Compute Library
 21.02
GCCol2ImKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2017-2020 Arm Limited.
3  *
4  * SPDX-License-Identifier: MIT
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to
8  * deal in the Software without restriction, including without limitation the
9  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10  * sell copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in all
14  * copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22  * SOFTWARE.
23  */
24 
26 
27 #include "arm_compute/core/Error.h"
33 #include "arm_compute/core/Types.h"
38 #include "support/StringSupport.h"
39 
40 using namespace arm_compute;
41 
43  : _input(nullptr), _output(nullptr), _convolved_dims()
44 {
45 }
46 
48  std::pair<unsigned int, unsigned int> convolved_dims)
49 {
52 
54  output_shape.set(0, convolved_dims.first);
55  output_shape.set(1, convolved_dims.second);
56  output_shape.set(2, input->info()->tensor_shape()[0]);
57 
58  // Output auto inizialitation if not yet initialized
59  auto_init_if_empty(*output->info(), input->info()->clone()->set_tensor_shape(output_shape));
60 
63 
64  _input = input;
65  _output = output;
66  _convolved_dims = convolved_dims;
67 
68  const DataType dt = input->info()->data_type();
69  const unsigned int local_size = 1;
70 
71  // Create kernel
72  std::set<std::string> build_opts;
73  build_opts.emplace("#define COL2IM ");
74  build_opts.emplace("#define WIDTH_OUTPUT " + support::cpp11::to_string(_convolved_dims.first));
75  const std::string dt_name = (dt == DataType::F32) ? "DATA_TYPE_FP32" : "DATA_TYPE_FP16";
76  build_opts.emplace(("#define " + dt_name));
77  build_opts.emplace("#define LOCAL_SIZE_X " + support::cpp11::to_string(local_size));
78  build_opts.emplace("#define LOCAL_SIZE_Y " + support::cpp11::to_string(local_size));
79  build_opts.emplace("#define LOCAL_SIZE_Z " + support::cpp11::to_string(local_size));
80 
81  _kernel = static_cast<GCKernel>(GCKernelLibrary::get().create_kernel("col2im", build_opts));
82 
83  // Configure window
84  const unsigned int num_elems_processed_per_iteration = (dt == DataType::F32) ? 1 : 2;
85 
86  Window win = calculate_max_window(*output->info(), Steps(num_elems_processed_per_iteration));
87 
89  const int input_padding = ceil_to_multiple(input->info()->dimension(0), 2) - input->info()->dimension(0);
90 
91  AccessWindowStatic input_access(input->info(), 0, 0, input->info()->dimension(0) + input_padding, input->info()->dimension(1) + 1);
92 
93  update_window_and_padding(win, input_access, output_access);
94 
95  output_access.set_valid_region(win, output->info()->valid_region());
96 
97  IGCKernel::configure(win);
98 }
99 
101 {
104 
105  _kernel.use();
106 
107  Window collapsed_window = window.collapse_if_possible(IGCKernel::window(), Window::DimZ);
108  Window slice = collapsed_window.first_slice_window_3D();
109 
110  // Set static kernel arguments
111  unsigned int idx = 2 * num_arguments_per_3D_tensor();
112  //_kernel.set_argument(idx++, _output->info()->strides_in_bytes()[3]);
113  _kernel.set_argument(idx++, uint(_output->info()->dimension(2)));
114  _kernel.set_argument(idx++, _input->info()->strides_in_bytes()[2]);
115 
116  do
117  {
118  // Set inputs
119  unsigned int idx = 0;
120  add_2D_tensor_argument(idx, _input, 1, slice);
121  add_3D_tensor_argument(idx, _output, 2, slice);
122  _kernel.update_shader_params();
123  enqueue(*this, slice);
124  }
125  while(collapsed_window.slide_window_slice_3D(slice));
126 }
Window calculate_max_window(const ValidRegion &valid_region, const Steps &steps, bool skip_border, BorderSize border_size)
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
GCCol2ImKernel()
Default constructor.
Shape of a tensor.
Definition: TensorShape.h:39
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
virtual size_t dimension(size_t index) const =0
Return the size of the requested dimension.
std::string to_string(T &&value)
Convert integer and float values to string.
virtual DataType data_type() const =0
Data type used for each element of the tensor.
1 channel, 1 F32 per channel
Interface for GLES Compute tensor.
Definition: IGCTensor.h:35
GCKernel class.
unsigned int num_arguments_per_3D_tensor() const
Returns the number of arguments enqueued per 3D tensor object.
Definition: IGCKernel.cpp:147
Copyright (c) 2017-2021 Arm Limited.
1 channel, 1 F16 per channel
virtual ValidRegion valid_region() const =0
Valid region of the tensor.
Implementation of a static rectangular access pattern.
DataType dt
bool update_window_and_padding(Window &win, Ts &&... patterns)
Update window and padding size for each of the access patterns.
Definition: WindowHelpers.h:46
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(f, w)
Definition: Validate.h:183
Window collapse_if_possible(const Window &full_window, size_t first, size_t last, bool *has_collapsed=nullptr) const
Collapse the dimensions between first and last if possible.
Definition: Window.inl:68
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
Manages all the GLES kernels compilation and caching, provides accessors for the GLES Context...
Class to describe a number of elements in each dimension.
Definition: Steps.h:40
Implementation of a row access pattern.
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...
virtual std::unique_ptr< T > clone() const =0
Provide a clone of the current object of class T.
void run(const Window &window) override
Enqueue the OpenGL ES shader to process the given window.
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor&#39;s metadata.
void add_2D_tensor_argument(unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
Add the passed 2D tensor&#39;s parameters to the object&#39;s kernel&#39;s arguments starting from the index idx...
Definition: IGCKernel.cpp:127
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
#define ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:790
Wrapper to configure the Khronos EGL and OpenGL ES C header.
static GCKernelLibrary & get()
Get the static instance of GCKernelLibrary.
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_DIMENSIONS(...)
Definition: Validate.h:286
static constexpr size_t DimZ
Alias for dimension 2 also known as Z dimension.
Definition: Window.h:47
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
virtual const Strides & strides_in_bytes() const =0
The strides in bytes for accessing each dimension of the tensor.
Window first_slice_window_3D() const
First 3D slice of the window.
Definition: Window.h:291
DataType
Available data types.
Definition: Types.h:77
Describe a multidimensional execution window.
Definition: Window.h:39
TensorShape & set(size_t dimension, size_t value, bool apply_dim_correction=true, bool increase_dim_unit=true)
Accessor to set the value of one of the dimensions.
Definition: TensorShape.h:79
SimpleTensor< T > slice(const SimpleTensor< T > &src, Coordinates starts, Coordinates ends)
void configure(const IGCTensor *input, IGCTensor *output, std::pair< unsigned int, unsigned int > convolved_dims)
Set the input and output of the kernel.