InferenceEngine::CPUStreamsExecutor Class Reference

CPU Streams executor implementation. The executor splits the CPU into groups of threads, that can be pinned to cores or NUMA nodes. It uses custom threads to pull tasks from single queue. More...

#include <ie_cpu_streams_executor.hpp>

Inheritance diagram for InferenceEngine::CPUStreamsExecutor:

Public Types
using	Ptr = std::shared_ptr< CPUStreamsExecutor >
	A shared pointer to a CPUStreamsExecutor object.
Public Types inherited from InferenceEngine::IStreamsExecutor
enum	ThreadBindingType : std::uint8_t {   NONE ,   CORES ,   NUMA }
	Defines thread binding type. More...
using	Ptr = std::shared_ptr< IStreamsExecutor >
Public Types inherited from InferenceEngine::ITaskExecutor
using	Ptr = std::shared_ptr< ITaskExecutor >

Public Member Functions
	CPUStreamsExecutor (const Config &config={})
	Constructor. More...
	~CPUStreamsExecutor () override
	A class destructor.
void	run (Task task) override
	Execute InferenceEngine::Task inside task executor context. More...
void	Execute (Task task) override
	Execute the task in the current thread using streams executor configuration and constraints. More...
int	GetStreamId () override
	Return the index of current stream. More...
int	GetNumaNodeId () override
	Return the id of current NUMA Node. More...
Public Member Functions inherited from InferenceEngine::IStreamsExecutor
	~IStreamsExecutor () override
	A virtual destructor.
Public Member Functions inherited from InferenceEngine::ITaskExecutor
virtual	~ITaskExecutor ()=default
	Destroys the object.
virtual void	runAndWait (const std::vector< Task > &tasks)
	Execute all of the tasks and waits for its completion. Default runAndWait() method implementation uses run() pure virtual method and higher level synchronization primitives from STL. The task is wrapped into std::packaged_task which returns std::future. std::packaged_task will call the task and signal to std::future that the task is finished or the exception is thrown from task Than std::future is used to wait for task execution completion and task exception extraction. More...

Detailed Description

CPU Streams executor implementation. The executor splits the CPU into groups of threads, that can be pinned to cores or NUMA nodes. It uses custom threads to pull tasks from single queue.

Constructor & Destructor Documentation

CPUStreamsExecutor()

InferenceEngine::CPUStreamsExecutor::CPUStreamsExecutor ( const Config &  config = {} )

explicit

Constructor.

Parameters

config

Stream executor parameters

Member Function Documentation

Execute()

void InferenceEngine::CPUStreamsExecutor::Execute ( Task  task )

overridevirtual

Execute the task in the current thread using streams executor configuration and constraints.

Parameters

task	A task to start

Implements InferenceEngine::IStreamsExecutor.

GetNumaNodeId()

int InferenceEngine::CPUStreamsExecutor::GetNumaNodeId ( )

overridevirtual

Return the id of current NUMA Node.

Returns

ID of current NUMA Node, or throws exceptions if called not from stream thread

Implements InferenceEngine::IStreamsExecutor.

GetStreamId()

int InferenceEngine::CPUStreamsExecutor::GetStreamId ( )

overridevirtual

Return the index of current stream.

Returns

An index of current stream. Or throw exceptions if called not from stream thread

Implements InferenceEngine::IStreamsExecutor.

run()

void InferenceEngine::CPUStreamsExecutor::run ( Task  task )

overridevirtual

Execute InferenceEngine::Task inside task executor context.

Parameters

task	A task to start

Implements InferenceEngine::ITaskExecutor.

---

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

• threading/ie_cpu_streams_executor.hpp