Convert shape.shape to tensor<?xindex> when lowering the shape to the
std dialect.
Details
- Reviewers
pifon2a jpienaar - Commits
- rG867bc41e85cb: [MLIR] Add type conversion for `shape.shape`
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
mlir/lib/Conversion/ShapeToStandard/ShapeToStandard.cpp | ||
---|---|---|
60 | why is it always dynamic? |
mlir/lib/Conversion/ShapeToStandard/ShapeToStandard.cpp | ||
---|---|---|
60 | The actual shape is not dynamic but the shape type is really not holding that information at compile time. |
mlir/lib/Conversion/ShapeToStandard/ShapeToStandard.cpp | ||
---|---|---|
60 | Yes this is where the type/value boundary gets spanned (what is part of the type in tensor is part of the value in shape). We could expand shape's type to hold rank (known, unknown, error) but starting with dynamic and having this be refined along with the lowering of the rest of the shape computations should work too. |
why is it always dynamic?