#include <Sequence.hpp>

Inheritance diagram for kp::Sequence:

Collaboration diagram for kp::Sequence:

Public Member Functions
	Sequence (std::shared_ptr< vk::PhysicalDevice > physicalDevice, std::shared_ptr< vk::Device > device, std::shared_ptr< vk::Queue > computeQueue, uint32_t queueIndex, uint32_t totalTimestamps=0)

	~Sequence ()

std::shared_ptr< Sequence >	record (std::shared_ptr< OpBase > op)

template<typename T , typename... TArgs>
std::shared_ptr< Sequence >	record (std::vector< std::shared_ptr< Tensor >> tensors, TArgs &&... params)

template<typename T , typename... TArgs>
std::shared_ptr< Sequence >	record (std::shared_ptr< Algorithm > algorithm, TArgs &&... params)

std::shared_ptr< Sequence >	eval ()

std::shared_ptr< Sequence >	eval (std::shared_ptr< OpBase > op)

template<typename T , typename... TArgs>
std::shared_ptr< Sequence >	eval (std::vector< std::shared_ptr< Tensor >> tensors, TArgs &&... params)

template<typename T , typename... TArgs>
std::shared_ptr< Sequence >	eval (std::shared_ptr< Algorithm > algorithm, TArgs &&... params)

std::shared_ptr< Sequence >	evalAsync ()

std::shared_ptr< Sequence >	evalAsync (std::shared_ptr< OpBase > op)

template<typename T , typename... TArgs>
std::shared_ptr< Sequence >	evalAsync (std::vector< std::shared_ptr< Tensor >> tensors, TArgs &&... params)

template<typename T , typename... TArgs>
std::shared_ptr< Sequence >	evalAsync (std::shared_ptr< Algorithm > algorithm, TArgs &&... params)

std::shared_ptr< Sequence >	evalAwait (uint64_t waitFor=UINT64_MAX)

void	clear ()

std::vector< std::uint64_t >	getTimestamps ()

void	begin ()

void	end ()

bool	isRecording () const

bool	isInit () const

void	rerecord ()

bool	isRunning () const

void	destroy ()

Detailed Description

Container of operations that can be sent to GPU as batch

Constructor & Destructor Documentation

◆ Sequence()

kp::Sequence::Sequence	(	std::shared_ptr< vk::PhysicalDevice >	physicalDevice,
		std::shared_ptr< vk::Device >	device,
		std::shared_ptr< vk::Queue >	computeQueue,
		uint32_t	queueIndex,
		uint32_t	totalTimestamps = `0`
	)

Main constructor for sequence which requires core vulkan components to generate all dependent resources.

Parameters

physicalDevice	Vulkan physical device
device	Vulkan logical device
computeQueue	Vulkan compute queue
queueIndex	Vulkan compute queue index in device
totalTimestamps	Maximum number of timestamps to allocate

◆ ~Sequence()

kp::Sequence::~Sequence ( )

Destructor for sequence which is responsible for cleaning all subsequent owned operations.

Member Function Documentation

◆ begin()

void kp::Sequence::begin ( )

Begins recording commands for commands to be submitted into the command buffer.

◆ clear()

void kp::Sequence::clear ( )

Clear function clears all operations currently recorded and starts recording again.

◆ destroy()

void kp::Sequence::destroy ( )

Destroys and frees the GPU resources which include the buffer and memory and sets the sequence as init=False.

◆ end()

void kp::Sequence::end ( )

Ends the recording and stops recording commands when the record command is sent.

◆ eval() [1/4]

std::shared_ptr<Sequence> kp::Sequence::eval ( )

Eval sends all the recorded and stored operations in the vector of operations into the gpu as a submit job synchronously (with a barrier).

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ eval() [2/4]

template<typename T , typename... TArgs>

std::shared_ptr<Sequence> kp::Sequence::eval	(	std::shared_ptr< Algorithm >	algorithm,
		TArgs &&...	params
	)

inline

Eval sends all the recorded and stored operations in the vector of operations into the gpu as a submit job with a barrier.

Parameters

algorithm	Algorithm to use for the record often used for OpAlgo operations
TArgs	Template parameters that are used to initialise operation which allows for extensible configurations on initialisation.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ eval() [3/4]

std::shared_ptr<Sequence> kp::Sequence::eval ( std::shared_ptr< OpBase > op )

Resets all the recorded and stored operations, records the operation provided and submits into the gpu as a submit job synchronously (with a barrier).

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ eval() [4/4]

template<typename T , typename... TArgs>

std::shared_ptr<Sequence> kp::Sequence::eval	(	std::vector< std::shared_ptr< Tensor >>	tensors,
		TArgs &&...	params
	)

inline

Eval sends all the recorded and stored operations in the vector of operations into the gpu as a submit job with a barrier.

Parameters

tensors	Vector of tensors to use for the operation
TArgs	Template parameters that are used to initialise operation which allows for extensible configurations on initialisation.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ evalAsync() [1/4]

std::shared_ptr<Sequence> kp::Sequence::evalAsync ( )

Eval Async sends all the recorded and stored operations in the vector of operations into the gpu as a submit job without a barrier. EvalAwait() must ALWAYS be called after to ensure the sequence is terminated correctly.

