"The Current State and Future of Rust Compiler Performance"
summary
Rust compiler performance has doubled over recent releases through many small optimizations and several larger efforts: parallel-rustc landed in nightly with about a 30% average speedup, profile-guided optimizations yield 10–20% PGO gains, and switching to LLD on Linux cuts link time by up to 40%. Work includes redesigned hybrid data structures that keep single-thread loss to ~1–3%, per-thread allocators, and tooling to measure regressions; remaining problems include incremental compilation lagging C++, work-stealing deadlocks, and crate-level caching.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:45Compiler performance challenges
- 3:15Benchmarking compilation speed
- 4:21Community performance efforts
- 6:52Linker optimizations reducing time
- 8:41Code-level optimizations and limits
- 10:54Parallelizing compiler frontend
- 13:25Thread-safe data structures
- 15:43Multi-threaded memory allocation
- 17:09Hash table optimizations
- 18:40Parallel compilation challenges
- 19:47Future linking and incremental builds
- 21:39Future compiler design ideas
- 22:21Closing remarks
