Magellan: Autonomous Discovery of Novel Compiler Optimization Heuristics with AlphaEvolve
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866912857785892864 |
|---|---|
| author | Chen, Hongzheng Novikov, Alexander Vũ, Ngân Alam, Hanna Zhang, Zhiru Grossman, Aiden Trofin, Mircea Yazdanbakhsh, Amir |
| author_facet | Chen, Hongzheng Novikov, Alexander Vũ, Ngân Alam, Hanna Zhang, Zhiru Grossman, Aiden Trofin, Mircea Yazdanbakhsh, Amir |
| contents | Modern compilers rely on hand-crafted heuristics to guide optimization passes. These human-designed rules often struggle to adapt to the complexity of modern software and hardware and lead to high maintenance burden. To address this challenge, we present Magellan, an agentic framework that evolves the compiler pass itself by synthesizing executable C++ decision logic. Magellan couples an LLM coding agent with evolutionary search and autotuning in a closed loop of generation, evaluation on user-provided macro-benchmarks, and refinement, producing compact heuristics that integrate directly into existing compilers. Across several production optimization tasks, Magellan discovers policies that match or surpass expert baselines. In LLVM function inlining, Magellan synthesizes new heuristics that outperform decades of manual engineering for both binary-size reduction and end-to-end performance. In register allocation, it learns a concise priority rule for live-range processing that matches intricate human-designed policies on a large-scale workload. We also report preliminary results on XLA problems, demonstrating portability beyond LLVM with reduced engineering effort. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_21096 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Magellan: Autonomous Discovery of Novel Compiler Optimization Heuristics with AlphaEvolve Chen, Hongzheng Novikov, Alexander Vũ, Ngân Alam, Hanna Zhang, Zhiru Grossman, Aiden Trofin, Mircea Yazdanbakhsh, Amir Artificial Intelligence Machine Learning Programming Languages Modern compilers rely on hand-crafted heuristics to guide optimization passes. These human-designed rules often struggle to adapt to the complexity of modern software and hardware and lead to high maintenance burden. To address this challenge, we present Magellan, an agentic framework that evolves the compiler pass itself by synthesizing executable C++ decision logic. Magellan couples an LLM coding agent with evolutionary search and autotuning in a closed loop of generation, evaluation on user-provided macro-benchmarks, and refinement, producing compact heuristics that integrate directly into existing compilers. Across several production optimization tasks, Magellan discovers policies that match or surpass expert baselines. In LLVM function inlining, Magellan synthesizes new heuristics that outperform decades of manual engineering for both binary-size reduction and end-to-end performance. In register allocation, it learns a concise priority rule for live-range processing that matches intricate human-designed policies on a large-scale workload. We also report preliminary results on XLA problems, demonstrating portability beyond LLVM with reduced engineering effort. |
| title | Magellan: Autonomous Discovery of Novel Compiler Optimization Heuristics with AlphaEvolve |
| topic | Artificial Intelligence Machine Learning Programming Languages |
| url | https://arxiv.org/abs/2601.21096 |