Abstract Task specialized for CUDA computation. More...

#include "abstract_cuda_task.h"

Inheritance diagram for hh::AbstractCUDATask< TaskOutput, TaskInputs >:

Collaboration diagram for hh::AbstractCUDATask< TaskOutput, TaskInputs >:

Public Member Functions
	AbstractCUDATask (size_t numberThreads=1)
	Default constructor for an AbstractCUDATask. More...

	AbstractCUDATask (std::string_view const &name, size_t numberThreads=1)
	Constructor for an AbstractCUDATask with name as mandatory parameter and the number of threads as optional parameter. More...

	AbstractCUDATask (std::string_view const &name, size_t numberThreads, bool automaticStart, bool enablePeerAccess)
	Constructor for an AbstractCUDATask with name, number of threads, automaticStart and enablePeerAccess as mandatory parameter. More...

void	initialize () final
	Initialize an AbstractCUDATask to bound it to a CUDA device, and do the peer access if enabled. At the end will call AbstractCUDATask::initializeCuda.

void	shutdown () final
	Shutdown an AbstractCUDATask to destroy the task's CUDA stream created during AbstractCUDATask::initialize. First calls AbstractCUDATask::shutdownCuda.

virtual void	initializeCuda ()
	Virtual initialization step, where user defined data structure can be initialized.

virtual void	shutdownCuda ()
	Virtual shutdown step, where user defined data structure can be destroyed.

bool	enablePeerAccess () const
	Accessor for peer access choice. More...

cudaStream_t	stream () const
	Getter for CUDA task's stream. More...

bool	hasPeerAccess (int peerDeviceId)
	Accessor for peer access enabled for a specific device id. More...

Public Member Functions inherited from hh::AbstractTask< TaskOutput, TaskInputs... >
	AbstractTask ()
	AbstractTask default constructor. More...

	AbstractTask (std::string_view const &name)
	Create a task with a custom name. More...

	AbstractTask (std::string_view const &name, size_t numberThreads)
	Create a task with a custom name and the task's cluster size. More...

	AbstractTask (std::string_view const &name, size_t numberThreads, bool automaticStart)
	Create a task with a custom name, the task's cluster size, and the automatic start. More...

	AbstractTask (std::string_view const name, size_t numberThreads, core::NodeType nodeType, bool automaticStart)
	Internal constructor needed to create nodes that derive from AbstractTask. More...

	AbstractTask (AbstractTask< TaskOutput, TaskInputs ... > *rhs)
	Copy constructor. More...

virtual	~AbstractTask ()=default
	Default destructor.

void	addResult (std::shared_ptr< TaskOutput > output)
	Add an output data. More...

std::string_view	name ()
	Task's name accessor. More...

size_t	numberThreads ()
	Task's number of threads accessor. More...

bool	automaticStart ()
	Task's automatic start accessor. More...

core::NodeType	nodeType ()
	Task's node type accessor. More...

int	deviceId ()
	Task's device ID accessor. More...

int	graphId ()
	Task's graph ID accessor. More...

std::shared_ptr< core::CoreNode >	core () final
	Task's core accessor. More...

std::shared_ptr< AbstractMemoryManager< TaskOutput > > const &	memoryManager () const
	Task's memory manager accessor. More...

virtual std::shared_ptr< AbstractTask< TaskOutput, TaskInputs... > >	copy ()
	Default copy overload, fail if cluster need to be copied. More...

void	connectMemoryManager (std::shared_ptr< AbstractMemoryManager< TaskOutput >> mm)
	Connect a memory manager to the task. More...

std::shared_ptr< TaskOutput >	getManagedMemory ()
	Memory manager accessor. More...

bool	canTerminate () override
	Method called to test the task;s termination, by default, the test is: no input nodes connected and, no data waiting to be treated. More...

Public Member Functions inherited from hh::behavior::Node
virtual std::string	extraPrintingInformation () const
	Adds node information to print in the dot file. More...

Public Member Functions inherited from hh::behavior::Execute< TaskInputs >
virtual void	execute (std::shared_ptr< TaskInputs >)=0
	Virtual declaration of execute function for a data of type Input.

Private Attributes
bool	enablePeerAccess_ = false
	Enable CUDA Peer Access through all CUDA devices available.

std::unordered_set< int >	peerDeviceIds_ = {}
	Sparse matrix of linked CUDA devices.

cudaStream_t	stream_ = {}
	CUDA stream linked to the task.

Additional Inherited Members
Public Types inherited from hh::behavior::MultiReceivers< TaskInputs... >
using	inputs_t = std::tuple< Inputs... >
	Tuple with the list of input types.

