Compute Library
 19.08
GCDepthwiseConvolutionLayer3x3Kernel Class Reference

Interface for the kernel to run a 3x3 depthwise convolution on a tensor. More...

#include <GCDepthwiseConvolutionLayer3x3Kernel.h>

Collaboration diagram for GCDepthwiseConvolutionLayer3x3Kernel:
[legend]

Public Member Functions

 GCDepthwiseConvolutionLayer3x3Kernel ()
 Default constructor. More...
 
 GCDepthwiseConvolutionLayer3x3Kernel (const GCDepthwiseConvolutionLayer3x3Kernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
GCDepthwiseConvolutionLayer3x3Kerneloperator= (const GCDepthwiseConvolutionLayer3x3Kernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
 GCDepthwiseConvolutionLayer3x3Kernel (GCDepthwiseConvolutionLayer3x3Kernel &&)=default
 Default Move Constructor. More...
 
GCDepthwiseConvolutionLayer3x3Kerneloperator= (GCDepthwiseConvolutionLayer3x3Kernel &&)=default
 Default move assignment operator. More...
 
void configure (const IGCTensor *input, const IGCTensor *weights, const IGCTensor *biases, IGCTensor *output, const PadStrideInfo &conv_info, unsigned int depth_multiplier=1)
 Initialize the function's source, destination, conv and border_size. More...
 
void run (const Window &window) override
 Enqueue the OpenGL ES shader to process the given window. More...
 
BorderSize border_size () const override
 The size of the border for that kernel. More...
 
- Public Member Functions inherited from IGCKernel
 IGCKernel ()
 Constructor. More...
 
GCKernelkernel ()
 Returns a reference to the GLES kernel of this object. More...
 
void add_1D_tensor_argument (unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
 Add the passed 1D tensor's parameters to the object's kernel's arguments starting from the index idx. More...
 
void add_2D_tensor_argument (unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
 Add the passed 2D tensor's parameters to the object's kernel's arguments starting from the index idx. More...
 
void add_3D_tensor_argument (unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
 Add the passed 3D tensor's parameters to the object's kernel's arguments starting from the index idx. More...
 
unsigned int num_arguments_per_1D_tensor () const
 Returns the number of arguments enqueued per 1D tensor object. More...
 
unsigned int num_arguments_per_2D_tensor () const
 Returns the number of arguments enqueued per 2D tensor object. More...
 
unsigned int num_arguments_per_3D_tensor () const
 Returns the number of arguments enqueued per 3D tensor object. More...
 
void set_lws_hint (gles::NDRange &lws_hint)
 Set the Local-Workgroup-Size hint. More...
 
void set_target (GPUTarget target)
 Set the targeted GPU architecture. More...
 
GPUTarget get_target () const
 Get the targeted GPU architecture. 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...
 
const Windowwindow () const
 The maximum window the kernel can be executed on. More...
 

Detailed Description

Interface for the kernel to run a 3x3 depthwise convolution on a tensor.

Definition at line 35 of file GCDepthwiseConvolutionLayer3x3Kernel.h.

Constructor & Destructor Documentation

◆ GCDepthwiseConvolutionLayer3x3Kernel() [1/3]

Default constructor.

Definition at line 41 of file GCDepthwiseConvolutionLayer3x3Kernel.cpp.

42  : _border_size(0), _input(), _output(), _weights(), _biases(), _conv_stride_x(0), _conv_stride_y(0), _conv_pad_left(0), _conv_pad_top(0), _lws(gles::NDRange(1U, 1U, 1U))
43 {
44 }
Class interface for specifying NDRange values.
Definition: OpenGLES.h:53

◆ GCDepthwiseConvolutionLayer3x3Kernel() [2/3]

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

◆ GCDepthwiseConvolutionLayer3x3Kernel() [3/3]

Member Function Documentation

◆ border_size()

BorderSize border_size ( ) const
overridevirtual

The size of the border for that kernel.

Returns
The width in number of elements of the border.

Reimplemented from IKernel.

Definition at line 46 of file GCDepthwiseConvolutionLayer3x3Kernel.cpp.

47 {
48  return _border_size;
49 }

◆ configure()

void configure ( const IGCTensor input,
const IGCTensor weights,
const IGCTensor biases,
IGCTensor output,
const PadStrideInfo conv_info,
unsigned int  depth_multiplier = 1 
)

Initialize the function's source, destination, conv and border_size.

