Operation computed on CPU. More...

#include <operation.h>

Inheritance diagram for Beatmup::NNets::CpuOperation:

Public Member Functions
bool	usesGpu () const
	Returns `true` if the operation is run on GPU. More...

Public Member Functions inherited from Beatmup::NNets::AbstractOperation
virtual	~AbstractOperation ()

virtual int	getInputCount () const
	Returns number of operation inputs. More...

virtual int	getOutputCount () const
	Returns number of operation outputs. More...

virtual bool	acceptsStorageInput (int index=0) const
	Returns `true` if the operation can take a Storage::View at a specific input. More...

virtual bool	acceptsVectorInput (int index=0) const
	Returns `true` if the operation can take a GL::Vector at a specific input. More...

virtual bool	acceptsTextureInput (int index=0) const
	Returns `true` if the operation can take a GL::TextureHandler at a specific input. More...

virtual bool	acceptsStorageOutput (int index=0) const
	Returns `true` if the operation can take a Storage::View at a specific output. More...

virtual bool	acceptsVectorOutput (int index=0) const
	Returns `true` if the operation can take a GL::Vector at a specific output. More...

virtual bool	acceptsTextureOutput (int index=0) const
	Returns `true` if the operation can take a GL::TextureHandler at a specific output. More...

virtual Size	getOutputSize (int outputIndex=0) const
	Returns full size of a specific operation output. More...

virtual Storage::View	getOutput (int index=0)
	Returns a storage view bound to a specific operation output. More...

virtual void	getOutput (GL::Vector *&vector, int index=0)
	Returns a GL::Vector bound to a specific operation output. More...

virtual void	getOutput (GL::TextureHandler *&vector, int index=0)
	Returns a GL::TextureHandler bound to a specific operation output. More...

virtual void	setInput (Storage::View &&storage, int index=0)

virtual void	setOutput (Storage::View &&storage, int index=0)

virtual void	setInput (GL::Vector &vector, int index=0)

virtual void	setOutput (GL::Vector &vector, int index=0)

virtual void	setInput (GL::TextureHandler &image, int index=0)

virtual void	setOutput (GL::TextureHandler &image, int index=0)

virtual std::map< std::string, std::string >	serialize () const =0
	Returns a serialized representation of th operation;. More...

virtual void	disconnect ()=0
	Assigns empty inputs and outputs. More...

virtual unsigned long	countMultiplyAdds () const
	Counts (approximate) number of multiply-adds used by this operation. More...

virtual unsigned long	countTexelFetches () const
	Counts (approximate) number of texels fetches. More...

std::string	getName () const

Protected Member Functions
	CpuOperation (const std::string &name)

void	prepare (GraphicPipeline &gpu, ChunkCollection &data, GL::ProgramBank &bank)
	Compiles GLSL shaders. More...

void	getSampledChannels (int index, int &min, int &max) const
	Retrieves range of input features channels sampled at the same time for a specific input. More...

virtual int	getAmountOfWork () const =0
	Returns amount of work in arbitrary units to be splitted among threads. More...

virtual void	beforeExecute (GraphicPipeline &gpu, const int threadCount)
	Called right before the operation is executed. More...

virtual void	afterExecute (const int threadCount)
	Called right after the operation is executed. More...

void	execute (TaskThread &thread, GraphicPipeline &gpu)
	Executes the operation. More...

void	execute (TaskThread &thread)
	Executes the operation within a specific CPU thread. More...

virtual void	execute (const int sliceStart, const int sliceStop, const int threadIdx, const int threadCount)=0
	Executes the operation body within a specific CPU thread. More...

Protected Member Functions inherited from Beatmup::NNets::AbstractOperation
	AbstractOperation (const std::string &name)

virtual int	getInputPadding (int index=0) const
	Retrieves minimum required size of zero padding for a given input. More...

Detailed Description

Operation computed on CPU.

Definition at line 434 of file operation.h.

Constructor & Destructor Documentation

◆ CpuOperation()

Beatmup::NNets::CpuOperation::CpuOperation ( const std::string & name )

inlineprotected

Definition at line 436 of file operation.h.

436 : AbstractOperation(name) {}

Beatmup::NNets::AbstractOperation::AbstractOperation

AbstractOperation(const AbstractOperation &)=delete

disabling copying constructor

Beatmup::NNets::AbstractOperation::name

std::string name

Definition: operation.h:50

Member Function Documentation

◆ prepare()

void Beatmup::NNets::CpuOperation::prepare	(	GraphicPipeline &	gpu,
		ChunkCollection &	data,
		GL::ProgramBank &	bank
	)

inlineprotectedvirtual

Compiles GLSL shaders.

Parameters

[in,out]	gpu	A graphic pipeline instance
[in,out]	data	Chunkfile containing operation data (e.g. weights and biases)
[in,out]	bank	A program bank with existing GLSL programs to be reused when possible. If a new program is built, it is added to the bank.

