ci: remove samply walltime profiler and pinned runner version

[not-matthias]

Summary

• Reverts the WIP commit that pinned the runner to 4.17.7-alpha.1 to test samply
• Removes the CODSPEED_WALLTIME_PROFILER: samply env var from the benchmarks workflow step

The CI workflow returns to using CodSpeedHQ/action@main with the default runner version and default walltime profiler.

Test plan

• CI runs the benchmarks step successfully

[greptile-apps]

Greptile Summary

This PR cleans up a WIP experiment: it removes the CODSPEED_WALLTIME_PROFILER: samply env var and the pinned runner-version: 4.17.7-alpha.1 from the benchmarks workflow step, returning to the default runner and walltime profiler via CodSpeedHQ/action@main. It also extends the cache to include the .capstone directory (with a bumped v2 key to bust stale caches) and reduces the job timeout from 30 to 15 minutes.

• Two stress-ng CPU-stress commands are removed from bench/generate_config.py, consistent with the samply experiment being rolled back.
• The .capstone cache addition and v2 key bump ensure cache hits no longer break the make install step that needs the Capstone headers.

Confidence Score: 5/5

Safe to merge — this reverts a WIP experiment and restores the CI workflow to its previous stable state.

All changes are revertive: the samply env var and pinned alpha runner version are removed, returning the workflow to its default configuration. The only additive changes (.capstone cache path, v2 key bump, timeout reduction) are well-reasoned and consistent with the diff's intent. No logic changes were made to benchmark generation beyond removing the two stress-ng commands that were only relevant to the samply experiment.

No files require special attention.

Important Files Changed

Filename	Overview
.github/workflows/codspeed.yml	Reverts samply profiler env var and pinned runner version; adds .capstone to the cache path with a bumped v2 key; reduces timeout from 30 to 15 minutes; all changes are intentional and consistent with the PR description.
bench/generate_config.py	Removes two stress-ng benchmark commands that were added for samply profiler testing; the remaining command list is clean, though stress-ng is still installed in the workflow step.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant GH as GitHub Actions
    participant Cache as actions/cache
    participant Build as Build Steps
    participant CodSpeed as CodSpeedHQ/action@main

    GH->>Cache: "Restore valgrind-{ver}-{os}-v2-{hash}"
    note over Cache: Paths: /tmp/valgrind-build + .capstone

    alt Cache miss
        Cache-->>Build: No hit
        Build->>Build: apt-get install deps
        Build->>Build: Build Capstone (local only)
        Build->>Build: "just build {valgrind}"
        Build->>Cache: Save cache (both paths)
    else Cache hit
        Cache-->>Build: Restored /tmp/valgrind-build + .capstone
    end

    Build->>Build: "just install {valgrind}"
    Build->>Build: "generate_config.py -> codspeed.yml"
    note over Build: 3 commands x 5 configs (minus requires_codspeed skips)

    Build->>CodSpeed: Run benchmarks (mode: walltime, default runner)
    note over CodSpeed: No CODSPEED_WALLTIME_PROFILER env var
    CodSpeed-->>GH: Results uploaded

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant GH as GitHub Actions
    participant Cache as actions/cache
    participant Build as Build Steps
    participant CodSpeed as CodSpeedHQ/action@main

    GH->>Cache: "Restore valgrind-{ver}-{os}-v2-{hash}"
    note over Cache: Paths: /tmp/valgrind-build + .capstone

    alt Cache miss
        Cache-->>Build: No hit
        Build->>Build: apt-get install deps
        Build->>Build: Build Capstone (local only)
        Build->>Build: "just build {valgrind}"
        Build->>Cache: Save cache (both paths)
    else Cache hit
        Cache-->>Build: Restored /tmp/valgrind-build + .capstone
    end

    Build->>Build: "just install {valgrind}"
    Build->>Build: "generate_config.py -> codspeed.yml"
    note over Build: 3 commands x 5 configs (minus requires_codspeed skips)

    Build->>CodSpeed: Run benchmarks (mode: walltime, default runner)
    note over CodSpeed: No CODSPEED_WALLTIME_PROFILER env var
    CodSpeed-->>GH: Results uploaded

Loading

_{Reviews (7): Last reviewed commit: "fix(bench): drop stress-ng benchmarks" | Re-trigger Greptile}

[codspeed-hq]

Merging this PR will degrade performance by 48.6%

⚡ 1 improved benchmark
❌ 1 regressed benchmark
✅ 40 untouched benchmarks
⏩ 88 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_valgrind[valgrind-3.25.1, echo Hello, World!, full-no-inline]	567.6 ms	3,101.9 ms	-81.7%
⚡	test_valgrind[valgrind-3.26.0, python3 testdata/test.py, full-no-inline]	10.3 s	7.1 s	+44.38%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing fix/remove-wip-commit (e0e62db) with master (2bc9a1c)}

Footnotes

1. 88 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