[MLIR] Nested affine loop differentiation may not be affine due to pop cache
Open
- Dominant language
- LLVM
- Stars
- 1.7k
- Forks
- 188
- Avg merge
- 1d 22h
- Merged PRs (30d)
- 26
Description
cc @pengmai
Contributor guide
Research direction
The issue concerns nested affine loop differentiation in MLIR and a possible pop-cache problem. Start by locating the relevant MLIR affine differentiation path and investigating how the pop cache affects nested loops; done means the differentiated result remains affine.
Written by the indexing model from the issue text.
Assessment
- Domain
- compilers
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100