[Perf] Windows/x64: 8 Regressions on 1/14/2026 1:50:43 AM +00:00

[performanceautofiler]

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	5d525b730bc668ca693f83b47520258a5fd7254e
Compare	e079636c9baa12b5c082958b9fa1a9b70f40a1f1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Collections.ContainsTrue<String>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Array - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	153.15 μs	423.02 μs	2.76	0.01	False

Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Collections.ContainsTrue<String>*'

Details

System.Collections.ContainsTrue<String>.Array(Size: 512)

ETL Files

Histogram

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	5d525b730bc668ca693f83b47520258a5fd7254e
Compare	e079636c9baa12b5c082958b9fa1a9b70f40a1f1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Collections.TryGetValueTrue<String, String>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
IDictionary - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	4.66 μs	7.93 μs	1.70	0.00	False

Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Collections.TryGetValueTrue<String, String>*'

Details

System.Collections.TryGetValueTrue<String, String>.IDictionary(Size: 512)

ETL Files

Histogram

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	5d525b730bc668ca693f83b47520258a5fd7254e
Compare	e079636c9baa12b5c082958b9fa1a9b70f40a1f1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Collections.TryGetValueTrue<SmallClass, SmallClass>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
ConcurrentDictionary - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	1.68 μs	2.88 μs	1.72	0.01	False
ImmutableDictionary - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	4.40 μs	5.95 μs	1.35	0.21	False

Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Collections.TryGetValueTrue<SmallClass, SmallClass>*'

Details

System.Collections.TryGetValueTrue<SmallClass, SmallClass>.ConcurrentDictionary(Size: 512)

ETL Files

Histogram

JIT Disasms

System.Collections.TryGetValueTrue<SmallClass, SmallClass>.ImmutableDictionary(Size: 512)

ETL Files

Histogram

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	5d525b730bc668ca693f83b47520258a5fd7254e
Compare	e079636c9baa12b5c082958b9fa1a9b70f40a1f1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Collections.ContainsKeyFalse<String, String>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
SortedDictionary - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	165.65 μs	228.05 μs	1.38	0.24	False

Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Collections.ContainsKeyFalse<String, String>*'

Details

System.Collections.ContainsKeyFalse<String, String>.SortedDictionary(Size: 512)

ETL Files

Histogram

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	3211693c5d1875116571962d5d0e3cf401148660
Compare	e079636c9baa12b5c082958b9fa1a9b70f40a1f1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Collections.Sort<IntClass>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Array - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	14.29 μs	21.96 μs	1.54	0.63	False
List - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	13.81 μs	21.05 μs	1.52	0.69	False

Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Collections.Sort<IntClass>*'

Details

System.Collections.Sort<IntClass>.Array(Size: 512)

ETL Files

Histogram

JIT Disasms

System.Collections.Sort<IntClass>.List(Size: 512)

ETL Files

Histogram

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	5d525b730bc668ca693f83b47520258a5fd7254e
Compare	e079636c9baa12b5c082958b9fa1a9b70f40a1f1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Collections.ContainsKeyTrue<String, String>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
IDictionary - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	4.71 μs	7.20 μs	1.53	0.02	False

Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Collections.ContainsKeyTrue<String, String>*'

Details

System.Collections.ContainsKeyTrue<String, String>.IDictionary(Size: 512)

ETL Files

Histogram

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

[DrewScoggins]

🔍 Automated Triage Analysis

Summary: All 8 reported Collection tests show genuine regressions (+35% to +176%), not noise. The most likely cause is commit 87b470c1 "Optimize more bound checks around logical operators" which made extensive changes to JIT bounds check optimization in rangecheck.cpp/h. The pattern of all collection operations regressing simultaneously is consistent with a JIT-level change affecting array/collection indexing.

Finding Confidence: 4/5 (analysis accuracy)
Regression Confidence: 4/5 (likelihood of true regression)

Likely Cause: 122023

📊 Full Analysis (click to expand)

Performance Issue Triage Analysis: #123391

Issue Overview

• Issue: #67621 - [Perf] Windows/x64: 8 Regressions on 1/14/2026 1:50:43 AM +00:00
• Date: 2026-01-14 | Queue: Windows.11.Amd64.Viper.Perf | OS/Arch: Windows/x64
• Commits: 5d525b730bc6 → e079636c9baa (37 commits)
• Labels: untriaged, perf-regression, os-windows, arch-x64, branch-refs/heads/main, runtime-coreclr, kind-micro
• Historical Data: Available (200-336 runs per test)

