TL;DR
Get monitors, keyboards and dev gear delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
A September 30 report by Rust compiler contributor Nicholas Nethercote says mean wall time across the compiler’s benchmarks fell 4.57% between July 29 and September 28, 2026. The gains came from multiple changes, including Clippy PGO, an LLVM 23 update and compiler optimizations; some new compiler components still slow a minority of workloads.
Rust compiler benchmarks recorded a 4.57% mean reduction in wall time between July 29 and September 28, according to compiler contributor Nicholas Nethercote, who published the results on September 30. Of 629 benchmark measurements, 555 improved and 74 regressed, a broad performance gain built from changes across the compiler rather than a single release or optimization.
Nethercote described the period’s results as “a sea of green” and said several benchmarks had double-digit percentage reductions. The figures summarize benchmark measurements over the stated two-month window; they do not mean every Rust project or compile experienced a 4.57% speedup. Results vary by workload, and the report does not provide a single expected improvement for all users.
One contributor change enabled profile-guided optimization (PGO) for Clippy. Pull request #159642, by Jakub Beránek, improved wall time across most Clippy benchmarks, with the best case reaching 18%, Nethercote reported. A separate update to LLVM 23, in pull request #158734 by Nikita Popov, lowered mean wall time across all benchmarks by 1.2%.
Other changes targeted specific compiler workloads. Jack Huey’s work on Polonius Alpha reduced instruction counts for the popular serde crate by 3% to 5% in one change. Nethercote also reported an approximately 30% wall-time reduction for a check build of cranelift-codegen after a dataflow-analysis traversal change, and a 17% instruction-count reduction on the match-stress benchmark from another optimization. These figures refer to different measures and workloads, so they should not be treated as directly comparable.
Where Rust Compile Time Fell
The results matter to developers because compiler speed affects the wait between editing code and testing it, as well as the cost of building software in continuous-integration systems. A mean reduction across a broad benchmark set suggests that several independent improvements added up, including work on linting, code generation, dataflow analysis and allocation-heavy paths.
The workload-specific results also show why compiler performance is tracked across many benchmarks. An optimization can make a substantial difference to one crate while barely affecting another. For teams using affected workloads, a targeted gain such as the reported cranelift-codegen reduction may be more relevant than the overall mean; others may see smaller changes or none. The report does not establish the precise effect for any individual project.
As an affiliate, we earn on qualifying purchases.
New Compiler Components Add Trade-Offs
The measurement period included development work on Polonius Alpha, a new borrow checker enabled on Rust Nightly. Nethercote said it is more precise than the existing checker and accepts some valid programs the older checker would reject. It also performs more work in a minority of cases, adding measurable compile time for some workloads, including serde. The optimizations described in the report address part of that cost, but Nethercote said further work remains.
A new trait solver was also enabled on Nightly. Nethercote said it, too, is slower in a minority of cases, while ongoing efforts have produced large compile-time improvements for certain outlier crates and a stress test. The source names the solver “Penelope Hammertime” but flags that name with an editorial question; the report does not resolve whether it is correct. These Nightly developments are separate from a claim that every stable Rust user has received the same changes or performance results.
The monthly performance report builds on work by multiple contributors. In addition to the named pull requests, Nethercote credited new contributor xmakro with changes to specialization-graph construction, incremental compilation data and allocation-heavy processing. The LLVM update and the individual compiler optimizations have distinct benchmark effects; the reported 4.57% is the overall mean for the period, not a simple sum of the highlighted changes.
“The mean wall-time reduction was 4.57%, which is a remarkable improvement in just two months.”
— Nicholas Nethercote, Rust compiler contributor
Rust programming language development hardware
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Workloads Still Show Regressions
The report gives aggregate benchmark results and selected examples, but does not specify in the supplied material the full benchmark methodology, the size or composition of each workload, or the practical effect on a particular user’s build. The 4.57% figure is a mean across measurements, not a guaranteed reduction in compile time for every crate, machine or build configuration.
Some regressions remain, particularly for a minority of workloads using the new borrow checker or trait solver. The report does not quantify the total number or severity of those regressions, nor does it state when all known issues will be addressed. It also describes compiler changes and Nightly features; the supplied information does not establish which changes are available in a stable Rust release.
As an affiliate, we earn on qualifying purchases.
Further Solver and Checker Tuning
Nethercote said work is continuing to reduce the remaining performance costs associated with Polonius Alpha and the new trait solver. Readers can track the specific compiler changes through the pull requests cited in the report and review the related technical posts by Noah Lev on rustdoc performance and Jana Dönszelmann on the new solver.
The next meaningful check will be how later benchmark reports measure these changes across additional workloads and whether the improvements reach release channels used by developers. The September report does not give a release timetable or promise a particular speedup in a future Rust version; those details remain open.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much faster did the Rust compiler get in this report?
The benchmark set recorded a 4.57% mean reduction in wall time from July 29 to September 28, 2026. That is an average across the measurements, not a speedup guaranteed for every project.
How many benchmark measurements improved?
Of 629 measurements, 555 improved and 74 regressed, according to Nicholas Nethercote’s September 30 report.
Which changes delivered the reported gains?
The report highlights Clippy PGO, an LLVM 23 update and optimizations to compiler analyses and data structures. It also describes workload-specific changes, including a roughly 30% wall-time reduction for a cranelift-codegen check build.
Will every Rust developer see a 4.57% improvement?
No such guarantee is reported. The figure is a mean across benchmarks, and the impact depends on the code, build and compiler configuration. The report also says some workloads regressed.
Are the new borrow checker and trait solver free of performance costs?
No. Nethercote said both are slower in a minority of cases, although ongoing optimizations have reduced costs for some workloads. Further performance work remains, and the report does not give a schedule for resolving all regressions.
Source: hn
Evergreen bestsellers Picks
bestsellers
As an affiliate, we earn on qualifying purchases.
