Writing a few intrinsics where necessary is not really comparable to the work required do reimplement something like CUDA on AMD (or equivalently deal with ROCm)