File size: 1,855 Bytes
be903e2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 | // Tencent is pleased to support the open source community by making ncnn available.
//
// Copyright (C) 2017 THL A29 Limited, a Tencent company. All rights reserved.
//
// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
// in compliance with the License. You may obtain a copy of the License at
//
// https://opensource.org/licenses/BSD-3-Clause
//
// Unless required by applicable law or agreed to in writing, software distributed
// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
// CONDITIONS OF ANY KIND, either express or implied. See the License for the
// specific language governing permissions and limitations under the License.
#include "bias_arm.h"
#if __ARM_NEON
#include <arm_neon.h>
#endif // __ARM_NEON
namespace ncnn {
int Bias_arm::forward_inplace(Mat& bottom_top_blob, const Option& opt) const
{
int w = bottom_top_blob.w;
int h = bottom_top_blob.h;
int d = bottom_top_blob.d;
int channels = bottom_top_blob.c;
int size = w * h * d;
const float* bias_ptr = bias_data;
#pragma omp parallel for num_threads(opt.num_threads)
for (int q = 0; q < channels; q++)
{
float* ptr = bottom_top_blob.channel(q);
float bias = bias_ptr[q];
#if __ARM_NEON
int nn = size >> 2;
int remain = size - (nn << 2);
#else
int remain = size;
#endif // __ARM_NEON
#if __ARM_NEON
float32x4_t _bias = vdupq_n_f32(bias);
for (; nn > 0; nn--)
{
float32x4_t _p = vld1q_f32(ptr);
float32x4_t _outp = vaddq_f32(_p, _bias);
vst1q_f32(ptr, _outp);
ptr += 4;
}
#endif // __ARM_NEON
for (; remain > 0; remain--)
{
*ptr = *ptr + bias;
ptr++;
}
}
return 0;
}
} // namespace ncnn
|