Compute Library
 22.02
ClElementwiseUnaryKernel.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 
28 #include "src/core/CL/CLValidate.h"
30 #include "support/Cast.h"
31 #include "support/StringSupport.h"
32 
33 namespace arm_compute
34 {
35 namespace opencl
36 {
37 namespace kernels
38 {
39 namespace
40 {
41 Status validate_arguments(const ITensorInfo &src, const ITensorInfo &dst, const ElementWiseUnary op)
42 {
45  {
47  }
48  else if(op == ElementWiseUnary::NEG)
49  {
51  }
52  else
53  {
55  }
56 
57  // Validate in case of configured dst
58  if(dst.total_size() > 0)
59  {
63  }
64 
65  return Status{};
66 }
67 } // namespace
68 
70 {
72 }
73 
74 void ClElementWiseUnaryKernel::configure(const CLCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst, const ElementWiseUnary &op)
75 {
77 
78  auto padding_info = get_padding_info({ src, dst });
79 
80  ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(*src, *dst, op));
81 
82  const std::string kernel_name = "elementwise_unary";
83  const int vec_size_x = 16 / dst->element_size();
84  const int dst_width_x = dst->tensor_shape().x();
85  const bool multi_access_x = (dst_width_x / vec_size_x > 0);
86 
87  // Set kernel build options
88  CLBuildOptions build_opts;
89  build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(src->data_type()));
90  build_opts.add_option_if(multi_access_x, "-DVEC_SIZE=" + support::cpp11::to_string(vec_size_x));
91  build_opts.add_option_if(multi_access_x, "-DLAST_ACCESSED_X=" + support::cpp11::to_string(std::max<int>(dst_width_x - vec_size_x, 0)));
92  switch(op)
93  {
95  build_opts.add_option("-DOPERATION=rsqrt_op");
96  break;
98  build_opts.add_option("-DOPERATION=exp_op");
99  break;
101  build_opts.add_option("-DOPERATION=neg_op");
102  break;
104  build_opts.add_option("-DOPERATION=sin_op");
105  break;
107  build_opts.add_option("-DOPERATION=fabs_op");
108  break;
110  build_opts.add_option("-DOPERATION=natural_log_op");
111  break;
113  build_opts.add_option("-DOPERATION=round_op");
114  break;
116  build_opts.add_option("-DOPERATION=logical_not_op");
117  break;
118  default:
119  ARM_COMPUTE_ERROR("Not implemented");
120  }
121 
122  // Create kernel
123  _kernel = create_kernel(compile_context, kernel_name, build_opts.options());
124 
125  // Configure kernel window
126  Window win = calculate_max_window(*dst);
127  if(multi_access_x)
128  {
129  win.set(Window::DimX,
130  Window::Dimension(win.x().start(), ceil_to_multiple(win.x().end(), vec_size_x), vec_size_x));
131  }
132  ICLKernel::configure_internal(win);
133 
135 }
136 
138 {
139  ARM_COMPUTE_UNUSED(op);
141  ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(*src, *dst, op));
142 
143  return Status{};
144 }
145 
146 void ClElementWiseUnaryKernel::run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue)
147 {
150 
152  Window slice = collapsed.first_slice_window_3D();
153 
154  const auto src = utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC));
155  auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST));
156 
157  do
158  {
159  unsigned int idx = 0;
160  add_3D_tensor_argument(idx, src, slice);
161  add_3D_tensor_argument(idx, dst, slice);
162  enqueue(queue, *this, slice, lws_hint());
163  }
164  while(collapsed.slide_window_slice_3D(slice));
165 }
166 } // namespace kernels
167 } // namespace opencl
168 } // 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
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
const StringSet & options() const
Gets the current options list set.
#define ARM_COMPUTE_ERROR(msg)
Print the given message then throw an std::runtime_error.
Definition: Error.h:352
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
Definition: ICLKernel.h:353
1 channel, 1 U8 per channel
#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.
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:466
Store the tensor&#39;s metadata.
Definition: ITensorInfo.h:40
#define ARM_COMPUTE_ERROR_THROW_ON(status)
Definition: Error.h:455
Describe one of the image&#39;s dimensions with a start, end and step.
Definition: Window.h:77
Status class.
Definition: Error.h:52
void add_3D_tensor_argument(unsigned int &idx, const ICLTensor *tensor, 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: ICLKernel.h:214
SimpleTensor< float > src
Definition: DFT.cpp:155
Copyright (c) 2017-2021 Arm Limited.
1 channel, 1 F16 per channel
#define ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(...)
Definition: Validate.h:159
1 channel, 1 S32 per channel
void add_option(std::string option)
Adds option to the existing build option list.
T x() const
Alias to access the size of the first dimension.
Definition: Dimensions.h:87
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
static constexpr size_t DimX
Alias for dimension 0 also known as X dimension.
Definition: Window.h:43
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
Definition: Error.h:152
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.
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
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
void add_option_if(bool cond, std::string option)
Adds option if a given condition is true;.
virtual size_t element_size() const =0
Element size in bytes calculated as data_size() * num_channels()
Elementeise CL kernel type.
Definition: CLTypes.h:84
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: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:552
CLCompileContext class.
ITensor * get_tensor(int id)
Get tensor of a given id from the pac.
Definition: ITensorPack.cpp:64
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(...)
Definition: Validate.h:439
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
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:788
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:537
Tensor packing service.
Definition: ITensorPack.h:39
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:157
Window first_slice_window_3D() const
First 3D slice of the window.
Definition: Window.h:291
std::string kernel_name
ElementWiseUnary
Available element wise unary operations.
Definition: Types.h:489
Describe a multidimensional execution window.
Definition: Window.h:39
void run_op(ITensorPack &tensors, 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_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:201
static Status validate(const ITensorInfo *src, const ITensorInfo *dst, const ElementWiseUnary &op)
Static function to check if given info will lead to a valid configuration.
SimpleTensor< T > slice(const SimpleTensor< T > &src, Coordinates starts, Coordinates ends)
void configure(const CLCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst, const ElementWiseUnary &op)
Initialise the kernel&#39;s srcs, dst.