[clang][amdgpu] Prefer not using `fp16` conversion intrinsics.

Reviewers: yaxunl, arsenm

Subscribers: kzhuravl, jvesely, wdng, nhaehnle, dstuttard, tpr, t-tye, kerbowa, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D81849

clang/lib/Basic/Targets/AMDGPU.h

@@ -219,6 +219,8 @@ public:
 
   ArrayRef<Builtin::Info> getTargetBuiltins() const override;
 
+  bool useFP16ConversionIntrinsics() const override { return false; }
+
   void getTargetDefines(const LangOptions &Opts,
                         MacroBuilder &Builder) const override;
 

clang/test/CodeGenHIP/half.hip

@@ -0,0 +1,16 @@
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -x hip -emit-llvm -fcuda-is-device -o - %s | FileCheck %s
+
+#define __device__ __attribute__((device))
+
+// CHECK-LABEL: @_Z2d0DF16_
+// CHECK: fpext
+__device__ float d0(_Float16 x) {
+  return x;
+}
+
+// CHECK-LABEL: @_Z2d1f
+// CHECK: fptrunc
+__device__ _Float16 d1(float x) {
+  return x;
+}