Jack 4b8ff4c63d
[mlir][math] Fix the semantics of math.clampf (#175012)
The `math.clampf` op is semantically incorrect when compared to both the
CUDA reference implementation and the SPIRV spec, both of which have a
clamp op.
- Fix the definition of `math.clampf` to agree with CUDA and SPIRV
- Explicitly state when `math.clampf` produces `ub.poison`
- Update the ExpandOps pass to reflect the corrected semantics
2026-01-19 10:49:01 +00:00
..