My Project
|
A view towards a smart pointer to GPU-allocated memory. More...
#include <gpu_smart_pointer.hpp>
Public Member Functions | |
PointerView (const PointerView &other)=default | |
PointerView (const std::shared_ptr< T > &ptr) | |
template<class Deleter > | |
PointerView (const std::unique_ptr< T, Deleter > &ptr) | |
PointerView (T *ptr) | |
OPM_HOST_DEVICE T * | get () const |
OPM_HOST_DEVICE T & | operator* () const |
OPM_HOST_DEVICE T * | operator-> () const |
A view towards a smart pointer to GPU-allocated memory.
This will emulate a smart pointer to GPU-allocated memory, but without ownership semantics, and being compatible with the requirements of the GPU kernels. This is useful when we want to pass a smart pointer to a GPU kernel, but we do not want to transfer the ownership of the memory.