Memory that describes the data. More...

#include <dnnl.hpp>

Inheritance diagram for dnnl::memory:

Collaboration diagram for dnnl::memory:

Classes
struct	desc
	A memory descriptor. More...

Public Types
enum	data_type { data_type::undef = dnnl_data_type_undef, data_type::f16 = dnnl_f16, data_type::bf16 = dnnl_bf16, data_type::f32 = dnnl_f32, data_type::s32 = dnnl_s32, data_type::s8 = dnnl_s8, data_type::u8 = dnnl_u8 }
	Data type specification. More...

enum	format_kind { format_kind::undef = dnnl_format_kind_undef, format_kind::any = dnnl_format_kind_any, format_kind::blocked = dnnl_blocked, format_kind::wino = dnnl_format_kind_wino, format_kind::packed = dnnl_format_kind_rnn_packed }
	Memory format kind. More...

enum	format_tag { format_tag::undef = dnnl_format_tag_undef, format_tag::any = dnnl_format_tag_any, format_tag::a = dnnl_a, format_tag::ab = dnnl_ab, format_tag::abc = dnnl_abc, format_tag::abcd = dnnl_abcd, format_tag::abcde = dnnl_abcde, format_tag::abcdef = dnnl_abcdef, format_tag::abdec = dnnl_abdec, format_tag::acb = dnnl_acb, format_tag::acbde = dnnl_acbde, format_tag::acdb = dnnl_acdb, format_tag::acdeb = dnnl_acdeb, format_tag::ba = dnnl_ba, format_tag::bac = dnnl_bac, format_tag::bacd = dnnl_bacd, format_tag::bcda = dnnl_bcda, format_tag::cba = dnnl_cba, format_tag::cdba = dnnl_cdba, format_tag::cdeba = dnnl_cdeba, format_tag::decab = dnnl_decab , format_tag::aBcd8b = dnnl_aBcd8b , format_tag::ABcd8b8a = dnnl_ABcd8b8a , format_tag::nc = dnnl_nc , format_tag::nchw = dnnl_nchw, format_tag::nhwc = dnnl_nhwc, format_tag::chwn = dnnl_chwn }
	Memory format tag specification. More...

Public Member Functions
	memory (const desc &md, const engine &aengine, void *ahandle)
	Constructs a memory. More...

	memory (const desc &md, const engine &aengine)
	Constructs a memory. More...

desc	get_desc () const
	Returns the descriptor of the memory.

engine	get_engine () const
	Returns the engine of the memory.

void *	get_data_handle () const
	Returns a handle of the data contained in the memory. More...

template<typename T = void>
T *	map_data () const
	Maps the data of the memory. More...

void	unmap_data (void *mapped_ptr) const
	Unmaps the previously mapped data for the memory. More...

cl_mem	get_ocl_mem_object () const
	Returns the OpenCL memory object associated with the memory.

void	set_ocl_mem_object (cl_mem mem_object)
	Sets the OpenCL memory object `mem_object` associated with the memory.

Public Member Functions inherited from dnnl::handle< dnnl_memory_t >
	handle ()=default
	Empty constructor. More...

	handle (dnnl_memory_t t, bool weak=false)
	Constructs a C handle wrapper from a C handle. More...

void	reset (dnnl_memory_t t, bool weak=false)
	Resets the value of a C handle. More...

dnnl_memory_t	get (bool allow_emtpy=false) const
	Returns the value of the underlying C handle.

Detailed Description

Memory that describes the data.

Examples:: cnn_inference_f32.cpp, cnn_inference_int8.cpp, cnn_training_f32.cpp, cpu_cnn_training_bf16.cpp, cpu_rnn_inference_f32.cpp, cpu_rnn_inference_int8.cpp, cross_engine_reorder.cpp, getting_started.cpp, gpu_opencl_interop.cpp, memory_format_propagation.cpp, performance_profiling.cpp, and rnn_training_f32.cpp.

Member Enumeration Documentation

◆ data_type

enum dnnl::memory::data_type

strong

Data type specification.

Enumerator
undef	Undefined data type, used for empty memory descriptors.
f16	16-bit/half-precision floating point.
bf16	non-standard 16-bit (bfloat16 w/ 7 bit mantissa) floating point.
f32	32-bit/single-precision floating point.
s32	32-bit signed integer.
s8	8-bit signed integer.
u8	8-bit unsigned integer.

