Compute Library
 21.02
ClCropKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2019-2021 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 
31 #include "arm_compute/core/Utils.h"
32 #include "src/core/CPP/Validate.h"
34 #include "support/Cast.h"
35 #include "support/StringSupport.h"
36 
37 #include <map>
38 
39 namespace arm_compute
40 {
41 namespace opencl
42 {
43 namespace kernels
44 {
45 void ClCropKernel::configure(const ITensorInfo *src, ITensorInfo *dst, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value, Window *dst_window)
46 {
47  configure(CLKernelLibrary::get().get_compile_context(), src, dst, start, end, batch_index, extrapolation_value, dst_window);
48 }
49 
50 void ClCropKernel::configure(const CLCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst, Coordinates2D start, Coordinates2D end, uint32_t batch_index,
51  float extrapolation_value, Window *dst_window)
52 {
54  ARM_COMPUTE_ERROR_THROW_ON(validate(src, dst, start, end, batch_index, extrapolation_value, dst_window));
55 
56  _start = start;
57  _batch_index = batch_index;
58  _extrapolation_value = extrapolation_value;
59 
60  const int vec_size_x = 4;
61  // Create and update the window (if needed)
62  Window win = calculate_max_window(*dst);
63 
64  if(dst_window != nullptr)
65  {
67  win = *dst_window;
68  }
69 
70  const int dst_width_x = win.num_iterations(0);
71  const bool multi_access_x = dst_width_x >= vec_size_x;
72  const bool remainder_x = dst_width_x % vec_size_x > 0;
73 
74  if(multi_access_x)
75  {
76  win.set(Window::DimX,
77  Window::Dimension(win.x().start(), ceil_to_multiple(win.x().end(), vec_size_x), vec_size_x));
78  }
79  ICLKernel::configure_internal(win);
80 
81  // Create kernel
82  CLBuildOptions build_opts;
83  build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(src->data_type()));
84  build_opts.add_option_if(multi_access_x, "-DVEC_SIZE=" + support::cpp11::to_string(vec_size_x));
85  build_opts.add_option_if(multi_access_x && remainder_x, "-DLAST_ACCESSED_X=" + support::cpp11::to_string(std::max<int>(dst_width_x - vec_size_x, 0)));
86  build_opts.add_option_if(start.x > end.x, "-DWIDTH_FLIPPED=");
87  build_opts.add_option_if(start.y > end.y, "-DHEIGHT_FLIPPED=");
88  _kernel = create_kernel(compile_context, "crop_tensor", build_opts.options());
89 }
90 
91 Status ClCropKernel::validate(const ITensorInfo *src, const ITensorInfo *dst, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value, Window *dst_window)
92 {
93  ARM_COMPUTE_UNUSED(extrapolation_value, dst_window);
98  ARM_COMPUTE_RETURN_ERROR_ON(start.x < 0 || start.y < 0 || end.x < 0 || end.y < 0);
99  ARM_COMPUTE_RETURN_ERROR_ON(start.x >= static_cast<int32_t>(src->dimension(1)) || start.y >= static_cast<int32_t>(src->dimension(2))
100  || end.x >= static_cast<int32_t>(src->dimension(1)) || end.y >= static_cast<int32_t>(src->dimension(2)));
101  ARM_COMPUTE_RETURN_ERROR_ON(batch_index >= src->dimension(3));
102  if(dst_window != nullptr)
103  {
104  ARM_COMPUTE_RETURN_ERROR_ON(dst_window->x().step() != 1);
105  }
106  if(dst->total_size() > 0)
107  {
111  }
112  return Status{};
113 }
114 
115 void ClCropKernel::run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue)
116 {
119 
120  const auto src = utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC));
121  auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST));
122 
123  Window in_slice = Window();
124  in_slice.use_tensor_dimensions(src->info()->tensor_shape());
125  in_slice.set(Window::DimX, Window::Dimension(in_slice.x().start(), ceil_to_multiple(in_slice.x().end(), window.x().step()), window.x().step()));
126  in_slice.set(3, Window::Dimension(_batch_index, _batch_index + 1, 1));
127 
128  unsigned int idx = 0;
129  add_3D_tensor_argument(idx, src, in_slice);
130  add_3D_tensor_argument(idx, dst, window);
131  add_argument(idx, _start.x);
132  add_argument(idx, _start.y);
133  enqueue(queue, *this, window, lws_hint());
134 }
135 } // namespace kernels
136 } // namespace opencl
137 } // 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
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_LAYOUT_NOT_IN(t,...)
Definition: Validate.h:746
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
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_LAYOUT(...)
Definition: Validate.h:494
#define ARM_COMPUTE_RETURN_ERROR_ON_CPU_F16_UNSUPPORTED(tensor)
Definition: Validate.h:108
virtual size_t dimension(size_t index) const =0
Return the size of the requested dimension.
const StringSet & options() const
Gets the current options list set.
constexpr int step() const
Return the step of the dimension.
Definition: Window.h:104
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
Definition: ICLKernel.h:276
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.
void configure(const ITensorInfo *src, ITensorInfo *dst, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value=0, Window *dst_window=nullptr)
Configure kernel.
1 channel, 1 F32 per channel
void add_argument(unsigned int &idx, T value)
Add the passed parameters to the object&#39;s kernel&#39;s arguments starting from the index idx...
Definition: ICLKernel.h:255
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
Describe one of the image&#39;s dimensions with a start, end and step.
Definition: Window.h:77
Status class.
Definition: Error.h:52
constexpr size_t num_iterations(size_t dimension) const
Return the number of iterations needed to iterate through a given dimension.
Definition: Window.inl:182
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
Definition: Error.h:296
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:172
SimpleTensor< float > src
Definition: DFT.cpp:155
Copyright (c) 2017-2021 Arm Limited.
int32_t x
X coordinates.
Definition: Types.h:465
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:40
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
int32_t y
Y coordinates.
Definition: Types.h:466
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
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:37
void add_option_if(bool cond, std::string option)
Adds option if a given condition is true;.
void end(TokenStream &in, bool &valid)
Definition: MLGOParser.cpp:290
void set(size_t dimension, const Dimension &dim)
Set the values of a given dimension.
Definition: Window.inl:49
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:941
CLCompileContext class.
ITensor * get_tensor(int id)
Get tensor of a given id from the pac.
Definition: ITensorPack.cpp:50
virtual size_t total_size() const =0
Returns the total size of the tensor in bytes.
Coordinate type.
Definition: Types.h:463
Manages all the OpenCL kernels compilation and caching, provides accessors for the OpenCL Context...
unsigned int num_dimensions() const
Returns the effective dimensionality of the tensor.
Definition: Dimensions.h:143
Num samples, height, width, channels.
Tensor packing service.
Definition: ITensorPack.h:37
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:161
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_NOT_IN(t,...)
Definition: Validate.h:694
constexpr int end() const
Return the end of the dimension.
Definition: Window.h:99
static Status validate(const ITensorInfo *src, const ITensorInfo *dst, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value=0, Window *dst_window=nullptr)
Static function to check if given info will lead to a valid configuration of CLStridedSliceKernel.
constexpr int start() const
Return the start of the dimension.
Definition: Window.h:94
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...
Describe a multidimensional execution window.
Definition: Window.h:39
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:205
constexpr const Dimension & x() const
Alias to access the first dimension of the window.
Definition: Window.h:145