Research Papers
Papers and reading notes from my compiler autotuning FYP. The project is finished, and I am not currently working through a reading list.
Papers Used in the FYP
GroupTuner: Efficient Group-Aware Compiler Auto-Tuning
Gao et al., 2025 · Paper · Code
This was the starting point of the project. I read the paper and reproduced its GCC experiments before moving on to my own Clang framework.
Reading notes · 中文笔记 · Reproduction and FYP follow-up
A Survey on Compiler Autotuning using Machine Learning
Ashouri et al. · Paper
A background reference in the report, covering the wider range of compiler autotuning methods. I referred to it when placing the project in context.
Efficient Compiler Autotuning via Bayesian Optimization
Chen et al., ICSE 2021 · Paper
The BOCA paper. I adapted the approach for comparison with my GA on the same Clang flag space. The comparison used a single seed, so the small differences between methods need further evaluation.
Earlier Reading Ideas
These were on the original list. I am keeping the titles here for reference, rather than treating them as an active queue or marking them all as completed.
- MILEPOST GCC: Machine Learning Based Research Compiler
- Compiler Auto-tuning through Multiple Phase Learning
- SRTuner: Effective Compiler Optimization Customization by Exposing Synergistic Relations
- Compiler Auto-Tuning via Critical Flag Selection
- End-to-end Deep Learning of Optimization Heuristics (DeepTune)
- CompilerGym: Robust, Performant Compiler Optimization Environments for AI Research
- Large Language Models for Compiler Optimization
The project overview summarises where the FYP ended up.
published: and updated: