Magellan: Autonomous Discovery of Novel Compiler Optimization Heuristics with AlphaEvolve
This addresses the high maintenance burden and adaptability issues of human-designed compiler heuristics for software and hardware developers, though it appears incremental as it builds on evolutionary search and LLM agents.
Magellan tackles the challenge of hand-crafted compiler optimization heuristics by autonomously evolving executable C++ decision logic, discovering policies that match or surpass expert baselines, such as outperforming decades of manual engineering in LLVM function inlining for binary-size reduction and performance.
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.