2023-07-18 Triage Log

A lot of spurious results in the regressions this week. However, we did see some real gains with PR #113609, with nearly 40 real-world benchmarks improving their check-build performance by >=1%.

Triage done by @pnkfelix. Revision range: 1d4f5aff..6b9236ed

Summary:

(instructions:u)meanrangecount
Regressions ❌
(primary)		1.5%[0.6%, 3.0%]11
Regressions ❌
(secondary)		1.4%[0.6%, 1.8%]11
Improvements ✅
(primary)		-1.6%[-3.7%, -0.6%]46
Improvements ✅
(secondary)		-1.9%[-4.2%, -0.4%]46
All ❌✅ (primary)-1.0%[-3.7%, 3.0%]57

5 Regressions, 5 Improvements, 5 Mixed; 2 of them in rollups 57 artifact comparisons made in total

Regressions

miri: protect Move() function arguments during the call #113569 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		0.8%[0.5%, 1.1%]13
Regressions ❌
(secondary)		0.9%[0.4%, 1.4%]11
Improvements ✅
(primary)		--0
Improvements ✅
(secondary)		--0
All ❌✅ (primary)0.8%[0.5%, 1.1%]13
  • RalfJ is investigating; has potential fix up in PR #113630, ...
  • ... but its not totally certain that PR is a real fix (i.e. the regression may already have been masked or otherwise resolved independently).
  • But meanwhile, I am hypothesizing that the regression reported here is spurious (discussion)
  • marking as triaged

Ignore flaky clippy tests. #113621 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		1.0%[0.6%, 1.2%]7
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		--0
Improvements ✅
(secondary)		--0
All ❌✅ (primary)1.0%[0.6%, 1.2%]7
  • already marked as triaged (its noise)

Rollup of 6 pull requests #113673 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		1.2%[1.2%, 1.2%]1
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		--0
Improvements ✅
(secondary)		--0
All ❌✅ (primary)1.2%[1.2%, 1.2%]1
  • addressed by PR #113697, already marked as triaged.

Add even more GHA log groups #113514 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		--0
Regressions ❌
(secondary)		2.9%[2.7%, 3.0%]6
Improvements ✅
(primary)		--0
Improvements ✅
(secondary)		--0
All ❌✅ (primary)--0
  • PR author says “this doesn't affect how the compiler was built at all, the perf regression must be spurious.”
  • I agree, marking as triaged

Rollup of 3 pull requests #113738 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		0.7%[0.7%, 0.7%]1
Regressions ❌
(secondary)		3.1%[2.7%, 3.5%]6
Improvements ✅
(primary)		--0
Improvements ✅
(secondary)		--0
All ❌✅ (primary)0.7%[0.7%, 0.7%]1

Improvements

Rewrite UnDerefer, again #113316 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		1.2%[1.2%, 1.2%]1
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		-0.9%[-1.7%, -0.4%]14
Improvements ✅
(secondary)		-1.2%[-2.0%, -0.2%]21
All ❌✅ (primary)-0.8%[-1.7%, 1.2%]15

(re-)tighten sourceinfo span of adjustments in MIR #112945 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		--0
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		-0.7%[-1.1%, -0.4%]18
Improvements ✅
(secondary)		-0.5%[-0.6%, -0.5%]5
All ❌✅ (primary)-0.7%[-1.1%, -0.4%]18

Bump bootstrap to 1.72 beta #113637 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		--0
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		-0.9%[-1.2%, -0.6%]12
Improvements ✅
(secondary)		--0
All ❌✅ (primary)-0.9%[-1.2%, -0.6%]12

Add a cache for maybe_lint_level_root_bounded #113609 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		--0
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		-1.4%[-3.4%, -0.5%]29
Improvements ✅
(secondary)		-1.9%[-5.9%, -0.2%]33
All ❌✅ (primary)-1.4%[-3.4%, -0.5%]29

Remove LLVMRustCoverageHashCString #113430 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		--0
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		-0.7%[-0.7%, -0.7%]2
Improvements ✅
(secondary)		-2.8%[-3.2%, -2.4%]6
All ❌✅ (primary)-0.7%[-0.7%, -0.7%]2

Mixed

Eliminate ZST allocations in Box and Vec #113113 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		1.7%[0.5%, 3.5%]3
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		-1.0%[-1.7%, -0.4%]2
Improvements ✅
(secondary)		--0
All ❌✅ (primary)0.6%[-1.7%, 3.5%]5
  • regressions here were anticipated and unavoidable. This is a bug fix.
  • Marking as triaged.

Enable MIR reference propagation by default #109025 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		1.2%[0.5%, 2.5%]15
Regressions ❌
(secondary)		0.8%[0.2%, 1.3%]7
Improvements ✅
(primary)		-0.9%[-1.0%, -0.8%]3
Improvements ✅
(secondary)		-0.6%[-1.1%, -0.4%]6
All ❌✅ (primary)0.8%[-1.0%, 2.5%]18
  • This is turning on a MIR pass at lower optimizations levels, so its expected that it would cause the compiler to do more work.
  • its clear from the perf runs on the PR itself that the PR author already put in much effort to make the pass faster than it had started out.
  • marking as triaged.

Resurrect: rustc_target: Add alignment to indirectly-passed by-value types, correcting the alignment of byval on x86 in the process. #112157 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		1.7%[1.5%, 2.0%]6
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		--0
Improvements ✅
(secondary)		-2.8%[-2.9%, -2.8%]6
All ❌✅ (primary)1.7%[1.5%, 2.0%]6
  • The six primary regressions were all to variants of diesel (all of check/debug/opt) for variious full and incr-full scenarios.
  • It isn't noise, there seems to be a clear cliff that starts with this PR when looking at the graph starting from 2023-0702.
  • not marking as triaged yet, but was tempted to do so, because this PR is a prerequiste for unlocking various memcpy optimizations added by pcwalton to LLVM

Remove unneeded handling for ExternalLocation::Unknown in rustdoc render context #113697 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		--0
Regressions ❌
(secondary)		3.0%[2.5%, 3.4%]6
Improvements ✅
(primary)		-1.1%[-1.1%, -1.1%]1
Improvements ✅
(secondary)		--0
All ❌✅ (primary)-1.1%[-1.1%, -1.1%]1
  • this was mentioned up above when I was talking about PR #113673
  • the secondary regressions are all to ctfe-stress-5, which lqd says was noisy at this time.
  • marked as triaged.

Add support for allocators in Rc & Arc #89132 (Comparison Link)

(instructions:u)meanrangecount
Regressions ❌
(primary)		0.8%[0.8%, 0.8%]1
Regressions ❌
(secondary)		--0
Improvements ✅
(primary)		-1.3%[-1.3%, -1.3%]1
Improvements ✅
(secondary)		-0.7%[-0.8%, -0.5%]4
All ❌✅ (primary)-0.3%[-1.3%, 0.8%]2

Visiting for perf-triage

  • primary regression was to image-0.24.1 opt full by 0.79%
  • I think this is just noise. From the graph, it seems like image has unpredictably jumped up and down between two plateaus since PR #113113 (a PR discussed up above that changed low level allocation procotol code in Box and Vec, and thus might be expected to have some weird follow-on effects).
  • marked as triaged

Untriaged Pull Requests

Nags requiring follow up

TODO: Nags