Measurements from 2026-07-29 to 2026-09-28 show a substantial improvement in Rust compiler performance: mean wall-time dropped 4.57% across 629 benchmarks (555 improved, 74 regressed). Major wins include rustdoc optimizations, Clippy enabled with PGO giving up to 18% wall-time improvements, and an LLVM upgrade to LLVM 23 yielding a 1.2% mean reduction. Enabling the new borrow checker Polonius Alpha introduced regressions for some code (notably serde), but targeted fixes - lazy liveness computation (#161938) and data-structure/inlining tweaks (#163027) - reduced serde instruction counts by 3-5% and produced small wins elsewhere. A broad “sea of green” resulted as many improvements outweighed regressions.
Several other focused changes delivered measurable wins. The new trait solver (nicknamed Pineapple Häagen-Dazs) required extensive tuning; a set of PRs produced dramatic reductions for outlier crates (examples: 50%, 25%, 15%). Contributor xmakro made multiple optimizations (impl handling −1.58% cycles, incremental-load improvements up to −6%, allocation reductions up to −2%, static-dispatch fixes sub-1%). Dataflow work (#160193) changed CFG traversal and cut a cranelift-codegen check build’s time by ~30%; EverInitializedPlaces tuning (#160033) reduced a match-stress benchmark by 17%. Miscellaneous fixes - raising compiler stack size, rollup merges to handle many perf PRs, and a small AST→HIR cleanup - added further sub-percent to few-percent gains. The author will join Hexcat to continue compiler performance work.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.