[Metal][DXC] WaveMultiPrefixCountBits fails with inactive lanes
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Newbie friendliness
- 65/100
- Issue type
- Bug
- Clarity
- Mostly clear
- Activity status
- Active
- Tech stack
- cpp
- Domain
- testing-qa
Research direction
Start with the WaveMultiPrefixCountBits.test reproduction from PR #1705 and the linked CI log; run it in the mtl configuration with DXC. Inspect the inactive-lane case and capture the returned counts, since the current test records only pass/fail flags. Done when the cause is confirmed and the test produces the expected result: all 128 OutDivergent flags equal 1.
Written by the indexing model from the issue text.
Description
The inactive-lane case in WaveMultiPrefixCountBits.test fails with DXC on Metal. Compilation and pipeline creation succeed, but the output check fails.
- Tests: https://github.com/llvm/offload-test-suite/pull/1705
- CI log: https://github.com/llvm/offload-test-suite/actions/runs/42183398383/job/127777973733
Reproduction
Run WaveMultiPrefixCountBits.test from PR #1705 in the mtl configuration with DXC.
The failing case calls the builtin only where WaveGetLaneIndex() % 3 != 1. It supplies a full mask. The reference counts true predicates in earlier active lanes and excludes inactive lanes.
Expected result
All 128 flags in OutDivergent equal 1.
Actual result
80 flags equal 0, and 48 equal 1. The runner reports:
Test failed: TestDivergent
Comparison Rule: BufferExact
Possible cause
The output pattern matches a model that incorrectly counts inactive lanes in 32-lane waves. However, the test stores only pass/fail flags. It does not record the returned counts, so the exact cause remains unconfirmed.
- Dominant language
- C++
- Stars
- 20
- Forks
- 41
- Avg merge
- 5d 2h
- Merged PRs (30d)
- 29
Getting set up
This project ships no dev container, Dockerfile or contributing guide, so setting up is up to you: start from its README, and see our first-contribution guide for the general steps.
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from llvm/offload-test-suite
-
Change Lavapipe installation on DXC CIPossibly taken @joaosaffran claimed this 1 day ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
llvm/offload-test-suite#1703 · 1 assignee ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
llvm/offload-test-suite#1698 ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
llvm/offload-test-suite#1562 ·
Maintainers usually reply within 1 day
-
[LongVec] Add a abs test casePossibly taken @farzonl claimed this 17 days ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 65/100
llvm/offload-test-suite#1441 · 2 comments · 1 assignee ·
Maintainers usually reply within 1 day
-
Difficulty 4/5 3-5 days Newbie friendliness 48/100
llvm/offload-test-suite#1702 ·
Maintainers usually reply within 1 day
All issues in llvm/offload-test-suite
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 66/100
xiaoyifang/goldendict-ng#3148 ·
Maintainers usually reply within 1 day
-
Difficulty 1/5 Under an hour Newbie friendliness 91/100
triton-inference-server/server#9069 ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
Maintainers usually reply within 4 days
-
documentation
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
-
[Bug]: Launching Space Calibrator causes Bluetooth mouse lag due to a 100% duty-cycle BLE scanPossibly taken @Yuuki204 claimed this today. Openbug
Difficulty 2/5 1-3 hours Newbie friendliness 78/100