the efficiency and effectiveness of compiler optimization and testing. Notably, the use of LLMs is enabling the detection of previously unknown bugs and optimizations, which is critical for ensuring the reliability and performance of software systems.

Some noteworthy papers in this area include: Leveraging Large Language Models to Detect Missed Peephole Optimizations, which proposes a novel automated framework that combines LLMs with rigorous correctness verification to detect missed peephole optimizations. Interleaving Large Language Models for Compiler Testing, which presents a novel compiler testing framework that decouples the testing process into offline and online phases, generating high-quality test programs by strategically combining small code pieces. Boosting Skeleton-Driven SMT Solver Fuzzing by Leveraging LLM to Produce Formula Generators, which introduces a novel LLM-assisted fuzzing framework that synthesizes reusable term generators to produce syntactically valid and semantically diverse formulas.