This is a helper wrapper around the multi-gpu data blocks owned by a worker. It's design is NOT final. Its so written this way to get something concrete in a short span of time. More...
#include <data.hpp>
Collaboration diagram for MLCommon::Matrix::Data< Type >:
Public Member Functions | |
| Data () | |
| Data (Type *_ptr, size_t _n_elements) | |
| size_t | numElements () const |
Public Attributes | |
| Type * | ptr = nullptr |
| size_t | totalSize = (size_t)0 |
Detailed Description
template<typename Type>
struct MLCommon::Matrix::Data< Type >
This is a helper wrapper around the multi-gpu data blocks owned by a worker. It's design is NOT final. Its so written this way to get something concrete in a short span of time.
- Todo:
- add support for custom allocators
Constructor & Destructor Documentation
◆ Data() [1/2]
template<typename Type >
|
inline |
◆ Data() [2/2]
template<typename Type >
|
inline |
Member Function Documentation
◆ numElements()
template<typename Type >
|
inline |
Return the number of elements of Type in ptr.
Member Data Documentation
◆ ptr
template<typename Type >
| Type* MLCommon::Matrix::Data< Type >::ptr = nullptr |
actual data block. This is just a linearly laid out buffer of all blocks owned by this worker
◆ totalSize
template<typename Type >
| size_t MLCommon::Matrix::Data< Type >::totalSize = (size_t)0 |
total size (in bytes) of this buffer. In future, this will be passed to the dealloc function underneath
The documentation for this struct was generated from the following file:
- include/cuml/prims/opg/matrix/data.hpp
