Compute Library
 21.02
NEBitwiseAndKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2016-2020 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 "arm_compute/core/Types.h"
33 
34 #include <arm_neon.h>
35 #include <cstdint>
36 
37 using namespace arm_compute;
38 using namespace arm_compute::wrapper;
39 
40 namespace arm_compute
41 {
42 class Coordinates;
43 } // namespace arm_compute
44 
45 namespace
46 {
47 template <typename T>
48 inline void bitwise_and(const T *__restrict input1, const T *__restrict input2, T *__restrict output)
49 {
51  const type val1 = vloadq(static_cast<const T *>(input1));
52  const type val2 = vloadq(static_cast<const T *>(input2));
53 
54  vstore(static_cast<T *>(output), vand(val1, val2));
55 }
56 } // namespace
57 
59  : _input1(nullptr), _input2(nullptr), _output(nullptr)
60 {
61 }
62 
63 void NEBitwiseAndKernel::configure(const ITensor *input1, const ITensor *input2, ITensor *output)
64 {
65  ARM_COMPUTE_ERROR_ON_NULLPTR(input1, input2, output);
66 
67  set_shape_if_empty(*output->info(), input1->info()->tensor_shape());
68 
72 
73  ARM_COMPUTE_ERROR_ON_MISMATCHING_SHAPES(input1, input2, output);
77  ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input1, input2, output);
78 
79  _input1 = input1;
80  _input2 = input2;
81  _output = output;
82 
83  constexpr unsigned int num_elems_processed_per_iteration = 16;
84 
85  // Configure kernel window
86  Window win = calculate_max_window(*input1->info(), Steps(num_elems_processed_per_iteration));
88 
92  output_access);
93 
95  input2->info()->valid_region());
96 
97  output_access.set_valid_region(win, valid_region);
98 
99  INEKernel::configure(win);
100 }
101 
103 {
104  ARM_COMPUTE_UNUSED(info);
107  Iterator input1(_input1, window);
108  Iterator input2(_input2, window);
109  Iterator output(_output, window);
110 
111  execute_window_loop(window, [&](const Coordinates &)
112  {
113  bitwise_and<uint8_t>(input1.ptr(), input2.ptr(), output.ptr());
114  },
115  input1, input2, output);
116 }
bool set_format_if_unknown(ITensorInfo &info, Format format)
Set the format, data type and number of channels to the specified value if the current data type is u...
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
1 channel, 1 U8 per channel
uint8x16_t vloadq(const uint8_t *ptr)
Definition: load.h:58
const ValidRegion valid_region
Definition: Scale.cpp:221
decltype(strategy::transforms) typedef type
Interface for Neon tensor.
Definition: ITensor.h:36
Copyright (c) 2017-2021 Arm Limited.
virtual ValidRegion valid_region() const =0
Valid region of the tensor.
void run(const Window &window, const ThreadInfo &info) override
Execute the kernel on the passed window.
bool update_window_and_padding(Window &win, Ts &&... patterns)
Update window and padding size for each of the access patterns.
Definition: WindowHelpers.h:46
#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.
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Definition: Validate.h:543
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.
uint8x8_t vand(const uint8x8_t &a, const uint8x8_t &b)
Definition: and.h:39
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_SHAPES(...)
Definition: Validate.h:441
NEBitwiseAndKernel()
Default constructor.
Create the appropriate Neon vector given its type and size in terms of bits.
Definition: traits.h:92
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor&#39;s metadata.
bool set_shape_if_empty(ITensorInfo &info, const TensorShape &shape)
Set the shape to the specified value if the current assignment is empty.
constexpr uint8_t * ptr() const
Return a pointer to the current pixel.
Definition: Helpers.inl:139
ValidRegion intersect_valid_regions(const Ts &... regions)
Intersect multiple valid regions.
Definition: WindowHelpers.h:74
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:941
#define ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:790
SimpleTensor< T > bitwise_and(const SimpleTensor< T > &src1, const SimpleTensor< T > &src2)
Definition: BitwiseAnd.cpp:35
ScaleKernelInfo info(interpolation_policy, default_border_mode, PixelValue(), sampling_policy, false)
Information about executing thread and CPU.
Definition: CPPTypes.h:235
void vstore(uint8_t *ptr, uint8x8_t val)
Definition: store.h:39
unsigned int num_elems_processed_per_iteration
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
Definition: Validate.h:161
void execute_window_loop(const Window &w, L &&lambda_function, Ts &&... iterators)
Iterate through the passed window, automatically adjusting the iterators and calling the lambda_funct...
Definition: Helpers.inl:77
Includes all wrapper headers at once.
Container for valid region of a window.
Definition: Types.h:188
Iterator updated by execute_window_loop for each window element.
Definition: Helpers.h:46
void configure(const ITensor *input1, const ITensor *input2, ITensor *output)
Initialise the kernel&#39;s inputs and output.
Describe a multidimensional execution window.
Definition: Window.h:39
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
Definition: Validate.h:205