test(rccl): Add doneEvent stream-ordering unit tests

[rahulvaidya20]

Motivation

Adds CI unit tests for the post-checkpoint RCCL hang caused by a dropped doneEvent ordering edge in the single-stream fast launch path.

Technical Details

Six process-isolated tests in DoneEventOrderingTests.cpp, registered in rccl-UnitTests and wired into ci-precheckin.json:

• FastPathStreamAlternation - ungrouped alternating-stream AllReduces; exercises the lastStream != launchStream doneEvent branch.
• PostReinitStreamAlternation - two ncclCommDestroy+ncclCommInitAll reinit cycles each followed by alternating-stream AllReduces; exercises the lastStreamValid reset path.
• SameStreamBaseline - same stream every iteration; negative control.
• GroupedPathBaseline - alternating-stream AllReduces inside ncclGroupStart/End; verifies group calls don't break doneEvent ordering.
• DefaultStreamToNamedStream - first AllReduce on hipStreamDefault (0), second on a named stream; exercises the lastStreamValid nullptr-handling path.
• SameStreamThenSwitch - kIterations same-stream AllReduces then a single stream switch; exercises the lazy hipEventRecord on lastStream in ncclLaunchPrepare.

JIRA ID

JIRA ID: AICOMRCCL-1301

Test Plan

Run rccl-UnitTests --gtest_filter='DoneEventOrdering.*' on a multi-GPU node.

Test Result

All 6 tests pass on banff (8-GPU MI300X, ROCm 7.2, RCCL 2.30.4), ~52 s total.

Submission Checklist

• Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

[therock-pr-bot]

❌ PR Check — Action Required

Check	Status	Details
🌿 Branch Name	✅ Pass	—
📝 PR Title/Description	✅ Pass	—
⛔ Forbidden Files	✅ Pass	—
🧪 Unit Test	❌ Fail	Error: Source/code files changed without an accompanying unit test. Expected: add at least one test file named like test_<name>.py / test_<name>.cpp (or <name>_test.*). Current: code file(s) changed: projects/rccl/test/DoneEventOrderingTests.cpp; no test file found
🔎 pre-commit	⏳ Pending	⏳ Still running…
🚫 Draft PR	🔜 To Be Enabled	—
🚩 Feature Flag	🔜 To Be Enabled	—
📊 Code Coverage	🔜 To Be Enabled	—

⚠️ 1 policy check(s) failed. Please address the issues above before this PR can be Reviewed.

🚫 Please fix the failed policies

• ❌ Unit Test

The Not ready to Review label was added to this PR. Once all policies pass, the label is removed automatically.

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

[therock-pr-bot]

🚫 Please fix the failed policies before requesting reviews.

The following policy checks failed:

• ❌ Unit Test

The Not ready to Review label has been added to this PR.
Once all policies pass, the label will be removed automatically.

[rahulvaidya20]

Unit Test checks by PR bot not applicable for RCCL.

[Copilot]

Pull request overview

Adds new RCCL unit tests that reproduce and guard against a single-stream fast-launch doneEvent stream-ordering regression (hang / incorrect ordering after checkpoint-like teardown/reinit), and wires them into CI precheckin.

Changes:

• Added DoneEventOrderingTests.cpp with six process-isolated gtest cases covering alternating streams, reinit cycles, grouped calls, default-stream transitions, and lazy stream-switch behavior.
• Registered the new test file in projects/rccl/test/CMakeLists.txt so it builds into rccl-UnitTests.
• Added the six tests to ci-precheckin.json so they run in CI with explicit per-test timeouts.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
projects/rccl/tools/scripts/test_runner/configs/ci-precheckin.json	Adds CI entries for the six new DoneEventOrdering tests.
projects/rccl/test/DoneEventOrderingTests.cpp	New process-isolated unit tests targeting doneEvent stream-ordering correctness.
projects/rccl/test/CMakeLists.txt	Includes the new test source in the unit test build.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.