[Mlir-commits] [mlir] [mlir][vector] Add pattern to break down reductions into arith ops (PR #75727)

[Andrzej Warzyński]

================
@@ -6212,6 +6214,7 @@ bool WarpExecuteOnLane0Op::areTypesCompatible(Type lhs, Type rhs) {
 
 Value mlir::vector::makeArithReduction(OpBuilder &b, Location loc,
                                        CombiningKind kind, Value v1, Value acc,
+                                       arith::FastMathFlagsAttr fastmath,
----------------
banach-space wrote:

The noise level is low, so that's not a concern. However, these attributes might affect things in a very subtle way and that's why I feel the deserve a dedicated commit. I would happily review that and accept (it's very straightforward).

https://github.com/llvm/llvm-project/pull/75727