[efficiency-improver] Add Scenario2: data-driven test performance benchmark

[github-actions]

Background

Recent PRs #9514 and #9617 eliminated significant allocations in data-driven test execution:

• #9514: cached MethodInfo.GetParameters() in TestMethodInfo.ParameterTypes
• #9617: eliminated O(N) allocations across data row iteration — CloneForDataDrivenIteration dictionary, TCS bridge fast path, ReflectionTestMethodInfo wrapper caching

The existing performance runner (test/Performance/MSTest.Performance.Runner/) only has Scenario1: 100 classes × 100 plain test methods (10,000 total). There is no scenario that exercises data-driven tests, so there is no regression baseline for the improvements in #9514 and #9617.

Proposed: Scenario2

Add a new scenario that measures data-driven test execution performance:

Suggested shape:

• 10 classes × 10 test methods × 100 data rows each = 10,000 test invocations
• Each method attributed with [DataRow] (100 rows) or a [DynamicData] source
• Mirrors total invocation count of Scenario1 so results are directly comparable

What this enables:

• Baseline measurement for the allocation improvements already merged
• Regression protection for future changes to data-driven hot paths (TestMethodRunner, ReflectionTestMethodInfo, TestMethodInfo.ResolveArguments)
• Comparison of wall-clock time and CPU time between plain and data-driven execution models (helpful for diagnosing overhead sources)

Why now

Without this scenario, any regression in data-driven allocation paths would go undetected. The infrastructure already supports this — Scenario1 demonstrates the pattern, and the runner is designed to be extended.

Acceptance criteria

1. Scenario2.cs added under test/Performance/MSTest.Performance.Runner/Scenarios/
2. Scenario runs end-to-end with PlainProcess and/or DotnetTestProcess
3. Results are comparable in format to Scenario1 output (JSON + compressed ZIP)
4. README or comment explains the scenario shape and its relationship to [efficiency-improver] perf: cache MethodInfo.GetParameters() to avoid per-row array allocations in data-driven tests #9514/perf: reduce allocations in data-driven test execution hot path #9617

Related

• Performance runner: test/Performance/MSTest.Performance.Runner/
• Existing scenario: Scenarios/Scenario1.cs
• Prior work: [efficiency-improver] perf: cache MethodInfo.GetParameters() to avoid per-row array allocations in data-driven tests #9514 (GetParameters caching), perf: reduce allocations in data-driven test execution hot path #9617 (data-driven allocation reductions)

🤖 Automated content by GitHub Copilot. Generated by the Efficiency Improver workflow. · 351.3 AIC · ⌖ 19.3 AIC · ⊞ 13.6K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/efficiency-improver.md@main

[github-actions]

This is now resolved. PR #9728 ("perf: add data-driven test Scenario2 to the performance runner"), merged 2026-07-08 by @Evangelink, added Scenario2 to the performance runner covering exactly this scenario: 100 classes × 10 methods × 10 DataRow rows = 10,000 invocations, alongside PlainProcess and DotnetTestProcess pipelines and a corresponding acceptance test. This issue can be closed.

🤖 Automated content by GitHub Copilot. Generated by the Perf Improver workflow. · 383 AIC · ⌖ 5.96 AIC · ⊞ 12.5K · [◷]( · ◷)

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/perf-improver.md@main

[Evangelink]

Resolved by #9728 (merged 2026-07-08). Scenario2 (data-driven: classes x methods x [DataRow] rows = 10,000 invocations) is in the performance runner, registered in Program.cs for both PlainProcess and DotnetTestProcess, with an explanatory comment referencing #9514/#9617. Closing.