Test Analysis

Test Name	Baseline	Compare	Δ%	μ	σ	n	z	Thresh%	Noise?	Assessment
System.Collections.ContainsTrue<String>.Array(Size: 512)	153.15 μs	423.02 μs	+176.2%	148.63 μs	2.65	200	-55925	5%	N	Severe regression, step change
System.Collections.TryGetValueTrue<String, String>.IDictionary(Size: 512)	4.66 μs	7.93 μs	+70.1%	4.66 μs	0.01	200	-442	5%	N	Severe regression, step change
System.Collections.TryGetValueTrue<SmallClass, SmallClass>.ConcurrentDictionary(Size: 512)	1.68 μs	2.88 μs	+71.6%	N/A	N/A	N/A	N/A	5%	N	Significant regression
System.Collections.TryGetValueTrue<SmallClass, SmallClass>.ImmutableDictionary(Size: 512)	4.40 μs	5.95 μs	+35.2%	N/A	N/A	N/A	N/A	5%	N	Significant regression
System.Collections.ContainsKeyFalse<String, String>.SortedDictionary(Size: 512)	165.65 μs	228.05 μs	+37.7%	N/A	N/A	N/A	N/A	5%	N	Significant regression
System.Collections.Sort<IntClass>.Array(Size: 512)	14.29 μs	21.96 μs	+53.7%	15.09 μs	1.18	336	5.07	15.6%	N	Significant regression, high variance
System.Collections.Sort<IntClass>.List(Size: 512)	13.81 μs	21.05 μs	+52.5%	15.09 μs	1.18	336	5.07	15.6%	N	Significant regression, high variance
System.Collections.ContainsKeyTrue<String, String>.IDictionary(Size: 512)	4.71 μs	7.20 μs	+52.8%	4.72 μs	0.31	336	8.05	13.1%	N	Clear regression, stable baseline

Legend: μ=mean, σ=stddev, n=runs, z=z-score, Thresh%=adaptive threshold

Confirmed Regressions

System.Collections.ContainsTrue<String>.Array(Size: 512) (+176.2%)

• Delta: 153.15 μs → 423.02 μs (+176.2%)
• Historical Context: μ=148.63 μs (σ=2.65, n=200); CV=1.79%; extremely stable baseline
• Assessment: Severe step change. This is the largest regression in the set.
• Finding Confidence: 4/5 - Clear signal with stable historical baseline
• Regression Confidence: 4/5 - Strong signal, step change pattern, multiple corroborating regressions
• Likely Cause: d2dda323d86e - "Optimize more bound checks around logical operators (Optimize more bound checks around logical operators #122263)"
  • Affected Areas: JIT / RangeCheck / Bounds checking optimization
  • Rationale: This commit makes extensive changes to rangecheck.cpp and rangecheck.h, adding new range operations for AND, OR, ShiftLeft, ShiftRight, and UnsignedMod. The changes to JIT bounds check optimization could have unintended side effects on array access patterns used by Contains operations. Multiple Collection tests regressed simultaneously, consistent with a JIT-level change.
  • Candidates Considered:
    • d2dda323d86e - JIT rangecheck optimization changes (HIGH likelihood)
    • 7ed6f1c3f408 - Bounds checks assertion merging changes (MEDIUM likelihood)

System.Collections.TryGetValueTrue<String, String>.IDictionary(Size: 512) (+70.1%)

• Delta: 4.66 μs → 7.93 μs (+70.1%)
• Historical Context: μ=4.66 μs (σ=0.01, n=200); CV=0.23%; ultra-stable baseline
• Assessment: Clear step change. Ultra-low variance makes this extremely reliable signal.
• Finding Confidence: 5/5 - Near-perfect historical stability
• Regression Confidence: 4/5 - Strong signal with corroboration from other dictionary tests
• Likely Cause: d2dda323d86e

System.Collections.ContainsKeyTrue<String, String>.IDictionary(Size: 512) (+52.8%)

• Delta: 4.71 μs → 7.20 μs (+52.8%)
• Historical Context: μ=4.72 μs (σ=0.31, n=336); z-score=8.05; CV=6.54%
• Finding Confidence: 4/5 | Regression Confidence: 4/5
• Likely Cause: d2dda323d86e

System.Collections.Sort<IntClass>.Array(Size: 512) (+53.7%) and List (+52.5%)

• Delta: 14.29 μs → 21.96 μs (Array), 13.81 μs → 21.05 μs (List)
• Historical Context: μ=15.09 μs (σ=1.18, n=336); z-score=5.07; CV=7.79%
• Finding Confidence: 3/5 | Regression Confidence: 3/5 - Higher variance
• Likely Cause: d2dda323d86e

Other Regressions (ConcurrentDictionary +71.6%, ImmutableDictionary +35.2%, SortedDictionary +37.7%)

• All show consistent pattern with dictionary/collection regressions
• Finding Confidence: 3/5 | Regression Confidence: 3/5
• Likely Cause: d2dda323d86e

Open Questions

• Historical Data: Partial analysis - 4 of 8 tests fully analyzed with historical statistics
• Confidence Limitations: Large commit range (37 commits) but clear attribution to JIT rangecheck changes
• Verification Needed: Bisection recommended to confirm attribution
• Bisection Status: Not performed

Recommended Actions

1. Verify attribution: Run bisection focusing on JIT rangecheck commits (d2dda323d86e and 7ed6f1c3f408)
2. Engage JIT team: Notify EgorBo (author of both rangecheck commits) about potential regression in bounds check optimization
3. Code review: Examine the rangecheck.cpp changes for potential edge cases in AND/OR/Shift range operations
4. Profiling: Collect hardware counters on affected tests to identify specific JIT code differences

---

Generated by PerfTriageAgent on 2026-01-20 17:05 UTC

[DrewScoggins]

Think it is related to #122023

[DrewScoggins]

We will do a bisection on this to confirm if it is related to this change.

[DrewScoggins]

Related:

• [Perf] Linux/x64: 4 Regressions on 1/13/2026 6:20:38 AM +00:00 perf-autofiling-issues#67641

[jkotas]

cc @hez2010

[dotnet-policy-service]

Tagging subscribers to this area: @JulieLeeMSFT, @dotnet/jit-contrib
See info in area-owners.md if you want to be subscribed.