Interface for the kernel to run a depthwise convolution native on a tensor. More...

#include <CpuDepthwiseConv2dNativeKernel.h>

Collaboration diagram for CpuDepthwiseConv2dNativeKernel:

Public Member Functions
	CpuDepthwiseConv2dNativeKernel ()=default

	ARM_COMPUTE_DISALLOW_COPY_ALLOW_MOVE (CpuDepthwiseConv2dNativeKernel)

void	configure (const ITensorInfo src, const ITensorInfo weights, const ITensorInfo biases, ITensorInfo dst, const ConvolutionInfo &info)
	Initialize the function's source, destination and parameters. More...

void	run_op (ITensorPack &tensors, const Window &window, const ThreadInfo &info) override
	Execute the kernel on the passed window. More...

const char *	name () const override
	Name of the kernel. More...

Public Member Functions inherited from ICPPKernel
virtual	~ICPPKernel ()=default
	Default destructor. More...

virtual void	run (const Window &window, const ThreadInfo &info)
	Execute the kernel on the passed window. More...

virtual void	run_nd (const Window &window, const ThreadInfo &info, const Window &thread_locator)
	legacy compatibility layer for implemantions which do not support thread_locator In these cases we simply narrow the interface down the legacy version More...

Public Member Functions inherited from IKernel
	IKernel ()
	Constructor. More...

virtual	~IKernel ()=default
	Destructor. More...

virtual bool	is_parallelisable () const
	Indicates whether or not the kernel is parallelisable. More...

virtual BorderSize	border_size () const
	The size of the border for that kernel. More...

const Window &	window () const
	The maximum window the kernel can be executed on. More...

bool	is_window_configured () const
	Function to check if the embedded window of this kernel has been configured. More...

Static Public Member Functions
static Status	validate (const ITensorInfo src, const ITensorInfo weights, const ITensorInfo biases, const ITensorInfo dst, const ConvolutionInfo &info)
	Static function to check if given info will lead to a valid configuration. More...

Detailed Description

Interface for the kernel to run a depthwise convolution native on a tensor.

Definition at line 43 of file CpuDepthwiseConv2dNativeKernel.h.

Constructor & Destructor Documentation

◆ CpuDepthwiseConv2dNativeKernel()

CpuDepthwiseConv2dNativeKernel ( )

default

Member Function Documentation

◆ ARM_COMPUTE_DISALLOW_COPY_ALLOW_MOVE()

ARM_COMPUTE_DISALLOW_COPY_ALLOW_MOVE ( CpuDepthwiseConv2dNativeKernel )

◆ configure()

void configure	(	const ITensorInfo *	src,
		const ITensorInfo *	weights,
		const ITensorInfo *	biases,
		ITensorInfo *	dst,
		const ConvolutionInfo &	info
	)

Initialize the function's source, destination and parameters.

Note: Supported data layouts: NHWC

Parameters

[in]	src	Source tensor. DataType supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
[in]	weights	Weights tensor. This is a 3D tensor with dimensions [IFM, W, H]. Data type supported: Same as `src` or QASYMM8/QASYMM8_SIGNED/QSYMM8_PER_CHANNEL when `src` is QASYMM8/QASYMM8_SIGNED.
[in]	biases	Biases tensor. A 1D tensor with dimensions [IFM]. Must be nullptr if not needed. Data type supported: Same as `src`, S32 when src is QASYMM8/QASYMM8_SIGNED.
[out]	dst	Destination tensor. Data type supported: Same as `src`.
[in]	info	Depthwise convolution meta-data.

Definition at line 806 of file CpuDepthwiseConv2dNativeKernel.cpp.

References ARM_COMPUTE_ERROR, ARM_COMPUTE_ERROR_ON_NULLPTR, ARM_COMPUTE_ERROR_THROW_ON, arm_compute::auto_init_if_empty(), arm_compute::calculate_max_window(), arm_compute::quantization::calculate_quantized_multiplier(), ICloneable< T >::clone(), arm_compute::misc::shape_calculator::compute_depthwise_convolution_shape(), ITensorInfo::data_type(), ConvolutionInfo::depth_multiplier, ConvolutionInfo::dilation, ITensorInfo::dimension(), arm_compute::F16, arm_compute::F32, arm_compute::is_data_type_quantized(), arm_compute::is_data_type_quantized_per_channel(), arm_compute::test::validation::output_shape, ConvolutionInfo::pad_stride_info, arm_compute::QASYMM8, arm_compute::QASYMM8_SIGNED, arm_compute::QSYMM8_PER_CHANNEL, ITensorInfo::quantization_info(), UniformQuantizationInfo::scale, QuantizationInfo::scale(), and QuantizationInfo::uniform().