Parameters
[in]inputSource tensor. DataType supported: F16.
[in]weightsWeights tensor. A 3D tensor with dimensions [3, 3, IFM]. Data type supported: Same as input.
[in]biases(Optional) Biases tensor. A 1D tensor with dimensions [IFM]. Must be nullptr if not needed. Data type supported: Same as input.
[out]outputDestination tensor. Data type supported: Same as input.
[in]conv_infoPadding and stride information to use for the convolution.
[in]depth_multiplier(Optional) Multiplier to apply to the input's depth in order to retrieve the output's depth. Defaults to 1.

Definition at line 51 of file GCDepthwiseConvolutionLayer3x3Kernel.cpp.

53 {
56  ARM_COMPUTE_ERROR_ON(weights->info()->dimension(0) != 3 || weights->info()->dimension(1) != 3);
57 
58  if(biases != nullptr)
59  {
61  ARM_COMPUTE_ERROR_ON(biases->info()->dimension(0) != weights->info()->dimension(2));
62  ARM_COMPUTE_ERROR_ON(biases->info()->num_dimensions() > 1);
63  }
64 
65  // Get convolved dimensions
66  const TensorShape output_shape = compute_depthwise_convolution_shape(*input->info(), *weights->info(), conv_info, depth_multiplier);
67 
68  // Output auto inizialitation if not yet initialized
69  auto_init_if_empty(*output->info(),
71  1,
72  input->info()->data_type());
73 
75  ARM_COMPUTE_ERROR_ON(output->info()->dimension(2) != weights->info()->dimension(2));
76 
77  _input = input;
78  _output = output;
79  _weights = weights;
80  _biases = biases;
81  _conv_stride_x = conv_info.stride().first;
82  _conv_stride_y = conv_info.stride().second;
83  _conv_pad_left = conv_info.pad_left();
84  _conv_pad_top = conv_info.pad_top();
85  _border_size = BorderSize(_conv_pad_top, conv_info.pad_right(), conv_info.pad_bottom(), _conv_pad_left);
86 
87  // Set build options
88  ARM_COMPUTE_ERROR_ON(_conv_stride_x < 1 || _conv_stride_x > 3);
89  std::set<std::string> options;
90 
91  options.emplace("#define DEPTH_MULTIPLIER " + support::cpp11::to_string(depth_multiplier));
92  options.emplace("#define LOCAL_SIZE_X " + support::cpp11::to_string(_lws[0]));
93  options.emplace("#define LOCAL_SIZE_Y " + support::cpp11::to_string(_lws[1]));
94  options.emplace("#define LOCAL_SIZE_Z " + support::cpp11::to_string(_lws[2]));
95  options.emplace("#define STRIDE_X " + support::cpp11::to_string(_conv_stride_x));
96  options.emplace("#define STRIDE_Y " + support::cpp11::to_string(_conv_stride_y));
97 
98  std::string dt_name = (input->info()->data_type() == DataType::F32) ? "DATA_TYPE_FP32" : "DATA_TYPE_FP16";
99  options.emplace(("#define " + dt_name));
100 
101  unsigned int num_elems_read_per_iteration_x = 8;
102  unsigned int num_elems_read_per_iteration_y = 1;
103  unsigned int num_elems_written_per_iteration_x = 4;
104  unsigned int num_elems_written_per_iteration_y = 1;
105  unsigned int num_elems_written_per_iteration_z = 1;
106 
107  if((_conv_stride_x == 1) && (_conv_stride_y == 1))
108  {
109  switch(input->info()->data_type())
110  {
111 #define PROCESS_4X_3Y_1Z
112 
113  case DataType::F16:
114 #if defined(PROCESS_4X_3Y_1Z)
115  options.emplace("#define PROCESS_4X_3Y_1Z");
116  num_elems_read_per_iteration_y = 5;
117  num_elems_written_per_iteration_y = 3;
118 #endif /* PROCESS_4X_3Y_1Z */
119 #undef PROCESS_4X_3Y_1Z
120  break;
121 
122  default:
123  ARM_COMPUTE_ERROR("Current data type is not supported");
124  break;
125  }
126  }
127  else
128  {
129  switch(input->info()->data_type())
130  {
131  case DataType::F16:
132  options.emplace("#define PROCESS_4X_1Y_1Z");
133  break;
134 
135  default:
136  ARM_COMPUTE_ERROR("Current data type is not supported");
137  break;
138  }
139  }
140 
141  if(_biases != nullptr)
142  {
143  options.emplace("#define BIAS");
144  }
145 
146  // Create kernel
147  std::string kernel_name = "depthwise_convolution_3x3";
148  _kernel = static_cast<GCKernel>(GCKernelLibrary::get().create_kernel(kernel_name, options));
149 
150  // Calculate output right and bottom border
151  const int output_width = output->info()->dimension(0);
152  const int output_height = output->info()->dimension(1);
153  const int output_padding_right = ceil_to_multiple(output_width, num_elems_written_per_iteration_x * _lws[0]) - output_width;
154  const int output_padding_bottom = ceil_to_multiple(output_height, num_elems_written_per_iteration_y * _lws[1]) - output_height;
155 
156  // Calculate input right and bottom border
157  const int input_width = input->info()->dimension(0);
158  const int input_height = input->info()->dimension(1);
159 
160  const int input_total_width = std::max(int(input->info()->padding().left), int(_conv_pad_left)) + input_width + std::max(int(input->info()->padding().right), int(_conv_pad_left));
161  const int input_total_height = std::max(int(input->info()->padding().top), int(_conv_pad_top)) + input_height + std::max(int(input->info()->padding().bottom), int(_conv_pad_top));
162 
163  const int input_padding_right = ceil_to_multiple(input_total_width, num_elems_read_per_iteration_x * _lws[0]) - input_width - _conv_pad_left;
164  const int input_padding_bottom = ceil_to_multiple(input_total_height, num_elems_read_per_iteration_y * _lws[1]) - input_height - _conv_pad_top;
