Memory description and management
Collaboration diagram for Memory description and management:
|
Namespaces | |
| cldnn::details | |
| Helper structs used in tensor constructor with dim_vec_kinds. | |
Classes | |
| struct | cldnn::type_to_data_type< T > |
| Converts C++ type to data_types . More... | |
| struct | cldnn::data_type_to_type< Data_Type > |
| Converts data_types to C++ type. More... | |
| struct | cldnn::data_type_traits |
| Helper class to identify key properties for data_types. More... | |
| struct | cldnn::padding |
| Represents data padding information. More... | |
| struct | cldnn::layout |
| Describes memory layout. More... | |
| struct | cldnn::pointer< T > |
| Helper class to get an access memory data. More... | |
| struct | cldnn::memory |
| Represents buffer with particular layout. More... | |
| struct | cldnn::format_traits |
| Format information helper class. More... | |
| struct | cldnn::format |
| Represents memory formats (orders). In CNN most of data is described as 4 dimensional blocks. In Intel(R) clDNN library we describe memory with 4 letters. More... | |
| struct | cldnn::tensor |
| N-dimensional vector. Mostly used to represent memory size. More... | |
Enumerations | |
| enum | cldnn::data_types : size_t { i8 = cldnn_i8, cldnn::data_types::u8 = cldnn_u8, f16 = cldnn_f16, f32 = cldnn_f32 } |
| Possible data types could be stored in memory. More... | |
Functions | |
| template<typename T > | |
| bool | cldnn::data_type_match (data_types data_type) |
Helper function to check if C++ type matches data_type. | |
| constexpr auto | cldnn::fuse (data_types dt, cldnn::format::type fmt) -> decltype(static_cast< std::underlying_type< data_types >::type >(dt)|static_cast< std::underlying_type< format::type >::type >(fmt)) |
| Helper function to get both data_types and format::type in a single, unique value. Useable in 'case' statement. | |
| template<typename ... InitTys> | |
| details::dim_vec_kind_init< details::dim_vec_kind::batch > | cldnn::batch (InitTys &&... inits) |
| template<typename ... InitTys> | |
| details::dim_vec_kind_init< details::dim_vec_kind::feature > | cldnn::feature (InitTys &&... inits) |
| template<typename ... InitTys> | |
| details::dim_vec_kind_init< details::dim_vec_kind::spatial > | cldnn::spatial (InitTys &&... inits) |
| tensor | cldnn::operator+ (const tensor &lhs, const tensor &rhs) |
Adds two tensors. | |
| tensor | cldnn::operator- (const tensor &lhs, const tensor &rhs) |
Subtracts two tensors. | |
| tensor | cldnn::operator* (const tensor &lhs, tensor::value_type rhs) |
Multiplies a tensor to a scalar. | |
| tensor | cldnn::operator/ (const tensor &lhs, tensor::value_type rhs) |
Divides a tensor by a scalar. | |
| void | cldnn::details::dim_vec_kind_init< Kind >::init_tensor_values (cldnn::tensor &t) |
Detailed Description
Class Documentation
◆ cldnn::type_to_data_type
| struct cldnn::type_to_data_type |
template<typename T>
struct cldnn::type_to_data_type< T >
Converts C++ type to data_types .
Definition at line 41 of file layout.hpp.
◆ cldnn::data_type_to_type
| struct cldnn::data_type_to_type |
template<data_types Data_Type>
struct cldnn::data_type_to_type< Data_Type >
Converts data_types to C++ type.
Definition at line 50 of file layout.hpp.
Enumeration Type Documentation
◆ data_types
|
strong |
Possible data types could be stored in memory.
| Enumerator | |
|---|---|
| u8 | Not supported in current HW. |
Definition at line 32 of file layout.hpp.
