lmp::tensor::detail::DispatchStub< Fn > Struct Template Reference

simple static registration class More...

#include <dispatch_stub.hpp>

Public Types
using	fn_type = Fn

Public Member Functions
void	register_kernel (DeviceType dev, fn_type f)
template<typename... Args>
decltype(auto)	operator() (DeviceType dev, Args &&... args) const

Public Attributes
std::array< fn_type, static_cast< size_t >(DeviceType::Count)>	table_ {}

Detailed Description

template<typename Fn>
struct lmp::tensor::detail::DispatchStub< Fn >

simple static registration class

this class keeps a dispatch table (table_), which directs to the correct overload during runtime. e.g. calling add_stub()(DeviceType::CUDA, a, b) calls add_cuda on the tensors a and b

See also

DeviceType

---

The documentation for this struct was generated from the following file:

• csrc/include/lamppp/tensor/dispatch_stub.hpp