[Mlir-commits] [mlir] [mlir] Add FP software implementation lowering pass: `arith-to-apfloat` (PR #166618)

[Matthias Springer]

matthias-springer wrote:

> I'd have imagined trunc and ext would also be part of initial coverage, but perhaps that already works (although messing up rounding is easy) or perhaps as you are focusing not on heterogeneous floating point computations (yet).

I was mainly trying to get the smallest possible skeleton merged, so that Max and me can start working on the other ops in parallel.

I believe the rounding modes won't be a problem. We just follow whatever the `arith` op folders are doing. (They are also going through APFloat, just at compile time.) In any case, for most `arith` ops, the rounding mode is unspecified at the moment. So there should be no expectation from the user's side that a particular rounding mode is chosen.


https://github.com/llvm/llvm-project/pull/166618