2023-11-21 Triage Log

Pretty quiet week, with only a small number of statistically significant changes landing.

Triage done by @simulacrum. Revision range: 173b6e686b158dbad7d072c64bef3ced2052312b..4f3da903a43f22ea33d2ca4435a24b42fc1f842a

Summary:

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.3%	[0.2%, 0.9%]	33
Regressions ❌ (secondary)	0.5%	[0.2%, 1.1%]	17
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.4%	[-0.6%, -0.3%]	7
All ❌✅ (primary)	0.3%	[0.2%, 0.9%]	33

1 Regressions, 1 Improvements, 1 Mixed; 0 of them in rollups 60 artifact comparisons made in total

Regressions

some additional region refactorings #117944 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.1%	[0.1%, 0.1%]	1
Regressions ❌ (secondary)	0.4%	[0.3%, 1.0%]	10
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.1%	[0.1%, 0.1%]	1

Slight regression to diesel incr-full in the primary benchmarks. That appears to be real, but the limitation to incr-full makes me suspect it's not worth further investigation, since that scenario is relatively rare.

Improvements

Update to LLVM 17.0.5 #117907 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.9%	[-2.9%, -2.9%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-2.9%	[-2.9%, -2.9%]	2

Mixed

Reenable effects in libcore #117825 (Comparison Link)

(instructions:u)	mean	range	count
Regressions ❌ (primary)	0.3%	[0.2%, 0.8%]	6
Regressions ❌ (secondary)	0.6%	[0.3%, 0.8%]	3
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.3%	[-1.3%, -1.3%]	1
All ❌✅ (primary)	0.3%	[0.2%, 0.8%]	6

Regressions seem real, likely due to downstream effects from libcore's metadata getting larger? But not entirely clear; the PR does not appear to directly change any functions/types in core.