1
// yala is pleased to support the open source community by making ncnn available.
4
// Copyright (C) 2022 yala <zhaojunchao@loongson.cn>;<junchao82@qq.com>. All rights reserved.
5
// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
6
// in compliance with the License. You may obtain a copy of the License at
8
// https://opensource.org/licenses/BSD-3-Clause
10
// Unless required by applicable law or agreed to in writing, software distributed
11
// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
12
// CONDITIONS OF ANY KIND, either express or implied. See the License for the
13
// specific language governing permissions and limitations under the License.
15
#include "absval_loongarch.h"
19
#endif // __loongarch_sx
23
AbsVal_loongarch::AbsVal_loongarch()
26
support_packing = true;
30
int AbsVal_loongarch::forward_inplace(Mat& bottom_top_blob, const Option& opt) const
32
int w = bottom_top_blob.w;
33
int h = bottom_top_blob.h;
34
int d = bottom_top_blob.d;
35
int channels = bottom_top_blob.c;
36
int elempack = bottom_top_blob.elempack;
37
int size = w * h * d * elempack;
39
#pragma omp parallel for num_threads(opt.num_threads)
40
for (int q = 0; q < channels; q++)
42
float* ptr = bottom_top_blob.channel(q);
46
for (; i + 3 < size; i += 4)
48
__builtin_prefetch(ptr + 16);
49
__m128i _p = __lsx_vld(ptr, 0);
50
__m128i _outp = __lsx_vbitclri_w(_p, 31);
51
__lsx_vst(_outp, ptr, 0);
55
#endif // __loongarch_sx
58
*ptr = *ptr > 0 ? *ptr : -*ptr;