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 */
24 #include "src/cpu/kernels/CpuConcatenateHeightKernel.h"
25
26 #include "arm_compute/core/Error.h"
27 #include "arm_compute/core/Helpers.h"
28 #include "arm_compute/core/ITensor.h"
29 #include "arm_compute/core/TensorInfo.h"
30 #include "arm_compute/core/Utils.h"
31 #include "arm_compute/core/Validate.h"
32 #include "arm_compute/core/Window.h"
33 #include "src/core/NEON/NEAsymm.h"
34 #include "src/core/NEON/wrapper/wrapper.h"
35 #include "src/core/helpers/AutoConfiguration.h"
36 #include "src/core/helpers/WindowHelpers.h"
37
38 #include <cstdint>
39
40 namespace arm_compute
41 {
42 namespace cpu
43 {
44 namespace kernels
45 {
46 namespace
47 {
validate_arguments(const ITensorInfo * src,unsigned int height_offset,const ITensorInfo * dst)48 Status validate_arguments(const ITensorInfo *src, unsigned int height_offset, const ITensorInfo *dst)
49 {
50 ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(src, dst);
51 // Note: ARM_COMPUTE_RETURN_ERROR_ON_CPU_F16_UNSUPPORTED(src) is not needed here as this kernel doesn't use CPU FP16 instructions.
52 ARM_COMPUTE_RETURN_ERROR_ON(src->data_type() == DataType::UNKNOWN);
53 ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(src, dst);
54 ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(Window::DimX) != dst->dimension(Window::DimX));
55 ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(Window::DimY) + height_offset > dst->dimension(Window::DimY));
56 for(size_t i = 2; i < Coordinates::num_max_dimensions; ++i)
57 {
58 ARM_COMPUTE_RETURN_ERROR_ON(src->dimension(i) != dst->dimension(i));
59 }
60
61 return Status{};
62 }
63 } // namespace
64
configure(const ITensorInfo * src,unsigned int height_offset,ITensorInfo * dst)65 void CpuConcatenateHeightKernel::configure(const ITensorInfo *src, unsigned int height_offset, ITensorInfo *dst)
66 {
67 ARM_COMPUTE_UNUSED(src);
68 ARM_COMPUTE_ERROR_ON_NULLPTR(src, dst);
69 ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(src, height_offset, dst));
70
71 _height_offset = height_offset;
72
73 // Configure kernel window
74 Window win = calculate_max_window(*dst, Steps());
75 ICpuKernel::configure(win);
76 }
77
validate(const ITensorInfo * src,unsigned int height_offset,const ITensorInfo * dst)78 Status CpuConcatenateHeightKernel::validate(const ITensorInfo *src, unsigned int height_offset, const ITensorInfo *dst)
79 {
80 ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(src, height_offset, dst));
81 return Status{};
82 }
83
run_op(ITensorPack & tensors,const Window & window,const ThreadInfo & info)84 void CpuConcatenateHeightKernel::run_op(ITensorPack &tensors, const Window &window, const ThreadInfo &info)
85 {
86 ARM_COMPUTE_UNUSED(info);
87 ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
88 ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICpuKernel::window(), window);
89
90 const auto src = tensors.get_const_tensor(TensorType::ACL_SRC);
91 auto dst = tensors.get_tensor(TensorType::ACL_DST);
92
93 // Offset destination pointer to the correct position
94 uint8_t *dst_ptr = dst->buffer() + dst->info()->offset_first_element_in_bytes() + _height_offset * dst->info()->strides_in_bytes()[Window::DimY];
95
96 const auto window_start_x = static_cast<int>(window.x().start());
97 const auto window_end_x = static_cast<int>(window.x().end()) * static_cast<int>(dst->info()->element_size());
98 const int window_step_x = 16;
99
100 Window win{ window };
101 win.set(Window::DimX, Window::Dimension(0, 1, 1));
102 win.set(Window::DimY, Window::Dimension(0, src->info()->tensor_shape().y(), 1));
103
104 // Create iterators
105 Iterator src_it(src, win);
106 Iterator dst_it(dst, win);
107
108 const DataType dt = src->info()->data_type();
109 const UniformQuantizationInfo &src_qinfo = src->info()->quantization_info().uniform();
110 const UniformQuantizationInfo &dst_qinfo = dst->info()->quantization_info().uniform();
111 if(dt == DataType::QASYMM8 && src_qinfo != dst_qinfo)
112 {
113 execute_window_loop(win, [&](const Coordinates &)
114 {
115 int x = window_start_x;
116 for(; x <= (window_end_x - window_step_x); x += window_step_x)
117 {
118 vst1q_u8(dst_ptr + dst_it.offset() + x, vquantize(vdequantize(vld1q_u8(src_it.ptr() + x), src_qinfo), dst_qinfo));
119 }
120
121 // Compute left-over elements
122 for(; x < window_end_x; ++x)
123 {
124 *(dst_ptr + dst_it.offset() + x) = quantize_qasymm8(dequantize_qasymm8(*(src_it.ptr() + x), src_qinfo), dst_qinfo);
125 }
126
127 },
128 src_it, dst_it);
129 }
130 else if(dt == DataType::QASYMM8_SIGNED && src_qinfo != dst_qinfo)
131 {
132 execute_window_loop(win, [&](const Coordinates &)
133 {
134 int x = window_start_x;
135 for(; x <= (window_end_x - window_step_x); x += window_step_x)
136 {
137 vst1q_s8(reinterpret_cast<int8_t *>(dst_ptr + dst_it.offset() + x),
138 vquantize_signed(vdequantize(vld1q_s8(reinterpret_cast<int8_t *>(src_it.ptr()) + x), src_qinfo), dst_qinfo));
139 }
140
141 // Compute left-over elements
142 for(; x < window_end_x; ++x)
143 {
144 *(dst_ptr + dst_it.offset() + x) = quantize_qasymm8_signed(dequantize_qasymm8_signed(*(src_it.ptr() + x), src_qinfo), dst_qinfo);
145 }
146 },
147 src_it, dst_it);
148 }
149 else
150 {
151 execute_window_loop(win, [&](const Coordinates &)
152 {
153 const auto in_ptr = src_it.ptr();
154 const auto out_ptr = dst_ptr + dst_it.offset();
155
156 int x = window_start_x;
157 for(; x <= (window_end_x - window_step_x); x += window_step_x)
158 {
159 wrapper::vstore(out_ptr + x, wrapper::vloadq(in_ptr + x));
160 }
161
162 // Compute left-over elements
163 for(; x < window_end_x; ++x)
164 {
165 *(out_ptr + x) = *(in_ptr + x);
166 }
167 },
168 src_it, dst_it);
169 }
170 }
171
name() const172 const char *CpuConcatenateHeightKernel::name() const
173 {
174 return "CpuConcatenateHeightKernel";
175 }
176 } // namespace kernels
177 } // namespace cpu
178 } // namespace arm_compute
179