Issue 242938
Summary Refine `KnownFPClass::fmul`/`KnownFPClass::fdiv`
Labels floating-point
Assignees ZERICO2005
Reporter ZERICO2005
    Tracking issue for the types of refinements we can make for `fmul`/`fdiv` without the magnitude refinement for `KnownFPClass` https://github.com/llvm/llvm-project/issues/240038

## fmul
- [ ] subnormal * normal never overflows.
- [ ] subnormal * subnormal always underflows (The result is either 0.0 or subnormal_min).
- [ ] directed rounding modes may saturate to normal_max instead of infinity.
- [ ] directed rounding modes may saturate to subnormal_min instead of zero.

## fdiv
- [ ] normal / subnormal never underflows.
- [ ] subnormal / normal never overflows.
- [ ] subnormal / subnormal will never overflow or underflow.
- [ ] directed rounding modes may saturate to normal_max instead of infinity.
- [ ] directed rounding modes may saturate to subnormal_min instead of zero.
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs

Reply via email to