Compute Library
 19.08
CLLocallyConnectedMatrixMultiplyKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2017-2018 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 
32 #include "arm_compute/core/Error.h"
34 #include "arm_compute/core/Types.h"
35 #include "arm_compute/core/Utils.h"
37 
38 #include <set>
39 #include <sstream>
40 #include <string>
41 
42 using namespace arm_compute;
43 
45  : _input0(nullptr), _input1(nullptr), _output(nullptr)
46 {
47 }
48 
49 namespace
50 {
51 Status validate_arguments(const ITensorInfo *input0, const ITensorInfo *input1, const ITensorInfo *output)
52 {
53  ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input0, input1, output);
59  ARM_COMPUTE_RETURN_ERROR_ON(input0->dimension(0) != input1->dimension(1));
60 
61  return Status{};
62 }
63 
64 std::tuple<Status, Window> validate_and_configure_window(ITensorInfo *input0, ITensorInfo *input1, ITensorInfo *output)
65 {
66  const unsigned int num_elems_processed_per_iteration_x = max_cl_vector_width / data_size_from_type(input0->data_type());
67 
68  Window win = calculate_max_window(*output, Steps(num_elems_processed_per_iteration_x));
69 
70  AccessWindowHorizontal input0_access(input0, 0, num_elems_processed_per_iteration_x);
71  AccessWindowHorizontal input1_access(input1, 0, num_elems_processed_per_iteration_x);
72  AccessWindowHorizontal output_access(output, 0, num_elems_processed_per_iteration_x);
73 
74  bool window_changed = update_window_and_padding(win, input0_access, input1_access, output_access);
75 
76  output_access.set_valid_region(win, ValidRegion(Coordinates(), output->tensor_shape()));
77 
78  Status err = (window_changed) ? ARM_COMPUTE_CREATE_ERROR(ErrorCode::RUNTIME_ERROR, "Insufficient Padding!") : Status{};
79 
80  return std::make_tuple(err, win);
81 }
82 } // namespace
83 
85 {
86  ARM_COMPUTE_ERROR_ON_NULLPTR(input0, input1, output);
87  ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input0->info(), input1->info(), output->info()));
88 
89  _input0 = input0;
90  _input1 = input1;
91  _output = output;
92 
93  cl::NDRange lws_hint;
94  if(output->info()->dimension(1) == 196)
95  {
96  lws_hint = cl::NDRange(1, 7);
97  }
98  else
99  {
100  lws_hint = cl::NDRange(8, 8);
101  }
102 
103  std::ostringstream mm_arguments;
104  std::set<std::string> build_opts;
105 
106  mm_arguments << "-DWIDTH_VECTOR_A=" << input0->info()->dimension(0) << " ";
107  build_opts.emplace(mm_arguments.str());
108 
109  // Create kernel
110  std::string data_type_name = lower_string(string_from_data_type(input0->info()->data_type()));
111  _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel(("gemm_lc_vm_" + data_type_name), build_opts));
112 
113  // Configure kernel window
114  auto win_config = validate_and_configure_window(input0->info(), input1->info(), output->info());
115 
116  ARM_COMPUTE_ERROR_THROW_ON(std::get<0>(win_config));
117 
118  ICLKernel::configure_internal(std::get<1>(win_config), lws_hint);
119 }
120 
122 {
123  ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input0, input1, output));
124  ARM_COMPUTE_RETURN_ON_ERROR(std::get<0>(validate_and_configure_window(input0->clone().get(), input1->clone().get(), output->clone().get())));
125 
126  return Status{};
127 }
128 
129 void CLLocallyConnectedMatrixMultiplyKernel::run(const Window &window, cl::CommandQueue &queue)
130 {
133 
135 
136  Window matrix_b_window;
137  matrix_b_window.use_tensor_dimensions(_input1->info()->tensor_shape());
138  Window slice_matrix_b = matrix_b_window.first_slice_window_3D();
139 
140  do
141  {
142  unsigned int idx = 0;
143  add_2D_tensor_argument(idx, _input0, slice);
144  add_3D_tensor_argument(idx, _input1, slice_matrix_b);
145  add_2D_tensor_argument(idx, _output, slice);
146  enqueue(queue, *this, slice, lws_hint());
147  }
149 }
Window first_slice_window_2D() const
First 2D slice of the window.
