[clang] [llvm] [NVPTX] Add builtins and intrinsics for conversions of new FP types (PR #134345)

[Srinivasa Ravi via cfe-commits]

================
@@ -1548,6 +1548,45 @@ let TargetPrefix = "nvvm" in {
       Intrinsic<[llvm_v2f16_ty], [llvm_i16_ty], [IntrNoMem, IntrNoCallback]>;
   def int_nvvm_e5m2x2_to_f16x2_rn_relu : ClangBuiltin<"__nvvm_e5m2x2_to_f16x2_rn_relu">,
       Intrinsic<[llvm_v2f16_ty], [llvm_i16_ty], [IntrNoMem, IntrNoCallback]>;
+  
+  def int_nvvm_ff_to_e2m3x2_rn : ClangBuiltin<"__nvvm_ff_to_e2m3x2_rn">,
----------------
Wolfram70 wrote:

Used foreach loops to simplify these definitions in the latest revision. Perhaps we could do this for the other CVT intrinsics too since many appear to be very similar in terms of their signatures and properties.

https://github.com/llvm/llvm-project/pull/134345