2024-01-30.md

2024-01-30 Triage Log

This was a very quiet week with only one PR having any real impact on overall compiler performance. The removal of the internal StructuralEq trait saw a roughly 0.4% improvement on average across nearly 50 real-world benchmarks.

Triage done by @rylev. Revision range: d6b151fc7..5c9c3c7

Summary:

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.5%	[0.3%, 0.7%]	5
Regressions ❌ (secondary)	0.5%	[0.2%, 1.4%]	10
Improvements ✅ (primary)	-0.5%	[-1.5%, -0.2%]	48
Improvements ✅ (secondary)	-2.3%	[-7.7%, -0.4%]	36
All ❌✅ (primary)	-0.4%	[-1.5%, 0.7%]	53

0 Regressions, 4 Improvements, 4 Mixed; 3 of them in rollups 37 artifact comparisons made in total

Regressions

Improvements

Only use dense bitsets in dataflow analyses #116152 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.4%	[-0.5%, -0.2%]	5
Improvements ✅ (secondary)	-0.8%	[-1.1%, -0.4%]	2
All ❌✅ (primary)	-0.4%	[-0.5%, -0.2%]	5

const-eval interning: get rid of type-driven traversal #119044 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.9%	[-1.1%, -0.8%]	3
Improvements ✅ (secondary)	-3.8%	[-8.0%, -0.6%]	14
All ❌✅ (primary)	-0.9%	[-1.1%, -0.8%]	3

Remove all ConstPropNonsense #119627 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.4%	[-0.6%, -0.2%]	11
Improvements ✅ (secondary)	-0.5%	[-0.5%, -0.5%]	3
All ❌✅ (primary)	-0.4%	[-0.6%, -0.2%]	11

remove StructuralEq trait #116167 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.3%	[0.3%, 0.3%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.4%	[-0.6%, -0.2%]	48
Improvements ✅ (secondary)	-1.3%	[-3.0%, -0.3%]	23
All ❌✅ (primary)	-0.4%	[-0.6%, 0.3%]	49

Mixed

Rollup of 10 pull requests #120335 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.4%	[0.3%, 0.8%]	9
Regressions ❌ (secondary)	0.7%	[0.6%, 0.9%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.7%	[-0.7%, -0.7%]	1
All ❌✅ (primary)	0.4%	[0.3%, 0.8%]	9

• Small regressions in doc benchmarks
• Most likely caused by addings AsyncFn family of traits.
• Author has been pinged, but it's probably acceptable cost given that we're documenting more.

Rollup of 9 pull requests #120375 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.3%	[0.3%, 0.4%]	3
Improvements ✅ (primary)	-0.2%	[-0.2%, -0.2%]	6
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.2%	[-0.2%, -0.2%]	6

• Small number of improvements outweigh the small number of regressions and so it's not really worth investigating further.

Rollup of 12 pull requests #120401 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.6%	[0.5%, 1.1%]	7
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.1%	[-3.1%, -3.1%]	1
All ❌✅ (primary)	-	-	0

• Very small regressions in match stress test is not worth investigating

Merge into larger interval set #120024 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.3%	[0.2%, 0.4%]	14
Regressions ❌ (secondary)	0.2%	[0.2%, 0.2%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.2%]	2
All ❌✅ (primary)	0.3%	[0.2%, 0.4%]	14

• The reviewer decided that the performance hit here was worth it given that we are now protecting against truly extreme cases that the perf benchmark suite doesn't test against.