[Mlir-commits] [mlir] [mlir][arith] Add minimumnumf and maximumnumf operations (PR #221658)

[lorenzo chelini]

================
@@ -765,13 +765,15 @@ Value xegpu::createReductionNeutralValue(OpBuilder &builder, Location loc,
 
   case vector::CombiningKind::MINNUMF:
   case vector::CombiningKind::MINIMUMF:
+  case vector::CombiningKind::MINIMUMNUMF:
     if (auto floatTy = dyn_cast<FloatType>(elemTy))
       return makeConst(builder.getFloatAttr(
           elemTy, APFloat::getInf(floatTy.getFloatSemantics())));
     return nullptr;
 
   case vector::CombiningKind::MAXNUMF:
   case vector::CombiningKind::MAXIMUMF:
+  case vector::CombiningKind::MAXIMUMNUMF:
----------------
chelini wrote:

yes, improved `getMaskNeutralValue` few lines above.

https://github.com/llvm/llvm-project/pull/221658