1 /*
2 * Copyright (c) 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/range/generic/neon/impl.h"
25
26 #include "arm_compute/core/Helpers.h"
27 #include "arm_compute/core/TensorInfo.h"
28 #include "src/core/NEON/wrapper/wrapper.h"
29 #include "src/core/common/Registrars.h"
30
31 namespace arm_compute
32 {
33 namespace cpu
34 {
35 template <typename T>
neon_range_function(ITensor * output,float start,float step,const Window & window)36 void neon_range_function(ITensor *output, float start, float step, const Window &window)
37 {
38 /** SIMD vector tag type. */
39 using ExactTagType = typename wrapper::traits::neon_bitvector<T, wrapper::traits::BitWidth::W128>::tag_type;
40
41 const auto step_vec = wrapper::vdup_n(static_cast<T>(step), ExactTagType{});
42 const auto start_vec = wrapper::vdup_n(static_cast<T>(start), ExactTagType{});
43 auto id_vec = wrapper::vdup_n(static_cast<T>(0.f), ExactTagType{});
44
45 const auto window_start_x = static_cast<int>(window.x().start());
46 const auto window_end_x = static_cast<int>(window.x().end());
47 const int window_step_x = 16 / sizeof(T);
48
49 Window win{ window };
50 win.set(Window::DimX, Window::Dimension(0, 1, 1));
51 Iterator output_it(output, win);
52
53 execute_window_loop(win, [&](const Coordinates &)
54 {
55 int x = window_start_x;
56 const auto out_ptr = reinterpret_cast<T *>(output_it.ptr());
57 for(; x <= (window_end_x - window_step_x); x += window_step_x)
58 {
59 for(int count = 0; count < window_step_x; ++count)
60 {
61 id_vec = wrapper::vsetlane(static_cast<T>(x + count), id_vec, count);
62 }
63
64 // start + step * id
65 const auto res_vec = wrapper::vmla(start_vec, id_vec, step_vec);
66 wrapper::vstore(out_ptr + x, res_vec);
67 }
68
69 // Compute left-over elements
70 for(; x < window_end_x; ++x)
71 {
72 const auto res = start + x * step;
73 *(out_ptr + x) = res;
74 }
75
76 },
77 output_it);
78 }
79
80 template void neon_range_function<uint8_t>(ITensor *output, float start, float step, const Window &window);
81 template void neon_range_function<uint16_t>(ITensor *output, float start, float step, const Window &window);
82 template void neon_range_function<uint32_t>(ITensor *output, float start, float step, const Window &window);
83 template void neon_range_function<int8_t>(ITensor *output, float start, float step, const Window &window);
84 template void neon_range_function<int16_t>(ITensor *output, float start, float step, const Window &window);
85 template void neon_range_function<int32_t>(ITensor *output, float start, float step, const Window &window);
86 template void neon_range_function<float32_t>(ITensor *output, float start, float step, const Window &window);
87
88 #if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS)
89 template void neon_range_function<float16_t>(ITensor *output, float start, float step, const Window &window);
90 #endif /* defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) */
91
92 } // namespace cpu
93 } // namespace arm_compute
94