Compute Library
 21.11
ClTransposeKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2017-2021 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"
34 #include "src/core/CL/CLValidate.h"
37 #include "support/Cast.h"
38 #include "support/StringSupport.h"
39 
40 namespace arm_compute
41 {
42 namespace opencl
43 {
44 namespace kernels
45 {
47 {
49 }
50 
52 {
54 
55  // Output auto initialization if not yet initialized
57  auto_init_if_empty(*dst, src->clone()->set_tensor_shape(dst_shape));
58 
60  auto padding_info = get_padding_info({ src, dst });
61 
62  // Create kernel
63  const unsigned int vec_size_x = adjust_vec_size(max_cl_vector_width / src->element_size(), src->dimension(0));
64  const int vec_size_x_leftovers = src->dimension(0) % vec_size_x;
65  const unsigned int vec_size_y = adjust_vec_size(max_cl_vector_width / src->element_size(), src->dimension(1));
66  const int vec_size_y_leftovers = src->dimension(1) % vec_size_y;
67 
68  CLBuildOptions build_opts;
69  build_opts.add_option("-DDATA_TYPE_IN_BYTES=" + support::cpp11::to_string(src->element_size()));
70  build_opts.add_option("-DVEC_SIZE_X=" + support::cpp11::to_string(vec_size_x));
71  build_opts.add_option("-DVEC_SIZE_LEFTOVER_X=" + support::cpp11::to_string(vec_size_x_leftovers));
72  build_opts.add_option("-DVEC_SIZE_Y=" + support::cpp11::to_string(vec_size_y));
73  build_opts.add_option("-DVEC_SIZE_LEFTOVER_Y=" + support::cpp11::to_string(vec_size_y_leftovers));
74 
75  _kernel = create_kernel(compile_context, "transpose", build_opts.options());
76 
77  // Configure kernel window
78  Window win = calculate_max_window(*src, Steps(vec_size_x, vec_size_y));
79  ICLKernel::configure_internal(win, cl::NDRange(2, 8));
81 }
82 
84 {
88  ARM_COMPUTE_RETURN_ERROR_ON_MSG(src->num_dimensions() > 2, "Transpose up to 2-D src tensor is supported");
89 
90  // Validate configured dst
91  if(dst->total_size() != 0)
92  {
93  const TensorInfo dst_info = src->clone()->set_tensor_shape(misc::shape_calculator::compute_transposed_shape(*src));
95 
98  }
99 
100  return Status{};
101 }
102 
103 void ClTransposeKernel::run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue)
104 {
107 
108  const auto src = utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC));
109  auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST));
110 
112 
113  do
114  {
115  unsigned int idx = 0;
116  add_2D_tensor_argument(idx, src, slice);
117  add_2D_tensor_argument(idx, dst, slice);
118  enqueue(queue, *this, slice, lws_hint());
119  }
120  while(window.slide_window_slice_2D(slice));
121 }
122 } // namespace kernels
123 } // namespace opencl
124 } // namespace arm_compute
Window first_slice_window_2D() const
First 2D slice of the window.
Definition: Window.h:283
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)
#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
Shape of a tensor.
Definition: TensorShape.h:39
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:606
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.
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
Definition: ICLKernel.h:318
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.
static Status validate(const ITensorInfo *src, const ITensorInfo *dst)
Static function to check if given info will lead to a valid configuration.
#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
Manages all the OpenCL kernels compilation and caching, provides accessors for the OpenCL Context...
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
void configure(const CLCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
Set the src and dst of the kernel.
SimpleTensor< float > src
Definition: DFT.cpp:155
bool slide_window_slice_2D(Window &slice) const
Slide the passed 2D window slice.
Definition: Window.h:323
Copyright (c) 2017-2021 Arm Limited.
TensorShape compute_transposed_shape(const ITensorInfo &input)
Calculate the transposed shape of a tensor.
#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
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(f, w)
Definition: Validate.h:179
Class to describe a number of elements in each dimension.
Definition: Steps.h:40
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 size_t element_size() const =0
Element size in bytes calculated as data_size() * num_channels()
Elementeise CL kernel type.
Definition: CLTypes.h:84
#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
Enqueue the OpenCL kernel to process the given window on the passed OpenCL command queue...
void add_2D_tensor_argument(unsigned int &idx, const ICLTensor *tensor, 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: ICLKernel.h:190
ITensor * get_tensor(int id)
Get tensor of a given id from the pac.
Definition: ITensorPack.cpp:64
virtual size_t total_size() const =0
Returns the total size of the tensor in bytes.
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(...)
Definition: Validate.h:439
#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
#define ARM_COMPUTE_RETURN_ERROR_ON_MSG(cond, msg)
If the condition is true, an error is returned.
Definition: Error.h:244
Tensor packing service.
Definition: ITensorPack.h:39
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:157
Store the tensor&#39;s metadata.
Definition: TensorInfo.h:43
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
Describe a multidimensional execution window.
Definition: Window.h:39
SimpleTensor< T > slice(const SimpleTensor< T > &src, Coordinates starts, Coordinates ends)