xref: /aosp_15_r20/external/ComputeLibrary/src/core/NEON/wrapper/scalar/add.h (revision c217d954acce2dbc11938adb493fc0abd69584f3)
1 /*
2  * Copyright (c) 2018-2020 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  */
24 #ifndef ARM_COMPUTE_WRAPPER_SCALAR_ADD_H
25 #define ARM_COMPUTE_WRAPPER_SCALAR_ADD_H
26 
27 #include <arm_neon.h>
28 
29 namespace arm_compute
30 {
31 namespace wrapper
32 {
add_sat(const uint8_t & a,const uint8_t & b)33 inline uint8_t add_sat(const uint8_t &a, const uint8_t &b)
34 {
35     const uint8x8_t va = { a, 0, 0, 0, 0, 0, 0, 0 };
36     const uint8x8_t vb = { b, 0, 0, 0, 0, 0, 0, 0 };
37     return vget_lane_u8(vqadd_u8(va, vb), 0);
38 }
39 
add_sat(const int16_t & a,const int16_t & b)40 inline int16_t add_sat(const int16_t &a, const int16_t &b)
41 {
42     const int16x4_t va = { a, 0, 0, 0 };
43     const int16x4_t vb = { b, 0, 0, 0 };
44     return vget_lane_s16(vqadd_s16(va, vb), 0);
45 }
46 
add_sat(const int32_t & a,const int32_t & b)47 inline int32_t add_sat(const int32_t &a, const int32_t &b)
48 {
49     const int32x2_t va = { a, 0 };
50     const int32x2_t vb = { b, 0 };
51     return vget_lane_s32(vqadd_s32(va, vb), 0);
52 }
53 
add_sat(const float & a,const float & b)54 inline float add_sat(const float &a, const float &b)
55 {
56     // No notion of saturation exists in floating point
57     return a + b;
58 }
59 
60 #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
add_sat(const float16_t & a,const float16_t & b)61 inline float16_t add_sat(const float16_t &a, const float16_t &b)
62 {
63     // No notion of saturation exists in floating point
64     return a + b;
65 }
66 #endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
67 } // namespace wrapper
68 } // namespace arm_compute
69 #endif /* ARM_COMPUTE_WRAPPER_SCALAR_ADD_H */
70