[RattataKing]

• Bump the benchmark corpus pin
• Dashboad v1 is still in the progress, updated benchmark workflow will collect data first and present partial data to match the existing dashboard's format
• ISSUE ID: [rocjitsu] Collect longer benchmark candidates and local runtimes for V1 #12611

[therock-pr-bot]

✅ All Checks Passed — Ready for Review

Check	Status	Details
📝 PR Description	✅ Pass	—
⛔ Forbidden Files	✅ Pass	—
🧪 Unit Test	✅ Pass	—
🔎 pre-commit	✅ Pass	—
🚫 Draft PR	🔜 To Be Enabled	—
🚩 Feature Flag	🔜 To Be Enabled	—
📊 Code Coverage	🔜 To Be Enabled	—
🤖 therock-pr-bot	✅ Pass	—

🎉 All checks passed! This PR is ready for review.

📖 Need help? See the Policy FAQ for details on every check and how to fix failures.

🙋 Wish to Override Policy?

[therock-pr-bot]

Pre-commit check failed

⛔ pre-commit failed

Please run locally:

• python -m pip install pre-commit
• pre-commit install
• pre-commit run --all-files --show-diff-on-failure

This repo uses .pre-commit-config.yaml.

[therock-pr-bot]

🎉 All checks passed! This PR is ready for review.

[kuhar]

Can we remove the unused scaffolding so these interfaces show what is actually exercised?

• In emulation/rocjitsu/website/dashboard/scripts/dashboard-data/tests/test_benchmark_result.py:5,63,85, importlib.util and manifest() are unused, and context() is consumed only for source_subject. Simplifying that fixture also removes unused revision/context fields that look as though they are being checked.
• At .github/workflows/rocjitsu-benchmarks.yml:44, the machine and source_commit_message job outputs have no downstream consumers; their step outputs already feed the converter locally.
• In the test file at line 235, the removed-script-name blacklist mostly records intermediate implementation choices. Assertions about the intended publication artifacts and validation would protect the contract more directly.

[kuhar]

Left 3 inline comments

[kuhar]

I think it would help to make these helper names and test fixtures more explicit:

• In emulation/rocjitsu/website/dashboard/scripts/dashboard-data/prepare-benchmark-result.py:29,33, use names such as parse_timestamp and build_published_run: the former parses a datetime, and the latter builds a whole run, whereas the dashboard contract uses “result” for a per-test record.
• In emulation/rocjitsu/website/dashboard/scripts/dashboard-data/tests/test_benchmark_result.py:15, name ROOT as REPO_ROOT to identify which root the seven-parent traversal targets.
• In the same test file at line 104, explain that the unusual indentation and trailing whitespace deliberately detect reserialization. The publication tests require the archived raw JSON to preserve those bytes.

[kuhar]

I think it would help to make these remaining names more explicit:

• In emulation/rocjitsu/website/dashboard/scripts/dashboard-data/tests/test_benchmark_result.py:411, rename test_workflow_shell_and_removed_paths to something like test_workflow_shell_and_publication_policy; its body now checks shell syntax, publication policy, dispatch inputs, and checkouts, with no removed-path assertions.
• In that test file at lines 18–21, use SOURCE_SHA (or ROCJITSU_SHA) and CORPUS_SHA for SHA and CORPUS. With separate workflow and Tensile revisions in the fixture, identifying each revision at its declaration makes the provenance assertions easier to follow.
• In emulation/rocjitsu/website/dashboard/scripts/dashboard-data/prepare-benchmark-result.py:23, a name such as encode_json_bytes would make the format and return type of encoded clear at its hashing and file-writing call sites.

[kuhar]

Left an inline reply and 3 new comments

[kuhar]

A few comments, plus an inline reply

[kuhar]

Left some inline comments

[kuhar]

Left an inline reply and a new comment

[kuhar]

LGTM

[systems-assistant]

TheRock Submodule Bump Activity

Newest first

• 2026-10-22 — Included in TheRock via ROCm/TheRock#9425.