2 * Copyright (c) 2016-2018 ARM Limited.
4 * SPDX-License-Identifier: MIT
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:
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
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
24 #include "arm_compute/core/NEON/kernels/NEBitwiseAndKernel.h"
26 #include "arm_compute/core/Helpers.h"
27 #include "arm_compute/core/ITensor.h"
28 #include "arm_compute/core/NEON/wrapper/wrapper.h"
29 #include "arm_compute/core/Types.h"
30 #include "arm_compute/core/Validate.h"
35 using namespace arm_compute;
36 using namespace arm_compute::wrapper;
41 } // namespace arm_compute
45 template <typename T, int S>
46 inline void bitwise_and(const T *__restrict input1, const T *__restrict input2, T *__restrict output)
48 using type = typename wrapper::traits::neon_vector<T, S>::type;
49 const type val1 = vloadq(static_cast<const T *>(input1));
50 const type val2 = vloadq(static_cast<const T *>(input2));
52 vstore(static_cast<T *>(output), vand(val1, val2));
56 NEBitwiseAndKernel::NEBitwiseAndKernel()
57 : _input1(nullptr), _input2(nullptr), _output(nullptr)
61 void NEBitwiseAndKernel::configure(const ITensor *input1, const ITensor *input2, ITensor *output)
63 ARM_COMPUTE_ERROR_ON_NULLPTR(input1, input2, output);
65 set_shape_if_empty(*output->info(), input1->info()->tensor_shape());
67 set_format_if_unknown(*output->info(), Format::U8);
68 set_format_if_unknown(*input1->info(), Format::U8);
69 set_format_if_unknown(*input2->info(), Format::U8);
71 ARM_COMPUTE_ERROR_ON_MISMATCHING_SHAPES(input1, input2, output);
72 ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input1, 1, DataType::U8);
73 ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input2, 1, DataType::U8);
74 ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::U8);
75 ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input1, input2, output);
81 constexpr unsigned int num_elems_processed_per_iteration = 16;
83 // Configure kernel window
84 Window win = calculate_max_window(*input1->info(), Steps(num_elems_processed_per_iteration));
85 AccessWindowHorizontal output_access(output->info(), 0, num_elems_processed_per_iteration);
87 update_window_and_padding(win,
88 AccessWindowHorizontal(input1->info(), 0, num_elems_processed_per_iteration),
89 AccessWindowHorizontal(input2->info(), 0, num_elems_processed_per_iteration),
92 const ValidRegion valid_region = intersect_valid_regions(input1->info()->valid_region(),
93 input2->info()->valid_region());
95 output_access.set_valid_region(win, valid_region);
97 INEKernel::configure(win);
100 void NEBitwiseAndKernel::run(const Window &window, const ThreadInfo &info)
102 ARM_COMPUTE_UNUSED(info);
103 ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
104 ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(INEKernel::window(), window);
105 Iterator input1(_input1, window);
106 Iterator input2(_input2, window);
107 Iterator output(_output, window);
109 execute_window_loop(window, [&](const Coordinates & id)
111 bitwise_and<uint8_t, 16>(input1.ptr(), input2.ptr(), output.ptr());
113 input1, input2, output);