Implements Beatmup::NNets::AbstractOperation.

Definition at line 438 of file operation.h.

438 {}

◆ getSampledChannels()

void Beatmup::NNets::CpuOperation::getSampledChannels	(	int	index,
		int &	min,
		int &	max
	)		const

inlineprotectedvirtual

Retrieves range of input features channels sampled at the same time for a specific input.

The operation would typically take the entire storage and sample it at once, if needed. If the number of textures in a storage exceeds the number of texture samplers that the GPU may use simultaneously, an exception occurs. This function provides the necessary information to limit the number of textures in the storage when allocating it. When the limit is reached, multiple channels are packed into a single texture in the storage.

Parameters

[in]	index	The input index. Expected to fall in the valid range, i.e. from zero to getInputCount() - 1 inclusive.
[out]	min	The minimum number of channels that can be sampled at once
[out]	max	The maximum number of channels that can be sampled at once

Implements Beatmup::NNets::AbstractOperation.

Definition at line 440 of file operation.h.

440 { min = max = 0; }

std::min

CustomPoint< numeric > min(const CustomPoint< numeric > &a, const CustomPoint< numeric > &b)

Definition: geometry.h:724

std::max

CustomPoint< numeric > max(const CustomPoint< numeric > &a, const CustomPoint< numeric > &b)

Definition: geometry.h:728

◆ getAmountOfWork()

virtual int Beatmup::NNets::CpuOperation::getAmountOfWork ( ) const

protectedpure virtual

Returns amount of work in arbitrary units to be splitted among threads.

Implemented in Beatmup::NNets::Softmax.

◆ beforeExecute()

virtual void Beatmup::NNets::CpuOperation::beforeExecute	(	GraphicPipeline &	gpu,
		const int	threadCount
	)

inlineprotectedvirtual

Called right before the operation is executed.

Reimplemented in Beatmup::NNets::Softmax.

Definition at line 450 of file operation.h.

450 {}

◆ afterExecute()

virtual void Beatmup::NNets::CpuOperation::afterExecute ( const int threadCount )

inlineprotectedvirtual

Called right after the operation is executed.

Reimplemented in Beatmup::NNets::Softmax.

Definition at line 455 of file operation.h.

455 {}

◆ execute() [1/3]

void CpuOperation::execute	(	TaskThread &	thread,
		GraphicPipeline &	gpu
	)

protectedvirtual

Executes the operation.

The operation should be prepared.

Parameters

[in,out]	thread	Calling CPU thread descriptor
[in,out]	gpu	A graphic pipeline instance

Implements Beatmup::NNets::AbstractOperation.

Definition at line 298 of file operation.cpp.

                                                                    {
     const int num = thread.numThreads();
     beforeExecute(gpu, num);
     const int amount = getAmountOfWork();
     thread.synchronize();
  
     // process the first hunk
     execute(0, amount / num, 0, num);
  
     thread.synchronize();
     afterExecute(num);
 }

◆ execute() [2/3]

void CpuOperation::execute ( TaskThread & thread )

protectedvirtual

Executes the operation within a specific CPU thread.

Parameters

[in,out] thread Calling CPU thread descriptor

Reimplemented from Beatmup::NNets::AbstractOperation.

Definition at line 312 of file operation.cpp.

                                              {
     thread.synchronize();
     if (thread.isTaskAborted())
         return;
     const int
         amount = getAmountOfWork(),
         idx = thread.currentThread(),
         num = thread.numThreads();
     execute(
          idx      * amount / num,
         (idx + 1) * amount / num,
         idx, num
     );
     thread.synchronize();
 }

◆ execute() [3/3]

virtual void Beatmup::NNets::CpuOperation::execute	(	const int	sliceStart,
		const int	sliceStop,
		const int	threadIdx,
		const int	threadCount
	)

protectedpure virtual

Executes the operation body within a specific CPU thread.

The threads can process different slices according to a given amount of work (see getAmountOfWork()).

Parameters

[in]	sliceStart	Current slice starting point (included in the slice)
[in]	sliceStop	Current slice end point (excluded from the slice)
[in]	threadIdx	Zero-based calling thread number
[in]	threadCount	Total number of threads executing the operation

Implemented in Beatmup::NNets::Softmax.

◆ usesGpu()

bool Beatmup::NNets::CpuOperation::usesGpu ( ) const

inlinevirtual

Returns true if the operation is run on GPU.

Otherwise, it is run on CPU.

Reimplemented from Beatmup::NNets::AbstractOperation.

Definition at line 470 of file operation.h.

470 { return false; }

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

core/nnets/operation.h
core/nnets/operation.cpp

Public Member Functions

Protected Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ CpuOperation()

Member Function Documentation

◆ prepare()

◆ getSampledChannels()

◆ getAmountOfWork()

◆ beforeExecute()

◆ afterExecute()

◆ execute() [1/3]

◆ execute() [2/3]

◆ execute() [3/3]

◆ usesGpu()