1 /** 2 * Copyright 2020 Huawei Technologies Co., Ltd 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 #ifdef ENABLE_AVX 17 #ifndef MINDSPORE_LITE_SRC_RUNTIME_KERNEL_ARM_FP32_CONVOLUTION_DEPTHWISE_SLIDEWINDOW_X86_FP32_H_ 18 #define MINDSPORE_LITE_SRC_RUNTIME_KERNEL_ARM_FP32_CONVOLUTION_DEPTHWISE_SLIDEWINDOW_X86_FP32_H_ 19 20 #include <vector> 21 #include "src/inner_kernel.h" 22 #include "src/runtime/kernel/arm/base/convolution_base.h" 23 #include "nnacl/fp32/conv_depthwise_fp32.h" 24 25 namespace mindspore::kernel { 26 class ConvolutionDepthwiseSWCPUKernelX86 : public ConvolutionBaseCPUKernel { 27 public: ConvolutionDepthwiseSWCPUKernelX86(OpParameter * parameter,const std::vector<lite::Tensor * > & inputs,const std::vector<lite::Tensor * > & outputs,const lite::InnerContext * ctx)28 ConvolutionDepthwiseSWCPUKernelX86(OpParameter *parameter, const std::vector<lite::Tensor *> &inputs, 29 const std::vector<lite::Tensor *> &outputs, const lite::InnerContext *ctx) 30 : ConvolutionBaseCPUKernel(parameter, inputs, outputs, ctx, inputs.at(kWeightIndex)->data(), 31 inputs.size() == kInputSize2 ? inputs.at(kBiasIndex)->data() : nullptr) {} 32 ~ConvolutionDepthwiseSWCPUKernelX86() override; 33 34 int Init() override; 35 int ReSize() override; 36 int Run() override; 37 38 int Execute(int task_id); 39 40 private: 41 void FreePackedInputOutput(); 42 int InitPackedInputOutput(); 43 int MallocWeightBiasData() override; 44 void PackWeight() override; 45 int oc_tile_ = C8NUM; // in x86 avx 46 SlidingWindowParam *sliding_ = nullptr; 47 float *packed_input_ = nullptr; 48 float *packed_output_ = nullptr; 49 bool input_need_align_ = false; 50 bool output_need_align_ = false; 51 }; 52 } // namespace mindspore::kernel 53 54 #endif 55 #endif // MINDSPORE_LITE_SRC_RUNTIME_KERNEL_ARM_FP32_CONVOLUTION_DEPTHWISE_SLIDEWINDOW_X86_FP32_H_ 56