Definition: Window.h:267
#define ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(tensor)
Definition: CLValidate.h:34
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
void configure(const ICLTensor *input0, const ICLTensor *input1, ICLTensor *output)
Initialise the kernel's input, output and alpha.
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:39
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Definition: Validate.h:545
std::pair< Status, Window > validate_and_configure_window(ITensorInfo *input, ITensorInfo *weights, ITensorInfo *biases, ITensorInfo *output, const PadStrideInfo &conv_info, unsigned int depth_multiplier, const Size2D &dilation)
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
Definition: ICLKernel.h:247
#define ARM_COMPUTE_RETURN_ON_ERROR(status)
Checks if a status contains an error and returns it.
Definition: Error.h:193
virtual DataType data_type() const =0
Data type used for each element of the tensor.
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:791
1 channel, 1 F32 per channel
static CLKernelLibrary & get()
Access the KernelLibrary singleton.
Store the tensor's metadata.
Definition: ITensorInfo.h:40
#define ARM_COMPUTE_ERROR_THROW_ON(status)
Definition: Error.h:327
Status class.
Definition: Error.h:52
#define ARM_COMPUTE_CREATE_ERROR(error_code,...)
Creates an error with a given message.
Definition: Error.h:167
std::string lower_string(const std::string &val)
Lower a given string.
Definition: Utils.cpp:327
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
Definition: Error.h:244
Window calculate_max_window(const ValidRegion &valid_region, const Steps &steps=Steps(), bool skip_border=false, BorderSize border_size=BorderSize())
Calculate the maximum window for a given tensor shape and border setting.
Definition: Helpers.cpp:28
void add_3D_tensor_argument(unsigned int &idx, const ICLTensor *tensor, const Window &window)
Add the passed 3D tensor's parameters to the object's kernel's arguments starting from the index idx.
Definition: ICLKernel.h:158
void use_tensor_dimensions(const TensorShape &shape, size_t first_dimension=Window::DimX)
Use the tensor's dimensions to fill the window dimensions.
Definition: Window.inl:250
bool slide_window_slice_2D(Window &slice) const
Slide the passed 2D window slice.
Definition: Window.h:307
Copyright (c) 2017-2018 ARM Limited.
1 channel, 1 F16 per channel
const std::string & string_from_data_type(DataType dt)
Convert a data type identity into a string.
Definition: Utils.cpp:144
bool update_window_and_padding(Window &win, Ts &&... patterns)
Update window and padding size for each of the access patterns.
Definition: Helpers.h:402
virtual const TensorShape & tensor_shape() const =0
Size for each dimension of the tensor.
Class to describe a number of elements in each dimension.
Definition: Steps.h:40
Coordinates of an item.
Definition: Coordinates.h:37
Implementation of a row access pattern.
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's metadata.
size_t data_size_from_type(DataType data_type)
The size in bytes of the data type.
Definition: Utils.h:109
std::unique_ptr< Kernel > create_kernel()
Helper function to create and return a unique_ptr pointed to a CL/GLES kernel object.
Definition: Helpers.h:86
#define ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(...)
Definition: Validate.h:163
void add_2D_tensor_argument(unsigned int &idx, const ICLTensor *tensor, const Window &window)
Add the passed 2D tensor's parameters to the object's kernel's arguments starting from the index idx.
Definition: ICLKernel.h:134
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:161
Interface for OpenCL tensor.
Definition: ICLTensor.h:42
void run(const Window &window, cl::CommandQueue &queue) override
Enqueue the OpenCL kernel to process the given window on the passed OpenCL command queue.
static Status validate(const ITensorInfo *input0, const ITensorInfo *input1, const ITensorInfo *output)
Static function to check if given info will lead to a valid configuration of CLLocallyConnectedMatrix...
Container for valid region of a window.
Definition: Types.h:174
Window first_slice_window_3D() const
First 3D slice of the window.
Definition: Window.h:275
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:205
Describe a multidimensional execution window.
Definition: Window.h:39
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:940
SimpleTensor< T > slice(const SimpleTensor< T > &src, Coordinates starts, Coordinates ends)