File size: 2,028 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 71 72 73 74 75 76 77 | // Leo is pleased to support the open source community by making ncnn available.
//
// Copyright (C) 2019 Leo <leo@nullptr.com.cn>. 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 "clip_mips.h"
#if __mips_msa
#include <msa.h>
#endif // __mips_msa
#include "mips_usability.h"
namespace ncnn {
Clip_mips::Clip_mips()
{
#if __mips_msa
support_packing = true;
#endif
}
int Clip_mips::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 elempack = bottom_top_blob.elempack;
int size = w * h * d * elempack;
#pragma omp parallel for num_threads(opt.num_threads)
for (int q = 0; q < channels; q++)
{
float* ptr = bottom_top_blob.channel(q);
int i = 0;
#if __mips_msa
v4f32 _max = (v4f32)__msa_fill_w_f32(max);
v4f32 _min = (v4f32)__msa_fill_w_f32(min);
for (; i + 3 < size; i += 4)
{
__builtin_prefetch(ptr + 16);
v4f32 _p = (v4f32)__msa_ld_w(ptr, 0);
_p = __msa_fmax_w(_p, _min);
_p = __msa_fmin_w(_p, _max);
__msa_st_w((v4i32)_p, ptr, 0);
ptr += 4;
}
#endif // __mips_msa
for (; i < size; i++)
{
if (*ptr < min)
*ptr = min;
if (*ptr > max)
*ptr = max;
ptr++;
}
}
return 0;
}
} // namespace ncnn
|