Files
alibaba--mnn/source/backend/metal/MetalConvolutionCommon.hpp
2026-07-13 13:33:03 +08:00

68 lines
1.9 KiB
C++

//
// MetalConvolutionCommon.hpp
// MNN
//
// Created by MNN on 2019/02/25.
// Copyright © 2018, Alibaba Group Holding Limited
//
#ifndef MetalConvolutionCommon_hpp
#define MetalConvolutionCommon_hpp
#import "core/ConvolutionCommon.hpp"
#import "MetalBackend.hpp"
#import "MetalExecution.hpp"
#import "MNNMetalContext.h"
#if MNN_METAL_ENABLED
namespace MNN {
class MetalConvolutionCommon : public MetalExecution {
public:
MetalConvolutionCommon(Backend *backend, const MNN::Op *op, std::shared_ptr<MNN::Tensor> bias);
virtual ~MetalConvolutionCommon() = default;
protected:
void loadWeight(const MNN::Op *op, bool loadWeightInt8 = false);
virtual std::shared_ptr<MNN::Tensor> weightTransform(int group, int oc, int ic, int kh, int kw, const float *src, bool int8Weight = false, bool int4Weight = false, id<MTLBuffer> srcGpuBuffer = nil, int subBits = 0);
private:
template <typename FType, typename TType>
void convertWeightFormat(int group, int oc, int ic, int kh, int kw, const FType *src, TType* dstOrigion, Tensor* dstTensor, id<MTLBuffer> srcGpuBuffer);
protected:
struct Param {
int input_size;
int input_slice;
int output_width;
int output_height;
int output_size;
int output_slice;
int output_channel;
int batch;
int block_size;
int activation;
float scale_coef;
};
protected:
int mKernelX = 0;
int mKernelY = 0;
int mStrideX = 0;
int mStrideY = 0;
int mDilateX = 0;
int mDilateY = 0;
int mActivationType = 0;
const MNN::Op *mOp = nullptr;
std::shared_ptr<MNN::Tensor> mWeight;
std::shared_ptr<MNN::Tensor> mBias;
std::shared_ptr<MNN::Tensor> mDequantScaleBias;
int mDequantBits;
float mScaleCoef;
id<MTLBuffer> mConstBuffer = nil;
};
} // namespace MNN
#endif /* MNN_METAL_ENABLED */
#endif /* MetalConvolutionCommon_hpp */