[Perf] Windows/x86: 5 Regressions on 6/19/2026 8:48:43 AM +00:00

[performanceautofiler]

Run Information

Name	Value
Architecture	x86
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	a50a8baa52210e32fb688a6cb7f59e40b79ff830
Compare	ae899ea0758628a22feb37a1401ae397cc5fc8d1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:sdk_scenarios

Regressions in SDK MSTest Test Project (.NET Core) Template Clean Build

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
SDK MSTest Test Project - Process Time ADX - Test Multi Config Graph	2758.10 ms	3103.68 ms	1.13	0.00	True

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 'SDK MSTest Test Project (.NET Core) Template Clean Build*'

Details

SDK MSTest Test Project (.NET Core) Template Clean Build

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x86
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	a50a8baa52210e32fb688a6cb7f59e40b79ff830
Compare	ae899ea0758628a22feb37a1401ae397cc5fc8d1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:sdk_scenarios

Regressions in Inner Loop Blazor Client

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Inner Loop Blazor Client - Generic Startup Diff ADX - Test Multi Config Graph	1817.35 ms	2640.53 ms	1.45	0.00	True

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 'Inner Loop Blazor Client*'

Details

Inner Loop Blazor Client

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x86
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	a50a8baa52210e32fb688a6cb7f59e40b79ff830
Compare	ae899ea0758628a22feb37a1401ae397cc5fc8d1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:sdk_scenarios

Regressions in Inner Loop MVC

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Inner Loop MVC - Generic Startup Diff ADX - Test Multi Config Graph	425.26 ms	608.32 ms	1.43	0.00	True

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 'Inner Loop MVC*'

Details

Inner Loop MVC

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x86
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	a50a8baa52210e32fb688a6cb7f59e40b79ff830
Compare	ae899ea0758628a22feb37a1401ae397cc5fc8d1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:sdk_scenarios

Improvements in Inner Loop Console

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Inner Loop Console - Generic Startup Diff ADX - Test Multi Config Graph	218.76 ms	183.50 ms	0.84	0.00	True

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 'Inner Loop Console*'

Details

Inner Loop Console

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x86
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	a50a8baa52210e32fb688a6cb7f59e40b79ff830
Compare	ae899ea0758628a22feb37a1401ae397cc5fc8d1
Diff	Diff
Configs	CompilationMode:tiered, RunKind:sdk_scenarios

Improvements in SDK MSTest Test Project (.NET Core) Template Build(No Change)

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
SDK MSTest Test Project - Process Time ADX - Test Multi Config Graph	959.14 ms	933.58 ms	0.97	0.00	True

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 'SDK MSTest Test Project (.NET Core) Template Build(No Change)*'

Details

SDK MSTest Test Project (.NET Core) Template Build(No Change)

ETL Files

Histogram

JIT Disasms

Docs

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

[LoopedBard3]

🔍 Automated Triage Analysis

Summary: All 3 reported regressions appear to be measurement noise or borderline signals in high-variance SDK scenario tests. Inner Loop Blazor Client (+45.3%) and Inner Loop MVC (+43.1%) exceed adaptive thresholds but have CV>18% with z-scores below 2.0, indicating extremely noisy measurements rather than true regressions. Finding confidence 3/5 (good historical data but commits inaccessible), regression confidence 1/5 (all signals within historical noise bounds).

Finding Confidence: 3/5 (analysis accuracy)
Regression Confidence: 1/5 (likelihood of true regression)
⚠️ Assessment: Likely noise - recommend closing or monitoring for recurrence.

🤖 Proposed Next Actions: Uncertain — flag for re-triage — 1 done, 1 dry-run (rule R5_look_again — see Full Analysis for the table). · 🦆 Re-analysis: ⚠️ disputes (inconsistent-noise-call, no-bisection-evidence, weak-evidence)

📊 Full Analysis (click to expand)

Summary: All 3 reported regressions appear to be measurement noise or borderline signals in high-variance SDK scenario tests. Inner Loop Blazor Client (+45.3%) and Inner Loop MVC (+43.1%) exceed adaptive thresholds but have CV>18% with z-scores below 2.0, indicating extremely noisy measurements rather than true regressions. Finding confidence 3/5 (good historical data but commits inaccessible), regression confidence 1/5 (all signals within historical noise bounds).

Issue Overview

• Issue: [Perf] Windows/x86: 5 Regressions on 6/19/2026 8:48:43 AM +00:00 #74787 - [Perf] Windows/x86: 5 Regressions on 6/19/2026 8:48:43 AM +00:00
• Date: 2026-06-19 | Queue: Windows.11.Amd64.Viper.Perf | OS/Arch: Windows 10.0.22631/x86
• Commits: a50a8baa → ae899ea0 (unknown commit count - range inaccessible)
• Labels: untriaged, perf-regression, arch-x86, os-windows, kind-sdk_scenarios, branch-11.0, compilationmode-tiered, runkind-sdk_scenarios
• Historical Data: Available (263-297 runs per test)
• Runtime stack: Desktop + CoreCLR (implicit)
• Build Repo: dotnet/sdk (branch 11.0)

Test Analysis

Test Name	Baseline	Compare	Δ%	μ	σ	n	z	Thresh%	Noise?	Assessment
SDK MSTest Template Clean Build	2758.10 ms	3103.68 ms	+12.53%	3065.65	197.89	297	0.19	12.91%	Y	Drift; compare returning toward historical mean
Inner Loop Blazor Client	1817.35 ms	2640.53 ms	+45.30%	2028.83	391.91	263	1.56	38.63%	Borderline	CV=19.3%; extreme variance; not unprecedented
Inner Loop MVC	425.26 ms	608.32 ms	+43.05%	496.29	91.87	297	1.22	37.02%	Borderline	CV=18.5%; extreme historical outliers (244-1311ms)
Inner Loop Console	218.76 ms	183.50 ms	-16.12%	204.09	45.33	297	-0.45	44.42%	Y	CV=22.2%; well within noise
SDK MSTest Template Build(No Change)	959.14 ms	933.58 ms	-2.66%	1000.37	45.67	295	-1.46	9.13%	Y	Long-term downward drift pattern

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

Finding Confidence: 3/5 - Good historical data (263-297 runs per test) with high statistical confidence in thresholds, but commit details inaccessible via GitHub API preventing attribution analysis.

