CLL2NormalizeLayer Class Reference

Basic function to perform a L2 normalization on a given axis. More...

#include <CLL2NormalizeLayer.h>

Collaboration diagram for CLL2NormalizeLayer:

Public Member Functions
	CLL2NormalizeLayer (std::shared_ptr< IMemoryManager > memory_manager=nullptr)
	Constructor. More...
	~CLL2NormalizeLayer ()
	Default Destructor. More...
	CLL2NormalizeLayer (const CLL2NormalizeLayer &)=delete
	Prevent instances of this class from being copied (As this class contains pointers) More...
	CLL2NormalizeLayer (CLL2NormalizeLayer &&)=default
	Default move constructor. More...
CLL2NormalizeLayer &	operator= (const CLL2NormalizeLayer &)=delete
	Prevent instances of this class from being copied (As this class contains pointers) More...
CLL2NormalizeLayer &	operator= (CLL2NormalizeLayer &&)=default
	Default move assignment operator. More...
void	configure (ICLTensor *input, ICLTensor *output, int axis, float epsilon=1e-12f)
	Set the input and output tensors. More...
void	configure (const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, int axis, float epsilon=1e-12f)
	Set the input and output tensors. More...
void	run () override
	Run the kernels contained in the function. More...
Public Member Functions inherited from IFunction
virtual	~IFunction ()=default
	Destructor. More...
virtual void	prepare ()
	Prepare the function for executing. More...

Static Public Member Functions
static Status	validate (const ITensorInfo *input, const ITensorInfo *output, int axis, float epsilon=1e-12f)
	Static function to check if given info will lead to a valid configuration of CLL2NormalizeLayer. More...

Detailed Description

Basic function to perform a L2 normalization on a given axis.

This function runs the following kernels:

1. CLReductionOperation
2. CLL2NormalizeLayerKernel

Definition at line 50 of file CLL2NormalizeLayer.h.

Constructor & Destructor Documentation

CLL2NormalizeLayer() [1/3]

CLL2NormalizeLayer

(

std::shared_ptr< IMemoryManager >

memory_manager = nullptr

)

Constructor.

Definition at line 45 of file CLL2NormalizeLayer.cpp.

    : _memory_group(std::move(memory_manager)),
      _reduce_func(),
      _normalize_kernel(std::make_unique<CLL2NormalizeLayerKernel>()),
      _sumsq()
{
}

~CLL2NormalizeLayer()

~CLL2NormalizeLayer ( )

default

Default Destructor.

CLL2NormalizeLayer() [2/3]

CLL2NormalizeLayer ( const CLL2NormalizeLayer &  )

delete

Prevent instances of this class from being copied (As this class contains pointers)

CLL2NormalizeLayer() [3/3]

CLL2NormalizeLayer ( CLL2NormalizeLayer &&  )

default

Default move constructor.

Member Function Documentation

configure() [1/2]

void configure	(	const CLCompileContext &	compile_context,
		ICLTensor *	input,
		ICLTensor *	output,
		int	axis,
		float	epsilon = 1e-12f
	)

Set the input and output tensors.

Parameters

[in]	compile_context	The compile context to be used.
[in]	input	Source tensor. Data types supported: F16/F32. Data layouts supported: NCHW/NHWC.
[out]	output	Destination tensor. Data types and data layouts supported: Same as input.
[in]	axis	Axis along which to reduce. Negative values wrap around. Maximum supported actual reduction axis : 2
[in]	epsilon	(Optional) Lower bound value for the normalization.

Definition at line 60 of file CLL2NormalizeLayer.cpp.

{
    ARM_COMPUTE_LOG_PARAMS(input, output, axis, epsilon);
 
    // Reset auxiliary tensor
    _sumsq.allocator()->init(TensorInfo());
 
    // Manage intermediate buffers
    _memory_group.manage(&_sumsq);
 
    // Configure kernels
    const uint32_t actual_axis = wrap_around(axis, max_input_tensor_dim);
    _reduce_func.configure(compile_context, input, &_sumsq, actual_axis, ReductionOperation::SUM_SQUARE);
    _normalize_kernel->configure(compile_context, input, &_sumsq, output, axis, epsilon);
 
    // Allocate intermediate tensor
    _sumsq.allocator()->allocate();
}

References CLTensorAllocator::allocate(), CLTensor::allocator(), ARM_COMPUTE_LOG_PARAMS, CLReductionOperation::configure(), arm_compute::quantization::epsilon, ITensorAllocator::init(), arm_compute::test::validation::input, MemoryGroup::manage(), arm_compute::SUM_SQUARE, and arm_compute::wrap_around().

configure() [2/2]

void configure	(	ICLTensor *	input,
		ICLTensor *	output,
		int	axis,
		float	epsilon = 1e-12f
	)

Set the input and output tensors.

