2023-01-24 Triage Log

Largely a win for compiler performance with 100 test cases in real-world crates showing some sort of change in performance with an average 1% improvement. These wins were a combination of many different changes including how doc(hidden) gets more efficiently encoded in metadata, some optimizations in the borrow checker, and simplification of the output from derive(Debug) for fieldless enums.

Triage done by @rylev. Revision range: 1f72129f..c8e6a9e8

Summary:

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.4%	[0.2%, 0.7%]	19
Regressions ❌ (secondary)	0.9%	[0.2%, 1.5%]	34
Improvements ✅ (primary)	-1.3%	[-17.2%, -0.2%]	81
Improvements ✅ (secondary)	-2.1%	[-7.1%, -0.2%]	64
All ❌✅ (primary)	-1.0%	[-17.2%, 0.7%]	100

2 Regressions, 5 Improvements, 3 Mixed; 1 of them in rollups 34 artifact comparisons made in total

Regressions

Check ADT fields for copy implementations considering regions #105102 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.5%	[0.4%, 0.5%]	6
Regressions ❌ (secondary)	1.7%	[1.4%, 2.0%]	3
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.5%	[0.4%, 0.5%]	6

• Soundness fix, combined with a small regression means we likely don't need to worry too much about this one.
• Asked the author and reviewer to comment.

ci: upgrade mingw-check to ubuntu:22.04 #106948 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.3%	[0.3%, 0.3%]	2
Regressions ❌ (secondary)	0.7%	[0.6%, 0.8%]	6
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.3%	[0.3%, 0.3%]	2

• This is noise correcting from spurious improvements in the previous run.

Improvements

rustdoc: Revert #104889 #107083 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-4.4%	[-15.7%, -0.5%]	7
Improvements ✅ (secondary)	-1.5%	[-1.6%, -1.4%]	3
All ❌✅ (primary)	-4.4%	[-15.7%, -0.5%]	7

Remove some ref patterns from the compiler #106090 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.2%	[-0.2%, -0.1%]	3
Improvements ✅ (secondary)	-0.3%	[-0.5%, -0.3%]	5
All ❌✅ (primary)	-0.2%	[-0.2%, -0.1%]	3

Lazy dominator tree construction in borrowck #106976 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.4%	[-0.6%, -0.3%]	26
Improvements ✅ (secondary)	-0.7%	[-1.0%, -0.3%]	14
All ❌✅ (primary)	-0.4%	[-0.6%, -0.3%]	26

Remap paths in UI tests by default #105924 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.3%	[-0.3%, -0.3%]	2
Improvements ✅ (secondary)	-0.6%	[-0.8%, -0.3%]	7
All ❌✅ (primary)	-0.3%	[-0.3%, -0.3%]	2

rustc_metadata: Encode doc(hidden) flag to metadata #107136 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.2%	[0.2%, 0.2%]	1
Improvements ✅ (primary)	-1.9%	[-5.7%, -0.5%]	18
Improvements ✅ (secondary)	-3.3%	[-5.8%, -0.8%]	28
All ❌✅ (primary)	-1.9%	[-5.7%, -0.5%]	18

Mixed

Rollup of 9 pull requests #107143 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.3%	[0.2%, 0.8%]	57
Regressions ❌ (secondary)	0.5%	[0.2%, 1.4%]	28
Improvements ✅ (primary)	-0.7%	[-1.1%, -0.3%]	2
Improvements ✅ (secondary)	-0.7%	[-1.2%, -0.5%]	8
All ❌✅ (primary)	0.3%	[-1.1%, 0.8%]	59

• This rollup contains a perf regression due to #104672
• Probably caused by some changes to inlining.

Simplify derive(Debug) output for fieldless enums #106884 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.8%	[0.7%, 1.0%]	7
Improvements ✅ (primary)	-0.5%	[-1.7%, -0.2%]	22
Improvements ✅ (secondary)	-1.7%	[-2.9%, -0.5%]	8
All ❌✅ (primary)	-0.5%	[-1.7%, -0.2%]	22

• Improvements outweigh the regressions so I think it's fine if we just take this performance as is.

Use UnordMap and UnordSet for id collections (DefIdMap, LocalDefIdMap, etc) #106977 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.3%	[0.2%, 0.3%]	2
Regressions ❌ (secondary)	0.6%	[0.4%, 0.8%]	7
Improvements ✅ (primary)	-0.5%	[-0.5%, -0.5%]	3
Improvements ✅ (secondary)	-0.3%	[-0.7%, -0.2%]	20
All ❌✅ (primary)	-0.2%	[-0.5%, 0.3%]	5

• perf got slightly worse from the original perf run, but it‘s still enough of a wash that I don’t think it needs investigating.