 {
     ARM_COMPUTE_ERROR_ON_NULLPTR(src, weights, dst);
     ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(src, weights, (biases != nullptr) ? biases : nullptr, dst, info));
 
     _conv_info        = info.pad_stride_info;
     _depth_multiplier = info.depth_multiplier;
     _dilation         = info.dilation;
     _has_biases       = (biases != nullptr);
 
     if(is_data_type_quantized(src->data_type()))
     {
         const auto input_scale  = src->quantization_info().uniform().scale;
         const auto output_scale = dst->quantization_info().uniform().scale;
 
         auto weights_scale = weights->quantization_info().scale();
         if(!is_data_type_quantized_per_channel(weights->data_type()))
         {
             for(size_t i = 1; i < weights->dimension(channel_idx); ++i)
             {
                 weights_scale.push_back(weights_scale.front());
             }
         }
 
         for(const auto &s : weights_scale)
         {
             int32_t     out_mult   = 0;
             int32_t     out_shift  = 0;
             const float multiplier = input_scale * s / output_scale;
             arm_compute::quantization::calculate_quantized_multiplier(multiplier, &out_mult, &out_shift);
 
             _output_multiplier.push_back(out_mult);
             _output_shift.push_back(out_shift);
         }
     }
 
     switch(weights->data_type())
     {
         case DataType::QASYMM8:
             _func = &CpuDepthwiseConv2dNativeKernel::run_depthwise<uint8_t, uint8_t>;
             break;
         case DataType::QASYMM8_SIGNED:
             _func = &CpuDepthwiseConv2dNativeKernel::run_depthwise<int8_t, int8_t>;
             break;
         case DataType::QSYMM8_PER_CHANNEL:
             if(src->data_type() == DataType::QASYMM8)
             {
                 _func = &CpuDepthwiseConv2dNativeKernel::run_depthwise<uint8_t, int8_t>;
             }
             else
             {
                 _func = &CpuDepthwiseConv2dNativeKernel::run_depthwise<int8_t, int8_t>;
             }
             break;
 #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
         case DataType::F16:
             _func = &CpuDepthwiseConv2dNativeKernel::run_depthwise<float16_t, float16_t>;
             break;
 #endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
         case DataType::F32:
             _func = &CpuDepthwiseConv2dNativeKernel::run_depthwise<float, float>;
             break;
         default:
             ARM_COMPUTE_ERROR("Data type not supported");
             break;
     }
 
     const TensorShape output_shape = misc::shape_calculator::compute_depthwise_convolution_shape(*src, *weights, info);
     auto_init_if_empty(*dst, src->clone()->set_is_resizable(true).reset_padding().set_tensor_shape(output_shape).set_quantization_info(dst->quantization_info()));
 
     Window win = calculate_max_window(*dst, Steps());
     ICpuKernel::configure(win);
 }

◆ name()

const char * name ( ) const

overridevirtual

Name of the kernel.

Returns: Kernel name

Implements ICPPKernel.

Definition at line 944 of file CpuDepthwiseConv2dNativeKernel.cpp.

 {
     return "CpuDepthwiseConv2dNativeKernel";
 }

◆ run_op()

void run_op	(	ITensorPack &	tensors,
		const Window &	window,
		const ThreadInfo &	info
	)

overridevirtual

Execute the kernel on the passed window.

Warning: If is_parallelisable() returns false then the passed window must be equal to window()

Note: The window has to be a region within the window returned by the window() method; The width of the window has to be a multiple of num_elems_processed_per_iteration().

Parameters

[in]	tensors	A vector containing the tensors to operate on.
[in]	window	Region on which to execute the kernel. (Must be a region of the window returned by window())
[in]	info	Info about executing thread and CPU.

Reimplemented from ICPPKernel.

Definition at line 930 of file CpuDepthwiseConv2dNativeKernel.cpp.

References arm_compute::ACL_DST, arm_compute::ACL_SRC_0, arm_compute::ACL_SRC_1, arm_compute::ACL_SRC_2, ARM_COMPUTE_ERROR_ON, ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW, ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL, ARM_COMPUTE_UNUSED, ITensorPack::get_const_tensor(), ITensorPack::get_tensor(), and IKernel::window().

 {
     ARM_COMPUTE_UNUSED(info);
     ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
     ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICpuKernel::window(), window);
     ARM_COMPUTE_ERROR_ON(_func == nullptr);
 
     const auto src     = tensors.get_const_tensor(TensorType::ACL_SRC_0);
     const auto weights = tensors.get_const_tensor(TensorType::ACL_SRC_1);
     const auto biases  = tensors.get_const_tensor(TensorType::ACL_SRC_2);
     auto       dst     = tensors.get_tensor(TensorType::ACL_DST);
     (this->*_func)(src, weights, biases, dst, window, _has_biases);
 }

◆ validate()

Status validate	(	const ITensorInfo *	src,
		const ITensorInfo *	weights,
		const ITensorInfo *	biases,
		const ITensorInfo *	dst,
		const ConvolutionInfo &	info
	)

static

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

Returns: a status

Definition at line 880 of file CpuDepthwiseConv2dNativeKernel.cpp.

References ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW, ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL, ARM_COMPUTE_RETURN_ON_ERROR, ITensorInfo::data_type(), arm_compute::test::validation::dst, ITensor::info(), arm_compute::is_data_type_quantized_per_channel(), arm_compute::test::validation::src, and IKernel::window().

 {
     ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(src, weights, biases, dst, info));
     return Status{};
 }

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

src/core/cpu/kernels/CpuDepthwiseConv2dNativeKernel.h
src/core/cpu/kernels/CpuDepthwiseConv2dNativeKernel.cpp

Public Member Functions

Static Public Member Functions