Compute Library
 21.02
CLGatherKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2018-2020 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  */
29 #include "support/StringSupport.h"
30 
31 #include <string>
32 
33 namespace arm_compute
34 {
35 namespace
36 {
37 inline Status validate_arguments(const ITensorInfo *input, const ITensorInfo *indices, const ITensorInfo *output, int axis)
38 {
39  ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, indices, output);
40  const uint32_t actual_axis = wrap_around(axis, static_cast<int>(input->num_dimensions()));
41  ARM_COMPUTE_RETURN_ERROR_ON(indices->num_dimensions() > 1);
42  ARM_COMPUTE_RETURN_ERROR_ON(input->num_dimensions() > 4);
43  ARM_COMPUTE_RETURN_ERROR_ON(actual_axis >= input->num_dimensions());
44  ARM_COMPUTE_RETURN_ERROR_ON(input->data_type() == DataType::UNKNOWN);
45 
46  if(output->total_size() != 0)
47  {
50  TensorShape output_shape = arm_compute::misc::shape_calculator::compute_gather_shape(input->tensor_shape(), indices->tensor_shape(), actual_axis);
51  ARM_COMPUTE_RETURN_ERROR_ON(output_shape.total_size() != output->tensor_shape().total_size());
52  }
53 
55 
56  return Status{};
57 }
58 
59 std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input, ITensorInfo *indices, ITensorInfo *output, int axis)
60 {
61  ARM_COMPUTE_ERROR_ON_NULLPTR(input, output, indices);
62  const uint32_t actual_axis = wrap_around(axis, static_cast<int>(input->num_dimensions()));
63  // Output auto initialization if not yet initialized
64  TensorShape output_shape = arm_compute::misc::shape_calculator::compute_gather_shape(input->tensor_shape(), indices->tensor_shape(), actual_axis);
65  auto_init_if_empty((*output), output_shape, 1, input->data_type());
66 
67  // Create window
68  Window win = calculate_max_window(*output, Steps());
69  output->set_valid_region(ValidRegion(Coordinates(), output->tensor_shape()));
70 
71  return std::make_pair(Status{}, win);
72 }
73 
74 } // namespace
75 
77  : _input(nullptr), _indices(nullptr), _output(nullptr), _axis(0)
78 {
79 }
80 
81 void CLGatherKernel::configure(const ICLTensor *input, const ICLTensor *indices, ICLTensor *output, int axis)
82 {
83  configure(CLKernelLibrary::get().get_compile_context(), input, indices, output, axis);
84 }
85 
86 void CLGatherKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *indices, ICLTensor *output, int axis)
87 {
88  ARM_COMPUTE_ERROR_ON_NULLPTR(input, output, indices);
89  auto padding_info = get_padding_info({ input, output, indices });
90  ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), indices->info(), output->info(), axis));
91 
92  // Configure kernel window
93  auto win_config = validate_and_configure_window(input->info(), indices->info(), output->info(), axis);
94  ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
95 
96  _input = input;
97  _output = output;
98  _indices = indices;
99  _axis = wrap_around(axis, static_cast<int>(input->info()->num_dimensions()));
100 
101  // Set build options
102  CLBuildOptions build_opts;
103  build_opts.add_option("-DDATA_TYPE=" + get_cl_unsigned_type_from_element_size(data_size_from_type(input->info()->data_type())));
104  build_opts.add_option("-DOUTPUT_DIM_Z=" + support::cpp11::to_string(output->info()->dimension(2)));
105  build_opts.add_option("-DINPUT_DIM_Z=" + support::cpp11::to_string(input->info()->dimension(2)));
106  build_opts.add_option("-DAXIS=" + support::cpp11::to_string(_axis));
107 
108  // Create kernel
109  _kernel = create_kernel(compile_context, "gather", build_opts.options());
110  ICLKernel::configure_internal(win_config.second);
112 }
113 
114 Status CLGatherKernel::validate(const ITensorInfo *input, const ITensorInfo *indices, const ITensorInfo *output, int axis)
115 {
116  ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, indices, output, axis));
117  ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(input->clone().get(), indices->clone().get(), output->clone().get(), axis).first);
118  return Status{};
119 }
120 
121 void CLGatherKernel::run(const Window &window, cl::CommandQueue &queue)
122 {
125 
126  Window window_collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
127  unsigned int idx = 0;
128  add_4D_tensor_argument(idx, _input, window_collapsed);
129  add_1D_tensor_argument(idx, _indices, window_collapsed);
130  add_4D_tensor_argument(idx, _output, window_collapsed);
131  enqueue(queue, *this, window_collapsed, lws_hint());
132 }
133 } // namespace arm_compute
virtual size_t num_dimensions() const =0
The number of dimensions of the tensor (rank)
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
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.
