Compute Library
 21.02
GCScaleKernel Class Reference

Interface for the scale kernel. More...

#include <GCScaleKernel.h>

Collaboration diagram for GCScaleKernel:
[legend]

Public Member Functions

void configure (const IGCTensor *input, IGCTensor *output, const ScaleKernelInfo &info)
 Initialise the kernel's inputs, output and interpolation policy. 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 IGCSimpleKernel
 IGCSimpleKernel ()
 Constructor. More...
 
 IGCSimpleKernel (const IGCSimpleKernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
IGCSimpleKerneloperator= (const IGCSimpleKernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
 IGCSimpleKernel (IGCSimpleKernel &&)=default
 Allow instances of this class to be moved. More...
 
IGCSimpleKerneloperator= (IGCSimpleKernel &&)=default
 Allow instances of this class to be moved. More...
 
 ~IGCSimpleKernel ()=default
 Default destructor. More...
 
void configure (const IGCTensor *input, IGCTensor *output, unsigned int num_elems_processed_per_iteration, bool border_undefined=false, const BorderSize &border_size=BorderSize())
 Configure the 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 scale kernel.

Definition at line 35 of file GCScaleKernel.h.

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 45 of file GCScaleKernel.cpp.

Referenced by GCScaleKernel::configure().

46 {
47  return BorderSize(1);
48 }
Container for 2D border size.
Definition: Types.h:273

◆ configure()

void configure ( const IGCTensor input,
IGCTensor output,
const ScaleKernelInfo info 
)

Initialise the kernel's inputs, output and interpolation policy.

Parameters
[in]inputSource tensor. Data types supported: F16
[out]outputDestination tensor. Data types supported: Same as input All but the lowest two dimensions must be the same size as in the input tensor, i.e. scaling is only performed within the XY-plane.
[in]infoScaleKernelInfo descriptor to be used to configure

Definition at line 50 of file GCScaleKernel.cpp.

References arm_compute::AREA, ARM_COMPUTE_ERROR_ON, ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN, ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES, ARM_COMPUTE_ERROR_ON_NULLPTR, ScaleKernelInfo::border_mode, GCScaleKernel::border_size(), BorderSize::bottom, arm_compute::calculate_max_window(), arm_compute::calculate_valid_region_scale(), arm_compute::ceil_to_multiple(), GCKernelLibrary::create_kernel(), ITensorInfo::dimension(), arm_compute::F16, GCKernelLibrary::get(), ITensor::info(), arm_compute::test::validation::input, ScaleKernelInfo::interpolation_policy, kernel_name, BorderSize::left, arm_compute::NEAREST_NEIGHBOR, IGCKernel::num_arguments_per_3D_tensor(), num_elems_processed_per_iteration, BorderSize::right, ScaleKernelInfo::sampling_policy, arm_compute::string_from_interpolation_policy(), ITensorInfo::tensor_shape(), arm_compute::support::cpp11::to_string(), arm_compute::utility::tolower(), BorderSize::top, arm_compute::TOP_LEFT, arm_compute::UNDEFINED, arm_compute::update_window_and_padding(), and ITensorInfo::valid_region().