165 
166  BorderSize border = BorderSize(0, output_padding_right, output_padding_bottom, 0);
167 
168  Window win = calculate_max_enlarged_window(*output->info(), Steps(num_elems_written_per_iteration_x, num_elems_written_per_iteration_y, num_elems_written_per_iteration_z), border);
169 
170  AccessWindowStatic input_access(input->info(), -_conv_pad_left, -_conv_pad_top, input_width + input_padding_right, input_height + input_padding_bottom);
171  AccessWindowStatic weights_access = AccessWindowStatic(nullptr, 0, 0, 0, 0);
172  AccessWindowStatic bias_access = AccessWindowStatic(nullptr, 0, 0, 0, 1);
173 
174  switch(weights->info()->data_type())
175  {
176  case DataType::F16:
177  weights_access = AccessWindowStatic(weights->info(), 0, 0, 4, 3);
178  if(_biases != nullptr)
179  {
180  bias_access = AccessWindowStatic(_biases->info(), 0, 0, _biases->info()->dimension(0) + 1, 1);
181  }
182  break;
183 
184  default:
185  ARM_COMPUTE_ERROR("Current data type is not supported");
186  break;
187  }
188 
189  AccessWindowStatic output_access(output->info(), 0, 0, output_width + output_padding_right, output_height + output_padding_bottom);
190 
191  if(_biases != nullptr)
192  {
193  update_window_and_padding(win, input_access, weights_access, bias_access, output_access);
194  }
195  else
196  {
197  update_window_and_padding(win, input_access, weights_access, output_access);
198  }
199 
200  output_access.set_valid_region(win, ValidRegion(Coordinates(), output->info()->tensor_shape()));
201 
202  IGCKernel::configure(win);
203 }
#define ARM_COMPUTE_ERROR(...)
Print the given message then throw an std::runtime_error.
Definition: Error.h:261
unsigned int top
top of the border
Definition: Types.h:339
virtual size_t num_dimensions() const =0
The number of dimensions of the tensor (rank)
Shape of a tensor.
Definition: TensorShape.h:39
TensorInfo * info() const override
Interface to be implemented by the child class to return the tensor's metadata.
Definition: CLTensor.cpp:35
TensorShape compute_depthwise_convolution_shape(const ITensorInfo &input, const ITensorInfo &weights, PadStrideInfo conv_info, unsigned int depth_multiplier, const Size2D &dilation=Size2D(1U, 1U))
Calculate the depthwise convolution output shape of a tensor.
virtual size_t dimension(size_t index) const =0
Return the size of the requested dimension.
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Definition: Validate.h:543
Container for 2D border size.
Definition: Types.h:259
std::string to_string(T &&value)
Convert integer and float values to string.
size_t dimension(size_t index) const override
Return the size of the requested dimension.
Definition: TensorInfo.h:223
virtual DataType data_type() const =0
Data type used for each element of the tensor.
1 channel, 1 F32 per channel
#define ARM_COMPUTE_ERROR_ON(cond)
If the condition is true then an error message is printed and an exception thrown.
Definition: Error.h:337
unsigned int bottom
bottom of the border
Definition: Types.h:341
bool auto_init_if_empty(ITensorInfo &info, const TensorShape &shape, int num_channels, DataType data_type, QuantizationInfo quantization_info=QuantizationInfo())
Auto initialize the tensor info (shape, number of channels and data type) if the current assignment i...
Definition: Helpers.inl:201
1 channel, 1 F16 per channel
Implementation of a static rectangular access pattern.
DataType data_type() const override
Data type used for each element of the tensor.
Definition: TensorInfo.h:256
Window calculate_max_enlarged_window(const ValidRegion &valid_region, const Steps &steps=Steps(), BorderSize border_size=BorderSize())
Calculate the maximum window for a given tensor shape and border setting.
Definition: Helpers.cpp:82
bool update_window_and_padding(Window &win, Ts &&... patterns)
Update window and padding size for each of the access patterns.
Definition: Helpers.h:402
virtual const TensorShape & tensor_shape() const =0
Size for each dimension of the tensor.
auto ceil_to_multiple(S value, T divisor) -> decltype(((value+divisor - 1)/divisor) *divisor)
Computes the smallest number larger or equal to value that is a multiple of divisor.
Definition: Utils.h:66
Class to describe a number of elements in each dimension.
Definition: Steps.h:40
Coordinates of an item.
Definition: Coordinates.h:37
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor's metadata.
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_DIMENSIONS(...)
Definition: Validate.h:286
virtual PaddingSize padding() const =0
Padding of tensor.
std::unique_ptr< Kernel > create_kernel()
Helper function to create and return a unique_ptr pointed to a CL/GLES kernel object.
Definition: Helpers.h:86
unsigned int left
left of the border
Definition: Types.h:342
unsigned int right
right of the border
Definition: Types.h:340
#define ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:789
static GCKernelLibrary & get()
Get the static instance of GCKernelLibrary.
Container for valid region of a window.
Definition: Types.h:174
Describe a multidimensional execution window.
Definition: Window.h:39

