|
| 1 | +# 2021-05-25 Triage Log |
| 2 | + |
| 3 | +A somewhat quiet week. Some PRs had performance runs performed on them, but the changes were merged despite this. Also, we still have issues with certain benchmarks being noisy. |
| 4 | + |
| 5 | +Triage done by **@rylev**. |
| 6 | +Revision range: [25a277f03df7e44643ddfcc240d034409cb2f505..cdbe2888979bb8797b05f0d58a6f6e60753983d2](https://perf.rust-lang.org/?start=25a277f03df7e44643ddfcc240d034409cb2f505&end=cdbe2888979bb8797b05f0d58a6f6e60753983d2&absolute=false&stat=instructions%3Au) |
| 7 | + |
| 8 | +2 Regressions, 2 Improvements, 1 Mixed |
| 9 | +0 of them in rollups |
| 10 | + |
| 11 | +#### Regressions |
| 12 | + |
| 13 | +Make building THIR a stealable query [#85273](https://github.com/rust-lang/rust/issues/85273) |
| 14 | +- Moderate regression in [instruction counts](https://perf.rust-lang.org/compare.html?start=126561cb31e8ebe1e2dd9dfd0d3ca621308dc56f&end=d568d63b1f9f5fc47e4202e2a2a84142ff6202d8&stat=instructions:u) (up to 3.7% on `incr-full` builds of `tuple-stress-check`) |
| 15 | +- Work needed for other work to land. A performance run was performed, but still the change was allowed to land. I've asked [why](https://github.com/rust-lang/rust/pull/85273#issuecomment-848017988). |
| 16 | + |
| 17 | +Implement the new desugaring from `try_trait_v2` [#84767](https://github.com/rust-lang/rust/issues/84767) |
| 18 | +- Moderate regression in [instruction counts](https://perf.rust-lang.org/compare.html?start=491cf5561efb1f5ff33c3234ccd0bc5cacbebe3e&end=4e3e6db011c5b482d2bef8ba02274657f93b5e0d&stat=instructions:u) (up to 2.9% on `full` builds of `cranelift-codegen-check`) |
| 19 | +- Moderate improvement in [instruction counts](https://perf.rust-lang.org/compare.html?start=491cf5561efb1f5ff33c3234ccd0bc5cacbebe3e&end=4e3e6db011c5b482d2bef8ba02274657f93b5e0d&stat=instructions:u) (up to -1.8% on `full` builds of `deeply-nested-async-check`) |
| 20 | +- A much larger performance hit than when the performance suite was run while the pull request was open. This is largely a performance regression with only one non-volatile benchmark showing improvement. |
| 21 | +- Biggest change seems to be in the `mir_built` query which seems to make sense given that probably more mir will be produced by this change. |
| 22 | +- Follow-up happening [here](https://github.com/rust-lang/rust/pull/84767#issuecomment-848036325). |
| 23 | + |
| 24 | +#### Improvements |
| 25 | + |
| 26 | +fix deallocation of immutable allocations [#85599](https://github.com/rust-lang/rust/issues/85599) |
| 27 | +- Moderate improvement in [instruction counts](https://perf.rust-lang.org/compare.html?start=6e92fb409816c65cd0a78a1fbcc71e2fbabdf50a&end=0f8cd43ee8c3614e04b5c624dd8a45758d7023da&stat=instructions:u) (up to -1.9% on `full` builds of `deeply-nested-async-check`) |
| 28 | + |
| 29 | +std: Attempt again to inline thread-local-init across crates [#84876](https://github.com/rust-lang/rust/issues/84876) |
| 30 | +- Moderate improvement in [instruction counts](https://perf.rust-lang.org/compare.html?start=3e827cc21e0734edd26170e8d1481f0d66a1426b&end=3bcaeb0bf9e1c29d18abc32928fd2f23d1bed0bd&stat=instructions:u) (up to -1.8% on `incr-full` builds of `externs-debug`) |
| 31 | +- Moderate regression in [instruction counts](https://perf.rust-lang.org/compare.html?start=3e827cc21e0734edd26170e8d1481f0d66a1426b&end=3bcaeb0bf9e1c29d18abc32928fd2f23d1bed0bd&stat=instructions:u) (up to 1.3% on `full` builds of `regex-opt`) |
| 32 | + |
| 33 | +#### Mixed |
| 34 | + |
| 35 | +Avoid zero-length memcpy in formatting [#85391](https://github.com/rust-lang/rust/issues/85391) |
| 36 | +- Moderate regression in [instruction counts](https://perf.rust-lang.org/compare.html?start=df70463ea5d701489d6f53dc780a2c16294d6143&end=a426fc37f2269093ef1a4dbb3e31b3247980fccc&stat=instructions:u) (up to 3.8% on `incr-patched: println` builds of `cargo-opt`) |
| 37 | +- Moderate improvement in [instruction counts](https://perf.rust-lang.org/compare.html?start=df70463ea5d701489d6f53dc780a2c16294d6143&end=a426fc37f2269093ef1a4dbb3e31b3247980fccc&stat=instructions:u) (up to -1.8% on `full` builds of `deeply-nested-async-debug`) |
| 38 | +- This seems to have some interactions with LLVM codegen that cause certain benchmarks to fluctuate in performance. These largely cancel each other out. |
| 39 | + |
| 40 | +#### Nags requiring follow up |
| 41 | + |
| 42 | +- Follow-up: |
| 43 | + [#84767](https://github.com/rust-lang/rust/pull/84767#issuecomment-848036325). |
| 44 | + |
| 45 | + |
0 commit comments