[Mlir-commits] [mlir] [mlir][arith] Add rounding mode flags to binary arithmetic operations (PR #188458)

[Mehdi Amini]

================
@@ -957,14 +988,17 @@ def Arith_NegFOp : Arith_FloatUnaryOp<"negf"> {
 // AddFOp
 //===----------------------------------------------------------------------===//
 
-def Arith_AddFOp : Arith_FloatBinaryOp<"addf", [Commutative]> {
+def Arith_AddFOp : Arith_FloatBinaryOpWithRoundingMode<"addf", [Commutative]> {
   let summary = "floating point addition operation";
   let description = [{
     The `addf` operation takes two operands and returns one result, each of
     these is required to be the same type. This type may be a floating point
     scalar type, a vector whose element type is a floating point type, or a
     floating point tensor.
 
+    If the value cannot be exactly represented, it is rounded using the
+    provided rounding mode or the default one if no rounding mode is provided.
----------------
joker-eph wrote:

Right: we need the arith dialect documentation to be self contained basically. We may refer to LLVM in some way, but we can't assume it in "the one environment" where arith operates right now.

https://github.com/llvm/llvm-project/pull/188458