Basically my exercises for the Tutorial wouldn't compile. This fixes the issues. Generally I think we may wanna have CUDA only abi compile on the host but throw or so if you actually call anything? For example how do we deep_copy(View,scalar) with cuda_warp storage as the view value_type etc.
Basically my exercises for the Tutorial wouldn't compile. This fixes the issues. Generally I think we may wanna have CUDA only abi compile on the host but throw or so if you actually call anything? For example how do we deep_copy(View,scalar) with cuda_warp storage as the view value_type etc.