test(KERNEL-QUANT-CIQ-GEMM-ROCM): match cooperative-tile toggle parsing
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 423
- Forks
- 53
- Avg merge
- 20h 26m
- Merged PRs (30d)
- 310
Description
Row: KERNEL-QUANT-CIQ-GEMM-ROCM
The cooperative-tile tests in PR #3036 select BIGTILE and SHARE_ACT by environment-variable presence. The production dispatch enables these options only when their value is exactly 1.
At head d295f00cb227cbde4fd1b0b50b31f1a057ab6459, tests/vt/test_backend_cross_device.cpp:3540 and :3665 therefore disagree with the production predicates in src/vt/rocm/rocm_grouped_gemm.hip:2144 and :2162.
For example, WIDE=1, SHARE_ACT=1, BIGTILE=0 selects the Shared kernel, but the tests expect BigTile counters. Explicit zero values must be treated as disabled. The tests must also respect the master WMMA switch.
Repair this in PR #3036. Run the toggle matrix on gfx1200 or gfx1201 and independently mutate the test predicate to prove regression coverage. Static finding confirmed; GPU reproduction remains pending.
Contributor guide
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.
Research direction
Start with the test predicates at tests/vt/test_backend_cross_device.cpp:3540 and :3665, then compare them with the production predicates in src/vt/rocm/rocm_grouped_gemm.hip:2144 and :2162. Run the toggle matrix on gfx1200 or gfx1201, including explicit zero values and the master WMMA switch. Done means the tests match production parsing and mutation of the test predicate demonstrates regression coverage.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- testing-qa
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 68/100