51 {
55  ARM_COMPUTE_ERROR_ON(output == input);
57 
58  _input = input;
59  _output = output;
60 
61  // Compute the ratio between source width/height and destination width/height
62  const auto wr = static_cast<float>(input->info()->dimension(0)) / static_cast<float>(output->info()->dimension(0));
63  const auto hr = static_cast<float>(input->info()->dimension(1)) / static_cast<float>(output->info()->dimension(1));
64 
65  // Compute actual border size
66  const bool border_undefined = info.border_mode == BorderMode::UNDEFINED;
67  BorderSize border = border_undefined ? BorderSize(0) : border_size();
68 
69  // Area interpolation behaves as Nearest Neighbour in case of up-sampling
70  auto interpolation_policy_to_use = info.interpolation_policy;
71  if(interpolation_policy_to_use == InterpolationPolicy::AREA && wr <= 1.f && hr <= 1.f)
72  {
73  interpolation_policy_to_use = InterpolationPolicy::NEAREST_NEIGHBOR;
74  }
75  else
76  {
77  ARM_COMPUTE_ERROR_ON(interpolation_policy_to_use == InterpolationPolicy::AREA);
78  }
79 
80  // Create kernel
81  std::set<std::string> build_opts;
82  build_opts.emplace("#define LOCAL_SIZE_X " + support::cpp11::to_string(1));
83  build_opts.emplace("#define LOCAL_SIZE_Y " + support::cpp11::to_string(1));
84  build_opts.emplace("#define LOCAL_SIZE_Z " + support::cpp11::to_string(1));
85 
86  build_opts.emplace("#define DATA_TYPE_FP16");
87  build_opts.emplace("#define BORDER_SIZE " + support::cpp11::to_string(border.right));
89  {
90  build_opts.emplace("#define SAMPLING_POLICY_TOP_LEFT");
91  }
92  else
93  {
94  build_opts.emplace("#define SAMPLING_POLICY_CENTER");
95  }
96 
97  // Configure kernel window
98  unsigned int num_elems_processed_per_iteration = 4;
99  unsigned int input_width_alignment = 2;
100 
101  // performance optimization for 2x upscaling with no border
102  if((fabs(wr - 0.5) < 1e-6) && (fabs(hr - 0.5) < 1e-6) && border_undefined)
103  {
104  num_elems_processed_per_iteration = 8;
105  input_width_alignment = 4;
106  build_opts.emplace("#define SCALE_NEAREST_8X");
107  }
108  else
109  {
110  build_opts.emplace("#define SCALE_NEAREST_GENERIC");
111  }
112 
113  std::string interpolation_name = string_from_interpolation_policy(interpolation_policy_to_use); // NOLINT
114  std::transform(interpolation_name.begin(), interpolation_name.end(), interpolation_name.begin(), ::tolower);
115  std::string kernel_name = "scale_" + interpolation_name;
116  _kernel = GCKernelLibrary::get().create_kernel(kernel_name, build_opts);
117 
118  Window win = calculate_max_window(*output->info(), Steps(num_elems_processed_per_iteration));
119 
120  const ValidRegion &input_valid_region = input->info()->valid_region();
121 
122  const int total_width = border.left + input_valid_region.anchor[0] + input_valid_region.shape[0] + border.right;
123  const int padding_right = ceil_to_multiple(total_width, input_width_alignment) - border.left - input_valid_region.anchor[0] - input_valid_region.shape[0];
124 
125  // Reads can occur within the valid region of the input
126  AccessWindowStatic input_access(input->info(),
127  input_valid_region.anchor[0] - border.left, input_valid_region.anchor[1] - border.top,
128  input_valid_region.anchor[0] + input_valid_region.shape[0] + padding_right,
129  input_valid_region.anchor[1] + input_valid_region.shape[1] + border.bottom);
130 
132 
133  update_window_and_padding(win, input_access, output_access);
134 
135  output_access.set_valid_region(win, calculate_valid_region_scale(*(input->info()),
136  output->info()->tensor_shape(),
138  info.sampling_policy,
139  border_undefined));
140 
141  IGCKernel::configure(win);
142 
143  unsigned int idx = 2 * num_arguments_per_3D_tensor(); //Skip the tensor parameters
144  _kernel.set_argument<float>(idx++, static_cast<float>(input->info()->dimension(0)));
145  _kernel.set_argument<float>(idx++, static_cast<float>(input->info()->dimension(1)));
146  _kernel.set_argument<float>(idx++, wr);
147  _kernel.set_argument<float>(idx++, hr);
148 }
BorderMode border_mode
Border mode policy.
unsigned int top
top of the border
Definition: Types.h:375
Window calculate_max_window(const ValidRegion &valid_region, const Steps &steps, bool skip_border, BorderSize border_size)
BorderSize border_size() const override
The size of the border for that kernel.
InterpolationPolicy interpolation_policy
Interpolation type to use.
virtual size_t dimension(size_t index) const =0
Return the size of the requested dimension.
Container for 2D border size.
Definition: Types.h:273
std::string to_string(T &&value)
Convert integer and float values to string.
#define ARM_COMPUTE_ERROR_ON(cond)
If the condition is true then an error message is printed and an exception thrown.
Definition: Error.h:466
unsigned int bottom
bottom of the border
Definition: Types.h:377
unsigned int num_arguments_per_3D_tensor() const
Returns the number of arguments enqueued per 3D tensor object.
Definition: IGCKernel.cpp:147
Output values are defined to match the source pixel whose center is nearest to the sample position...
1 channel, 1 F16 per channel
virtual ValidRegion valid_region() const =0
Valid region of the tensor.
Implementation of a static rectangular access pattern.
std::string tolower(std::string string)
Convert string to lower case.
Definition: Utility.h:203
bool update_window_and_padding(Window &win, Ts &&... patterns)
Update window and padding size for each of the access patterns.
Definition: WindowHelpers.h:46
SamplingPolicy sampling_policy
Sampling policy used by the interpolation.
virtual const TensorShape & tensor_shape() const =0
Size for each dimension of the tensor.
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Definition: Validate.h:543
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:71
Class to describe a number of elements in each dimension.
Definition: Steps.h:40
Implementation of a row access pattern.
std::string kernel_name
const std::string & string_from_interpolation_policy(InterpolationPolicy policy)
Translates a given interpolation policy to a string.
Definition: Utils.cpp:212
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor&#39;s metadata.
Samples are taken at pixel top left corner.
unsigned int left
left of the border
Definition: Types.h:378
unsigned int right
right of the border
Definition: Types.h:376
Output values are determined by averaging the source pixels whose areas fall under the area of the de...
#define ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:790
ValidRegion calculate_valid_region_scale(const ITensorInfo &src_info, const TensorShape &dst_shape, InterpolationPolicy interpolate_policy, SamplingPolicy sampling_policy, bool border_undefined)
Helper function to calculate the Valid Region for Scale.
Definition: Helpers.cpp:28
static GCKernelLibrary & get()
Get the static instance of GCKernelLibrary.
Borders are left undefined.
GCKernel create_kernel(const std::string &shader_name, const StringSet &build_options_set={}) const
Creates a kernel from the kernel library.
unsigned int num_elems_processed_per_iteration
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:161
Container for valid region of a window.
Definition: Types.h:188
Describe a multidimensional execution window.
Definition: Window.h:39