Valid data layouts:

• NHWC
• NCHW

Valid data type configurations:

src	dst
F16	F16
F32	F32

Parameters

[in]	input	Source tensor. Data types supported: F16/F32. Data layouts supported: NCHW/NHWC.
[out]	output	Destination tensor. Data types and data layouts supported: Same as input.
[in]	axis	Axis along which to reduce. Negative values wrap around. Maximum supported actual reduction axis : 2
[in]	epsilon	(Optional) Lower bound value for the normalization.

Definition at line 55 of file CLL2NormalizeLayer.cpp.

{
    configure(CLKernelLibrary::get().get_compile_context(), input, output, axis, epsilon);
}

References arm_compute::quantization::epsilon, CLKernelLibrary::get(), and arm_compute::test::validation::input.

operator=() [1/2]

CLL2NormalizeLayer& operator= ( CLL2NormalizeLayer &&  )

default

Default move assignment operator.

operator=() [2/2]

CLL2NormalizeLayer& operator= ( const CLL2NormalizeLayer &  )

delete

Prevent instances of this class from being copied (As this class contains pointers)

run()

void run ( )

overridevirtual

Run the kernels contained in the function.

For CPU kernels:

• Multi-threading is used for the kernels which are parallelisable.
• By default std::thread::hardware_concurrency() threads are used.

Note

CPPScheduler::set_num_threads() can be used to manually set the number of threads

For OpenCL kernels:

• All the kernels are enqueued on the queue associated with CLScheduler.
• The queue is then flushed.

Note

The function will not block until the kernels are executed. It is the user's responsibility to wait.

Will call prepare() on first run if hasn't been done

Implements IFunction.

Definition at line 100 of file CLL2NormalizeLayer.cpp.

{
    MemoryGroupResourceScope scope_mg(_memory_group);
 
    _reduce_func.run();
    CLScheduler::get().enqueue(*_normalize_kernel, true);
}

References CLScheduler::enqueue(), CLScheduler::get(), and CLReductionOperation::run().

validate()

Status validate ( const ITensorInfo *  input, const ITensorInfo *  output, int  axis, float  epsilon = 1e-12f  )

static

Static function to check if given info will lead to a valid configuration of CLL2NormalizeLayer.

Parameters

[in]	input	Source tensor info. Data types supported: F16/F32. Data layouts supported: NCHW/NHWC.
[in]	output	Destination tensor info. Data types and data layouts supported: Same as input.
[in]	axis	Axis along which to reduce. Negative values wrap around. Maximum supported actual reduction axis : 2
[in]	epsilon	(Optional) Lower bound value for the normalization.

Returns

a status

Definition at line 79 of file CLL2NormalizeLayer.cpp.

{
    TensorShape shape(input->tensor_shape());
 
    // Create intermediate tensor info
    TensorInfo sum_sq;
    sum_sq.set_data_type(input->data_type());
    sum_sq.set_tensor_shape(shape);
 
    const uint32_t actual_axis = wrap_around(axis, max_input_tensor_dim);
    ARM_COMPUTE_RETURN_ON_ERROR(CLReductionOperation::validate(input, &sum_sq, actual_axis, ReductionOperation::SUM_SQUARE));
 
    // Reduce shape on axis
    shape.set(actual_axis, 1);
    sum_sq.set_tensor_shape(shape);
 
    ARM_COMPUTE_RETURN_ON_ERROR(CLL2NormalizeLayerKernel::validate(input, &sum_sq, output, axis, epsilon));
 
    return Status{};
}

References ARM_COMPUTE_RETURN_ON_ERROR, arm_compute::quantization::epsilon, arm_compute::test::validation::input, TensorInfo::set_data_type(), TensorInfo::set_tensor_shape(), arm_compute::test::validation::shape, arm_compute::SUM_SQUARE, CLL2NormalizeLayerKernel::validate(), CLReductionOperation::validate(), and arm_compute::wrap_around().

---

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

• arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h
• src/runtime/CL/functions/CLL2NormalizeLayer.cpp