Compute Library
 21.11
ClWidthConcatenateKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2018-2021 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  */
25 
30 #include "arm_compute/core/Utils.h"
31 #include "src/core/CL/CLValidate.h"
33 #include "support/Cast.h"
34 
35 #include "support/StringSupport.h"
36 
37 namespace arm_compute
38 {
39 namespace opencl
40 {
41 namespace kernels
42 {
43 namespace
44 {
45 Status validate_arguments(const ITensorInfo *src, unsigned int width_offset, const ITensorInfo *dst)
46 {
50 
52  ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(0) + width_offset > dst->dimension(0));
53 
54  for(size_t i = 1; i < Coordinates::num_max_dimensions; ++i)
55  {
56  ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(i) != dst->dimension(i));
57  }
58  ARM_COMPUTE_RETURN_ERROR_ON(src->num_dimensions() > 4);
59 
60  return Status{};
61 }
62 } // namespace
63 
65 {
67 }
68 
69 Status ClWidthConcatenateKernel::validate(const ITensorInfo *src, unsigned int width_offset, const ITensorInfo *dst)
70 {
71  ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(src, width_offset, dst));
72  return Status{};
73 }
74 
75 void ClWidthConcatenateKernel::configure(const CLCompileContext &compile_context, ITensorInfo *src, unsigned int width_offset, ITensorInfo *dst)
76 {
78  ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(src, width_offset, dst));
79 
80  auto padding_info = get_padding_info({ src, dst });
81 
82  const unsigned int num_elems_processed_per_iteration = adjust_vec_size(16, src->dimension(0));
83 
84  // Add build options
85  CLBuildOptions build_opts;
86  build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(src->data_type()));
87  build_opts.add_option("-DVEC_SIZE=" + support::cpp11::to_string(num_elems_processed_per_iteration));
88  build_opts.add_option("-DVEC_SIZE_LEFTOVER=" + support::cpp11::to_string(src->dimension(0) % num_elems_processed_per_iteration));
89  build_opts.add_option("-DWIDTH_OFFSET=" + support::cpp11::to_string(width_offset));
90  build_opts.add_option("-DDEPTH=" + support::cpp11::to_string(src->dimension(2)));
91 
93  {
94  const UniformQuantizationInfo iqinfo = src->quantization_info().uniform();
95  const UniformQuantizationInfo oqinfo = dst->quantization_info().uniform();
96 
97  build_opts.add_option("-DOFFSET_IN1=" + float_to_string_with_full_precision(iqinfo.offset));
98  build_opts.add_option("-DOFFSET_OUT=" + float_to_string_with_full_precision(oqinfo.offset));
99  build_opts.add_option("-DSCALE_IN1=" + float_to_string_with_full_precision(iqinfo.scale));
100  build_opts.add_option("-DSCALE_OUT=" + float_to_string_with_full_precision(oqinfo.scale));
101  }
102 
103  // Create kernel
104  _kernel = create_kernel(compile_context, "concatenate_width", build_opts.options());
105  // Configure kernel window
106  Window win = calculate_max_window(*src, Steps(num_elems_processed_per_iteration));
107  ICLKernel::configure_internal(win.collapse(win, Window::DimZ));
108 
110 }
111 
112 void ClWidthConcatenateKernel::run_op(ITensorPack &tensors, const Window &window, ::cl::CommandQueue &queue)
113 {
116 
117  const auto src = utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC));
118  auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST));
119 
120  unsigned int idx = 0;
121  add_4D_tensor_argument(idx, src, window);
122  add_4D_tensor_argument(idx, dst, window);
123  enqueue(queue, *this, window, lws_hint());
124 }
125 } // namespace kernels
126 } // namespace opencl
127 } // namespace arm_compute
Window calculate_max_window(const ValidRegion &valid_region, const Steps &steps, bool skip_border, BorderSize border_size)
#define ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(tensor)
Definition: CLValidate.h:35
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
virtual size_t dimension(size_t index) const =0
Return the size of the requested dimension.
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:32
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
Definition: ICLKernel.h:318
#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
Store the tensor&#39;s metadata.
Definition: ITensorInfo.h:40
#define ARM_COMPUTE_ERROR_THROW_ON(status)
Definition: Error.h:455
Quantization info when assuming per layer quantization.
Manages all the OpenCL kernels compilation and caching, provides accessors for the OpenCL Context...
Status class.
Definition: Error.h:52
static Status validate(const ITensorInfo *src, unsigned int width_offset, const ITensorInfo *dst)
Static function to check if given info will lead to a valid configuration.
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
Definition: Error.h:296
SimpleTensor< float > src
Definition: DFT.cpp:155
Copyright (c) 2017-2021 Arm Limited.
#define ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(...)
Definition: Validate.h:159
void add_option(std::string option)
Adds option to the existing build option list.
const ITensor * get_const_tensor(int id) const
Get constant tensor of a given id.
Definition: ITensorPack.cpp:54
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:391
std::string float_to_string_with_full_precision(float val)
Create a string with the float in full precision.
Definition: Utils.h:1075
Class to describe a number of elements in each dimension.
Definition: Steps.h:40
void configure(const CLCompileContext &compile_context, ITensorInfo *src, unsigned int width_offset, ITensorInfo *dst)
Initialise the kernel&#39;s source and destination.
unsigned int num_elems_processed_per_iteration
UniformQuantizationInfo uniform() const
Return per layer quantization info.
std::string get_cl_type_from_data_type(const DataType &dt)
Translates a tensor data type to the appropriate OpenCL type.
Definition: CLHelpers.cpp:39
Elementeise CL kernel type.
Definition: CLTypes.h:84
virtual QuantizationInfo quantization_info() const =0
Get the quantization settings (scale and offset) of the tensor.
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:915
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:533
CLCompileContext class.
void run_op(ITensorPack &tensors, const Window &window, ::cl::CommandQueue &queue) override
bool is_data_type_quantized_asymmetric(DataType dt)
Check if a given data type is of asymmetric quantized type.
Definition: Utils.h:1003
ITensor * get_tensor(int id)
Get tensor of a given id from the pac.
Definition: ITensorPack.cpp:64
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:541
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:518
Tensor packing service.
Definition: ITensorPack.h:39
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:157
unsigned int adjust_vec_size(unsigned int vec_size, size_t dim0)
Returns the adjusted vector size in case it is less than the input&#39;s first dimension, getting rounded down to its closest valid vector size.
Definition: Utils.h:1171
static constexpr size_t num_max_dimensions
Number of dimensions the tensor has.
Definition: Dimensions.h:46
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:224
Describe a multidimensional execution window.
Definition: Window.h:39
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:201