References ARM_COMPUTE_ERROR, ARM_COMPUTE_ERROR_ON, ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN, ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES, ARM_COMPUTE_ERROR_ON_MISMATCHING_DIMENSIONS, arm_compute::auto_init_if_empty(), BorderSize::bottom, arm_compute::calculate_max_enlarged_window(), arm_compute::ceil_to_multiple(), arm_compute::misc::shape_calculator::compute_depthwise_convolution_shape(), arm_compute::test::validation::conv_info, arm_compute::create_kernel(), ITensorInfo::data_type(), TensorInfo::data_type(), ITensorInfo::dimension(), TensorInfo::dimension(), arm_compute::F16, arm_compute::F32, GCKernelLibrary::get(), ITensor::info(), CLTensor::info(), BorderSize::left, ITensorInfo::num_dimensions(), arm_compute::test::validation::output_shape, ITensorInfo::padding(), BorderSize::right, ITensorInfo::tensor_shape(), arm_compute::support::cpp11::to_string(), BorderSize::top, arm_compute::update_window_and_padding(), and arm_compute::test::validation::weights.

◆ operator=() [1/2]

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

◆ operator=() [2/2]

Default move assignment operator.

◆ run()

void run ( const Window window)
overridevirtual

Enqueue the OpenGL ES shader to process the given window.

Parameters
[in]windowRegion on which to execute the kernel. (Must be a valid region of the window returned by window()).

Implements IGCKernel.

Definition at line 205 of file GCDepthwiseConvolutionLayer3x3Kernel.cpp.