Public Types inherited from hh::behavior::Sender< TaskOutput >
using	output_t = TaskOutput
	Output Type.

Detailed Description

template<class TaskOutput, class ... TaskInputs>
class hh::AbstractCUDATask< TaskOutput, TaskInputs >

Abstract Task specialized for CUDA computation.

At initialization, the device is set to the task (cudaSetDevice), and a stream is created and bound to the task (cudaStreamCreate). During shutdown, the stream is destroyed (cudaStreamDestroy).

Virtual functions: Execute::execute (one for each of TaskInputs)
AbstractCUDATask::copy (only used if number of threads is greater than 1 or used in an ExecutionPipeline)
AbstractCUDATask::initializeCuda
AbstractCUDATask::shutdownCuda
Node::canTerminate
Node::extraPrintingInformation

Template Parameters

TaskOutput	Task output type
TaskInputs	Task input types

Definition at line 78 of file abstract_cuda_task.h.

Constructor & Destructor Documentation

◆ AbstractCUDATask() [1/3]

template<class TaskOutput , class ... TaskInputs>

hh::AbstractCUDATask< TaskOutput, TaskInputs >::AbstractCUDATask ( size_t numberThreads = 1 )

inlineexplicit

Default constructor for an AbstractCUDATask.

Default constructor with as optional parameter numberThread set to 1. The default name set is "CudaTask", the task won't start automatically, and the peer access is enable.

Parameters

numberThreads Task's number of threads

Definition at line 92 of file abstract_cuda_task.h.

Here is the call graph for this function:

◆ AbstractCUDATask() [2/3]

template<class TaskOutput , class ... TaskInputs>

hh::AbstractCUDATask< TaskOutput, TaskInputs >::AbstractCUDATask	(	std::string_view const &	name,
		size_t	numberThreads = `1`
	)

inlineexplicit

Constructor for an AbstractCUDATask with name as mandatory parameter and the number of threads as optional parameter.

The task won't start automatically, and the peer access is enable.

Parameters

name	Task's name
numberThreads	Task's number of threads

Definition at line 103 of file abstract_cuda_task.h.

Here is the call graph for this function:

◆ AbstractCUDATask() [3/3]

template<class TaskOutput , class ... TaskInputs>

hh::AbstractCUDATask< TaskOutput, TaskInputs >::AbstractCUDATask	(	std::string_view const &	name,
		size_t	numberThreads,
		bool	automaticStart,
		bool	enablePeerAccess
	)

inline

Constructor for an AbstractCUDATask with name, number of threads, automaticStart and enablePeerAccess as mandatory parameter.

Parameters

name	Task's name
numberThreads	Task's number of threads
automaticStart	Task's automatic start
enablePeerAccess	Peer access enabling

Definition at line 115 of file abstract_cuda_task.h.

Here is the call graph for this function:

Member Function Documentation

◆ enablePeerAccess()

template<class TaskOutput , class ... TaskInputs>

bool hh::AbstractCUDATask< TaskOutput, TaskInputs >::enablePeerAccess ( ) const

inline

Accessor for peer access choice.

Returns: True if peer access is enabled, else False

Definition at line 169 of file abstract_cuda_task.h.

◆ hasPeerAccess()

template<class TaskOutput , class ... TaskInputs>

bool hh::AbstractCUDATask< TaskOutput, TaskInputs >::hasPeerAccess ( int peerDeviceId )

inline

Accessor for peer access enabled for a specific device id.

Parameters

peerDeviceId Device id to test

Returns: True if peer access enable for device id peerDeviceId, else False

Definition at line 178 of file abstract_cuda_task.h.

◆ stream()

template<class TaskOutput , class ... TaskInputs>

cudaStream_t hh::AbstractCUDATask< TaskOutput, TaskInputs >::stream ( ) const

inline

Getter for CUDA task's stream.

Returns: CUDA stream

Definition at line 173 of file abstract_cuda_task.h.

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

/home/tjb3/work/hedgehog-nist/hedgehog/api/abstract_cuda_task.h

Public Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

template<class TaskOutput, class ... TaskInputs> class hh::AbstractCUDATask< TaskOutput, TaskInputs >

Constructor & Destructor Documentation

◆ AbstractCUDATask() [1/3]

◆ AbstractCUDATask() [2/3]

◆ AbstractCUDATask() [3/3]

Member Function Documentation

◆ enablePeerAccess()

◆ hasPeerAccess()

◆ stream()

template<class TaskOutput, class ... TaskInputs>
class hh::AbstractCUDATask< TaskOutput, TaskInputs >