llvm-project

d27672b5 - [AMDGPU] Modifies raytracing and wmma builtin def to take _Float16 for HIP/C++ (#175039)

[AMDGPU] Modifies raytracing and wmma builtin def to take _Float16 for HIP/C++ (#175039) For raytrace and wmma builtins, using 'x' in the def to take _Float16 for HIP/C++ and half for OpenCL.