206 {
209 
210  _kernel.use();
211 
212  _output->set_needs_shifting(true);
213 
214  // Create input window and adjust
215  Window win_in = window;
216  win_in.adjust(Window::DimX, -_conv_pad_left, true);
217  win_in.adjust(Window::DimY, -_conv_pad_top, true);
218  win_in.set_dimension_step(Window::DimX, window.x().step() * _conv_stride_x);
219  win_in.set_dimension_step(Window::DimY, window.y().step() * _conv_stride_y);
220 
221  Window slice_in = win_in.first_slice_window_3D();
222  Window slice_out = window.first_slice_window_3D();
223  Window slice_weights = window.first_slice_window_3D();
224  slice_weights.set_dimension_step(Window::DimX, 0);
225  slice_weights.set_dimension_step(Window::DimY, 0);
226 
227  // Set biases
228  if(_biases != nullptr)
229  {
230  unsigned int idx = 3 * num_arguments_per_3D_tensor();
231  Window slice_biases;
232  slice_biases.use_tensor_dimensions(_biases->info()->tensor_shape());
233  add_1D_tensor_argument(idx, _biases, 4, slice_biases);
234  }
235 
236  slice_out.shift(Window::DimX, -(_output->info()->padding()).left);
237 
238  do
239  {
240  unsigned int idx = 0;
241  add_3D_tensor_argument(idx, _input, 1, slice_in);
242  add_3D_tensor_argument(idx, _output, 2, slice_out);
243  add_3D_tensor_argument(idx, _weights, 3, slice_weights);
244 
245  _kernel.update_shader_params();
246  enqueue(*this, slice_out, _lws);
247  }
248  while(window.slide_window_slice_3D(slice_out) && win_in.slide_window_slice_3D(slice_in));
249 }
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
void add_3D_tensor_argument(unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
Add the passed 3D tensor's parameters to the object's kernel's arguments starting from the index idx.
Definition: IGCKernel.cpp:132
void shift(size_t dimension, int shift_value)
Shift the values of a given dimension by the given shift_value.
Definition: Window.inl:119
void enqueue(cl::CommandQueue &queue, ICLKernel &kernel, const Window &window, const cl::NDRange &lws_hint=CLKernelLibrary::get().default_ndrange(), bool use_dummy_work_items=false)
Add the kernel to the command queue with the given window.
Definition: ICLKernel.cpp:39
constexpr int step() const
Return the step of the dimension.
Definition: Window.h:102
unsigned int num_arguments_per_3D_tensor() const
Returns the number of arguments enqueued per 3D tensor object.
Definition: IGCKernel.cpp:147
void use_tensor_dimensions(const TensorShape &shape, size_t first_dimension=Window::DimX)
Use the tensor's dimensions to fill the window dimensions.
Definition: Window.inl:250
static constexpr size_t DimX
Alias for dimension 0 also known as X dimension.
Definition: Window.h:43
virtual const TensorShape & tensor_shape() const =0
Size for each dimension of the tensor.
void set_needs_shifting(bool needs_shifting)
Set the flag indicating whether or not a tensor needs shifting.
Definition: IGCTensor.cpp:61
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor's metadata.
virtual PaddingSize padding() const =0
Padding of tensor.
bool slide_window_slice_3D(Window &slice) const
Slide the passed 3D window slice.
Definition: Window.h:319
static constexpr size_t DimY
Alias for dimension 1 also known as Y dimension.
Definition: Window.h:45
void set_dimension_step(size_t dimension, int step)
Set the step of a given dimension.
Definition: Window.inl:153
constexpr const Dimension & y() const
Alias to access the second dimension of the window.
Definition: Window.h:152
void adjust(size_t dimension, int adjust_value, bool is_at_start)
Adjust the start or end of a given dimension by the given value.
Definition: Window.inl:126
Window first_slice_window_3D() const
First 3D slice of the window.
Definition: Window.h:275
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:205
Describe a multidimensional execution window.
Definition: Window.h:39
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:940
void add_1D_tensor_argument(unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
Add the passed 1D tensor's parameters to the object's kernel's arguments starting from the index idx.
Definition: IGCKernel.cpp:122
constexpr const Dimension & x() const
Alias to access the first dimension of the window.
Definition: Window.h:143

References IGCKernel::add_1D_tensor_argument(), IGCKernel::add_3D_tensor_argument(), Window::adjust(), ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW, ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL, Window::DimX, Window::DimY, arm_compute::enqueue(), Window::first_slice_window_3D(), ITensor::info(), IGCKernel::num_arguments_per_3D_tensor(), ITensorInfo::padding(), Window::set_dimension_step(), IGCTensor::set_needs_shifting(), Window::shift(), Window::slide_window_slice_3D(), Window::Dimension::step(), ITensorInfo::tensor_shape(), Window::use_tensor_dimensions(), IKernel::window(), Window::x(), and Window::y().


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