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_QUANTIZATION_INFO(...)
Definition: Validate.h:610
const StringSet & options() const
Gets the current options list set.
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
Definition: ICLKernel.h:276
void run(const Window &window, cl::CommandQueue &queue) override
Enqueue the OpenCL kernel to process the given window on the passed OpenCL command queue...
#define ARM_COMPUTE_RETURN_ON_ERROR(status)
Checks if a status contains an error and returns it.
Definition: Error.h:204
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.
#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
static CLKernelLibrary & get()
Access the KernelLibrary singleton.
Store the tensor&#39;s metadata.
Definition: ITensorInfo.h:40
#define ARM_COMPUTE_ERROR_THROW_ON(status)
Definition: Error.h:455
Status class.
Definition: Error.h:52
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
Definition: Error.h:296
Copyright (c) 2017-2021 Arm Limited.
#define ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(...)
Definition: Validate.h:163
1 channel, 1 S32 per channel
void add_option(std::string option)
Adds option to the existing build option list.
T wrap_around(T x, T m)
Wrap-around a number within the range 0 <= x < m.
Definition: Helpers.h:231
cl::Kernel create_kernel(const CLCompileContext &ctx, const std::string &kernel_name, const std::set< std::string > &build_opts=std::set< std::string >())
Creates an opencl kernel using a compile context.
Definition: CLHelpers.cpp:403
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
1 channel, 1 U32 per channel
void configure(const ICLTensor *input, const ICLTensor *indices, ICLTensor *output, int axis=0)
Initialise the kernel&#39;s inputs and outputs.
static Status validate(const ITensorInfo *input, const ITensorInfo *indices, const ITensorInfo *output, int axis=0)
Static function to check if given info will lead to a valid configuration of CLGatherKernel.
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.
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor&#39;s metadata.
size_t data_size_from_type(DataType data_type)
The size in bytes of the data type.
Definition: Utils.h:106
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:941
bool has_padding_changed(const std::unordered_map< const ITensorInfo *, PaddingSize > &padding_map)
Check if the previously stored padding info has changed after configuring a kernel.
Definition: Utils.cpp:528
CLCompileContext class.
Interface for OpenCL tensor.
Definition: ICLTensor.h:42
static constexpr size_t DimZ
Alias for dimension 2 also known as Z dimension.
Definition: Window.h:47
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Definition: Validate.h:545
std::string get_cl_unsigned_type_from_element_size(size_t element_size)
Translates the element size to an unsigned integer data type.
Definition: CLHelpers.cpp:103
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:792
std::unordered_map< const ITensorInfo *, PaddingSize > get_padding_info(std::initializer_list< const ITensorInfo *> infos)
Stores padding information before configuring a kernel.
Definition: Utils.cpp:513
Status validate_arguments(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output, const GEMMLowpOutputStageInfo *output_stage)
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:161
void add_1D_tensor_argument(unsigned int &idx, const ICLTensor *tensor, const Window &window)
Add the passed 1D tensor&#39;s parameters to the object&#39;s kernel&#39;s arguments starting from the index idx...
Definition: ICLKernel.h:124
void add_4D_tensor_argument(unsigned int &idx, const ICLTensor *tensor, const Window &window)
Add the passed 4D tensor&#39;s parameters to the object&#39;s kernel&#39;s arguments starting from the index idx...
Definition: ICLKernel.h:182
Describe a multidimensional execution window.
Definition: Window.h:39
CLGatherKernel()
Default constructor.
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:205
TensorShape compute_gather_shape(const TensorShape &input_shape, const TensorShape &indices_shape, uint32_t actual_axis)