Candidate Stack Relevance

Unable to assess candidate stack relevance — both baseline and compare commits (a50a8baa, ae899ea0) are not accessible via the GitHub API for dotnet/sdk. The commit range comparison also returned "not found." This prevents examination of changed files and commit attribution.

Confirmed Regressions

No confirmed regressions. All tests classified as Noise (3 tests) or Borderline (2 tests) by statistical analysis:

Inner Loop Blazor Client (+45.3%) — Borderline

• Delta: 1817.35 ms → 2640.53 ms (+45.30%)
• Historical Context: μ=2028.83 (σ=391.91, n=263); z-score=1.56; adaptive threshold=38.63%
• Assessment: Borderline — delta exceeds adaptive threshold but z-score is below 2.0. CV=19.3% indicates fundamentally noisy measurement. Baseline was below historical mean; compare is elevated but not unprecedented in this test's history.
• Finding Confidence: 3/5 - Adequate historical data but cannot verify commits
• Regression Confidence: 1/5 - z=1.56 in borderline range, CV>18%, no clear step change pattern
• Likely Cause: No attribution possible (commits inaccessible)
• Verification Status: Not verified

Inner Loop MVC (+43.1%) — Borderline

• Delta: 425.26 ms → 608.32 ms (+43.05%)
• Historical Context: μ=496.29 (σ=91.87, n=297); z-score=1.22; adaptive threshold=37.02%
• Assessment: Borderline — delta exceeds adaptive threshold but z-score is well below 1.5. CV=18.5% with extreme historical outliers (range 244-1311ms) signals very high measurement variance. Baseline was below historical mean — conflicting signals.
• Finding Confidence: 3/5 - Adequate historical data but cannot verify commits
• Regression Confidence: 1/5 - z=1.22 below significance threshold, extreme variance
• Likely Cause: No attribution possible (commits inaccessible)
• Verification Status: Not verified

SDK MSTest Template Clean Build (+12.5%) — Noise

• Delta: 2758.10 ms → 3103.68 ms (+12.53%)
• Historical Context: μ=3065.65 (σ=197.89, n=297); z-score=0.19; adaptive threshold=12.91%
• Assessment: Noise — delta is just inside adaptive threshold, z-score is negligible. Baseline was at the low end of a long-term downward drift (early mean ~3365 → recent mean ~2882). Compare is returning toward the historical mean rather than showing a new regression.
• Finding Confidence: 3/5 - Strong historical data confirms drift pattern
• Regression Confidence: 1/5 - z=0.19 is negligible; natural variance recovery

Open Questions

• Commit Data: Both baseline (a50a8baa) and compare (ae899ea0) commits are inaccessible via GitHub API — may be internal SDK build commits or from a private fork. Cannot determine commit range size or inspect changed files.
• Historical Data: All tests have excellent historical coverage (263-297 runs), but the two borderline tests (Blazor Client, MVC) have CV>18% making regression detection unreliable.
• Confidence Limitations: Cannot perform attribution analysis without commit access. Regression confidence capped at 1/5 due to all signals being within or near historical noise bounds.
• Measurement Quality: Inner Loop Console has historical values below zero suggesting measurement artifacts. Blazor Client and MVC have extreme outliers suggesting test harness instability.
• Bisection Status: Not performed — no confirmed regression to bisect and commits are inaccessible.

Recommended Actions

1. Monitor subsequent runs - Re-evaluate if Blazor Client and MVC values remain elevated in the next 2-3 runs; a sustained elevation would upgrade from Borderline to Regression
2. Investigate test stability - Blazor Client (CV=19.3%) and MVC (CV=18.5%) have fundamentally unstable measurements; consider investigating test harness or measurement methodology improvements
3. Verify commit accessibility - Determine why commits a50a8baa and ae899ea0 are not accessible in dotnet/sdk; may need to check if these are internal build SHAs from a different repository or branch

🤖 Proposed Next Actions

Rule: R5_look_again — Uncertain — flag for re-triage.

Finding confidence 3/5 is below the required 4/5. Revisit once more historical data is available.

#	Action	Status	Details
1	Add label	✅ Done	Look Again
2	Remove label	⏭️ Skipped (dry-run)	untriaged

🦆 Re-analysis (advisory)

⚠️ Disputes the analysis — advisory only, not applied

Flags: inconsistent-noise-call, no-bisection-evidence, weak-evidence

Rationale: Two findings are marked noise=False yet carry 1/5 change confidence, which is contradictory — if we are only 1/5 confident the change is real, the appropriate noise classification is True (or at best indeterminate). With no bisection performed, no culprit identified, and uniformly minimal change confidence, finding confidence should be lowered to 2/5 because there is insufficient evidence to support the noise=False designation with any meaningful certainty.

Suggested confidence: Finding Confidence → 2/5 (advisory — not applied)

This critique is advisory in the current configuration and did not change the confidences, culprit, or proposed actions above.

---

Generated by PerfTriageAgent on 2026-06-19 08:54 UTC