◆ 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()).

Reimplemented from IGCSimple3DKernel.

Definition at line 150 of file GCScaleKernel.cpp.

References IGCKernel::add_3D_tensor_argument(), ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW, ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL, Window::DimX, arm_compute::enqueue(), Window::first_slice_window_3D(), Tensor::info(), ITensorInfo::padding(), Window::shift(), arm_compute::test::validation::reference::slice(), Window::slide_window_slice_3D(), and IKernel::window().

151 {
154 
155  _kernel.use();
156 
157  _output->set_needs_shifting(true);
158 
160  Window slice_in = window.first_slice_window_3D();
161 
162  slice.shift(Window::DimX, -(_output->info()->padding()).left);
163 
164  do
165  {
166  unsigned int idx = 0;
167  add_3D_tensor_argument(idx, _input, 1, slice_in);
168  add_3D_tensor_argument(idx, _output, 2, slice);
169  _kernel.update_shader_params();
170  enqueue(*this, slice);
171  }
172  while(window.slide_window_slice_3D(slice) && window.slide_window_slice_3D(slice_in));
173 }
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&#39;s parameters to the object&#39;s kernel&#39;s arguments starting from the index idx...
Definition: IGCKernel.cpp:132
void enqueue(IGCKernel &kernel, const Window &window, const gles::NDRange &lws=gles::NDRange(1U, 1U, 1U))
Add the kernel to the command queue with the given window.
Definition: IGCKernel.cpp:41
void shift(size_t dimension, int shift_value)
Shift the values of a given dimension by the given shift_value.
Definition: Window.inl:133
ITensorInfo * info() const override
Interface to be implemented by the child class to return the tensor&#39;s metadata.
Definition: Tensor.cpp:33
static constexpr size_t DimX
Alias for dimension 0 also known as X dimension.
Definition: Window.h:43
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:335
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:941
Window first_slice_window_3D() const
First 3D slice of the window.
Definition: Window.h:291
Describe a multidimensional execution window.
Definition: Window.h:39
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:205
SimpleTensor< T > slice(const SimpleTensor< T > &src, Coordinates starts, Coordinates ends)

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