[Mlir-commits] [mlir] [MLIR][NVGPU] Add convert.fpext and convert.fptrunc Ops (PR #199700)

[Guray Ozen]

================
@@ -671,4 +671,80 @@ def NVGPU_RcpOp : NVGPU_Op<"rcp", [Pure,
   let hasVerifier = 1;
 }
 
+//===----------------------------------------------------------------------===//
+// NVGPU Conversion Ops
+//===----------------------------------------------------------------------===//
+
+def Int8OrFloatLike : TypeConstraint<
+    Or<[FloatLike.predicate,
+        I8.predicate,
+        ValueSemanticsContainerOf<[I8]>.predicate]>,
+    "scalar, vector, or tensor of i8 or floats">;
+def AnyI32Like : TypeOrValueSemanticsContainer<I32, "scalar i32 or vector of i32">;
+
+def NVGPU_FPTruncOp : NVGPU_Op<"convert.fptrunc",
----------------
grypp wrote:

I am wondering why don't we have `convert` OP that handles truncate or extension? Because what it happens is on the in/out types. Can we have a single OP? 

https://github.com/llvm/llvm-project/pull/199700