Returns: Boolean stating whether execution was successful.

◆ evalAsync() [2/4]

template<typename T , typename... TArgs>

std::shared_ptr<Sequence> kp::Sequence::evalAsync	(	std::shared_ptr< Algorithm >	algorithm,
		TArgs &&...	params
	)

inline

Eval sends all the recorded and stored operations in the vector of operations into the gpu as a submit job with a barrier.

Parameters

algorithm	Algorithm to use for the record often used for OpAlgo operations
TArgs	Template parameters that are used to initialise operation which allows for extensible configurations on initialisation.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ evalAsync() [3/4]

std::shared_ptr<Sequence> kp::Sequence::evalAsync ( std::shared_ptr< OpBase > op )

Clears currnet operations to record provided one in the vector of operations into the gpu as a submit job without a barrier. EvalAwait() must ALWAYS be called after to ensure the sequence is terminated correctly.

Returns: Boolean stating whether execution was successful.

◆ evalAsync() [4/4]

template<typename T , typename... TArgs>

std::shared_ptr<Sequence> kp::Sequence::evalAsync	(	std::vector< std::shared_ptr< Tensor >>	tensors,
		TArgs &&...	params
	)

inline

Eval sends all the recorded and stored operations in the vector of operations into the gpu as a submit job with a barrier.

Parameters

tensors	Vector of tensors to use for the operation
TArgs	Template parameters that are used to initialise operation which allows for extensible configurations on initialisation.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ evalAwait()

std::shared_ptr<Sequence> kp::Sequence::evalAwait ( uint64_t waitFor = UINT64_MAX )

Eval Await waits for the fence to finish processing and then once it finishes, it runs the postEval of all operations.

Parameters

waitFor Number of milliseconds to wait before timing out.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ getTimestamps()

std::vector<std::uint64_t> kp::Sequence::getTimestamps ( )

Return the timestamps that were latched at the beginning and after each operation during the last eval() call.

◆ isInit()

bool kp::Sequence::isInit ( ) const

Returns true if the sequence has been initialised, and it's based on the GPU resources being referenced.

Returns: Boolean stating if is initialized

◆ isRecording()

bool kp::Sequence::isRecording ( ) const

Returns true if the sequence is currently in recording activated.

Returns: Boolean stating if recording ongoing.

◆ isRunning()

bool kp::Sequence::isRunning ( ) const

Returns true if the sequence is currently running - mostly used for async workloads.

Returns: Boolean stating if currently running.

◆ record() [1/3]

template<typename T , typename... TArgs>

std::shared_ptr<Sequence> kp::Sequence::record	(	std::shared_ptr< Algorithm >	algorithm,
		TArgs &&...	params
	)

inline

Record function for operation to be added to the GPU queue in batch. This template requires classes to be derived from the OpBase class. This function also requires the Sequence to be recording, otherwise it will not be able to add the operation.

Parameters

algorithm	Algorithm to use for the record often used for OpAlgo operations
TArgs	Template parameters that are used to initialise operation which allows for extensible configurations on initialisation.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ record() [2/3]

std::shared_ptr<Sequence> kp::Sequence::record ( std::shared_ptr< OpBase > op )

Record function for operation to be added to the GPU queue in batch. This template requires classes to be derived from the OpBase class. This function also requires the Sequence to be recording, otherwise it will not be able to add the operation.

Parameters

op	Object derived from kp::BaseOp that will be recoreded by the sequence which will be used when the operation is evaluated.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ record() [3/3]

template<typename T , typename... TArgs>

std::shared_ptr<Sequence> kp::Sequence::record	(	std::vector< std::shared_ptr< Tensor >>	tensors,
		TArgs &&...	params
	)

inline

Record function for operation to be added to the GPU queue in batch. This template requires classes to be derived from the OpBase class. This function also requires the Sequence to be recording, otherwise it will not be able to add the operation.

Parameters

tensors	Vector of tensors to use for the operation
TArgs	Template parameters that are used to initialise operation which allows for extensible configurations on initialisation.

Returns: shared_ptr<Sequence> of the Sequence class itself

◆ rerecord()

void kp::Sequence::rerecord ( )

Clears command buffer and triggers re-record of all the current operations saved, which is useful if the underlying kp::Tensors or kp::Algorithms are modified and need to be re-recorded.

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

/home/alejandro/Programming/kubernetes/seldon/vulkan-kompute/src/include/kompute/Sequence.hpp

Public Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ Sequence()

◆ ~Sequence()

Member Function Documentation

◆ begin()

◆ clear()

◆ destroy()

◆ end()

◆ eval() [1/4]

◆ eval() [2/4]

◆ eval() [3/4]

◆ eval() [4/4]

◆ evalAsync() [1/4]

◆ evalAsync() [2/4]

◆ evalAsync() [3/4]

◆ evalAsync() [4/4]

◆ evalAwait()

◆ getTimestamps()

◆ isInit()

◆ isRecording()

◆ isRunning()

◆ record() [1/3]

◆ record() [2/3]

◆ record() [3/3]

◆ rerecord()