Examples:: cnn_inference_f32.cpp, cnn_inference_int8.cpp, cnn_training_f32.cpp, cpu_cnn_training_bf16.cpp, and rnn_training_f32.cpp.

◆ format_kind

enum dnnl::memory::format_kind

strong

Memory format kind.

Enumerator
undef	Undefined memory format kind, used for empty memory descriptors.
any	Unspecified format kind. The primitive selects a format automatically.
blocked	A tensor in a generic format described by the stride and blocking values in each dimension. See dnnl_blocking_desc_t for more information.
wino	Weights format used in 8bit Winograd convolution.
packed	Packed weights format used in RNN.

◆ format_tag

enum dnnl::memory::format_tag

strong

Memory format tag specification.

See dnnl_format_tag_t for a detailed description.

Enumerator
undef	Undefined memory format tag.
any	Placeholder memory format tag. The primitive selects a format automatically.
a	plain 1D tensor
ab	plain 2D tensor
abc	plain 3D tensor
abcd	plain 4D tensor
abcde	plain 5D tensor
abcdef	plain 6D tensor
abdec	permuted 5D tensor
acb	permuted 3D tensor
acbde	permuted 5D tensor
acdb	permuted 4D tensor
acdeb	permuted 5D tensor
ba	permuted 2D tensor
bac	permuted 3D tensor
bacd	permuted 4D tensor
bcda	permuted 4D tensor
cba	permuted 3D tensor
cdba	permuted 4D tensor
cdeba	permuted 5D tensor
decab	permuted 5D tensor
aBcd8b	4D tensor blocked by 2nd dimension with block size 8
ABcd8b8a	4D tensor blocked by 1st and 2nd dimension with block size 8
nc	2D CNN activations tensor, an alias to dnnl::memory::format_tag::ab
nchw	4D CNN activations tensor, an alias to dnnl::memory::format_tag::abcd
nhwc	4D CNN activations tensor, an alias to dnnl::memory::format_tag::acdb
chwn	4D CNN activations tensor, an alias to dnnl::memory::format_tag::bcda

Examples:: cnn_inference_f32.cpp, cnn_inference_int8.cpp, cnn_training_f32.cpp, cpu_cnn_training_bf16.cpp, and rnn_training_f32.cpp.

Constructor & Destructor Documentation

◆ memory() [1/2]

dnnl::memory::memory	(	const desc &	md,
		const engine &	aengine,
		void *	ahandle
	)

inline

Constructs a memory.

Parameters

md	Memory descriptor.
aengine	Engine.
ahandle	handle.

◆ memory() [2/2]

dnnl::memory::memory	(	const desc &	md,
		const engine &	aengine
	)

inline

Constructs a memory.

Parameters

md	Memory descriptor.
aengine	Engine.

Member Function Documentation

◆ get_data_handle()

void* dnnl::memory::get_data_handle ( ) const

inline

Returns a handle of the data contained in the memory.

On the CPU engine, this is a pointer to the allocated memory.

Examples:: cpu_rnn_inference_f32.cpp, and cpu_rnn_inference_int8.cpp.

◆ map_data()

template<typename T = void>

T* dnnl::memory::map_data ( ) const

inline

Maps the data of the memory.

Mapping allows to read/write directly from/to the memory contents for engines that do not support direct memory access.

Mapping is an exclusive operation - a memory object cannot be used in other operations until this memory object is unmapped.

Template Parameters

T	Type of the pointer to be mapped.

Note: Any primitives working with the memory should be completed before mapping. Use stream::wait() to synchronize the corresponding execution stream.; Map/unmap API is provided mainly for debug/testing purposes and its performance may be suboptimal.

Examples:: cross_engine_reorder.cpp.

◆ unmap_data()

void dnnl::memory::unmap_data ( void * mapped_ptr ) const

inline

Unmaps the previously mapped data for the memory.

Any changes of the mapped data are synchronized back to the memory after the call is complete. The mapped pointer must be obtained through a map_data() call.

Note: Map/unmap API is provided mainly for debug/testing purposes and its performance may be suboptimal.

Examples:: cross_engine_reorder.cpp.

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

include/dnnl.hpp

Classes

Public Types

Public Member Functions

Detailed Description

Member Enumeration Documentation

◆ data_type

◆ format_kind

◆ format_tag

Constructor & Destructor Documentation

◆ memory() [1/2]

◆ memory() [2/2]

Member Function Documentation

◆ get_data_handle()

◆ map_data()

◆ unmap_data()