Rust compiler speeds up 4.57% in two months
Original: How to speed up the Rust compiler in September 2026
Why This Matters
Faster compile times directly reduce developer iteration cycles across the entire Rust ecosystem.
Nicholas Nethercote's September 2026 report shows the Rust compiler achieved a mean 4.57% wall-time reduction over two months, with 555 of 629 benchmarks improving. Key drivers include PGO for Clippy (up to 18% faster), LLVM 23 upgrade (1.2% mean reduction), and new borrow checker and trait solver work.
Over the period July 29 to September 28, 2026, the Rust compiler saw widespread benchmark improvements that Nethercote calls 'a sea of green.' Jakub Beránek's PR enabling Profile-Guided Optimization for Clippy delivered up to 18% wall-time gains. Nikita Popov's upgrade to LLVM 23 contributed a 1.2% mean wall-time reduction across all benchmarks — modest on paper but significant for a single dependency bump.
Two major new features also landed on Nightly: Polonius Alpha, the new borrow checker, and the new trait solver. Both are more precise but slower in edge cases. Polonius Alpha caused measurable regressions on popular crates like serde; Jack Huey addressed these by making liveness computations lazy, cutting serde instruction counts by 3–5%. The new trait solver regressions are being tracked by Jana Dönszelmann, whose detailed post covers the broad effort. Nethercote himself contributed six PRs with outlier-crate improvements of 15–50%.
New contributor xmakro made four notable PRs: optimizing specialization graph handling (1.58% mean cycle count reduction), improving incremental compilation data loading (up to 6% instruction count reduction), reducing allocations in obligations processing (up to 2%), and switching trait solver selection from dynamic to static dispatch for sub-1% gains across many benchmarks.