【caffe Blob】caffe中与Blob相关的代码注释、使用举例

首先，Blob使用的小例子（通过运行结果即可知道相关功能）：

#include <vector>
#include <caffe/blob.hpp>
#include <caffe/util/io.hpp>//磁盘读写
#include <iostream>

using namespace std;
using namespace caffe;

int  main()
{
    Blob<float> a;
    cout<<"Size: "<<a.shape_string()<<endl;
    a.Reshape(,,,);
    cout<<"Size: "<<a.shape_string()<<endl;
    a.Reshape(,,,);
    cout<<"Size: "<<a.shape_string()<<endl;

    float* p=a.mutable_cpu_data();
    float* q=a.mutable_cpu_diff();
    for(int i=;i<a.count();i++)
    {
        p[i]=i;
        q[i]=a.count()--i;
    }
    cout<<"L1: "<<a.asum_data()<<endl;
    cout<<"L2: "<<a.sumsq_data()<<endl;
    //a.Update();

    //磁盘读写
    BlobProto bp;
    a.ToProto(&bp,true);//a序列化，连带diff（默认不带）
    WriteProtoToBinaryFile(bp,"a.blob");
    BlobProto bp2;
    ReadProtoFromBinaryFileOrDie("a.blob",&bp2);
    Blob<float> b;
    b.FromProto(bp2,true);//从序列化对象中克隆b（连同形状）
    b.Update();
    cout<<"L1: "<<b.asum_data()<<endl;
    cout<<"L2: "<<b.sumsq_data()<<endl;

    return ;
}

编译：

export LD_LIBRARY_PATH=./build/lib/:$LD_LIBRARY_PATH

g++ -o app ./bambootry/try.cpp -I ./include/ -D CPU_ONLY \
-I ./.build_release/src/ \
-L ./.build_release/lib/ -lcaffe -lglog -lboost_system

运行结果：

进入正题，代码注释

src/caffe/proto/caffe.proto中Blob部分

// Specifies the shape (dimensions) of a Blob.表示Blob每个维度的大小
message BlobShape {
  repeated int64 dim =  [packed = true];//packed表示这些值在内存中紧密排布没有空洞
}
//该结构描述Blob在磁盘中序列化的形态
message BlobProto {
  optional BlobShape shape = ; //可选，包括一个BlobShape对象
  repeated float data =  [packed = true]; //包括若干浮点元素，存储数据或权重，元素数目由shape或（num，channels，height，weight）决定
  repeated float diff =  [packed = true]; //包括若干浮点元素，用于存储增量信息，维度与data数组一致
  repeated double double_data =  [packed = true]; //data 类型double
  repeated double double_diff =  [packed = true]; //diff 类型double

  // 4D dimensions -- deprecated.  Use "shape" instead.可选，维度信息。新版本推荐使用shape
  optional int32 num =  [default = ];
  optional int32 channels =  [default = ];
  optional int32 height =  [default = ];
  optional int32 width =  [default = ];
}

include/caffe/blob.hpp （其中使用了SyncedMemory类，具体有关该类的内容在其他文件中） 作为基本计算单元服务 Layer Net Solver等

 #ifndef CAFFE_BLOB_HPP_
 #define CAFFE_BLOB_HPP_

 #include <algorithm>
 #include <string>
 #include <vector>

 #include "caffe/common.hpp"
 #include "caffe/proto/caffe.pb.h" //由protoc生成的头文件，声明了BlobProto、BlobShape等遵循caffe.proto协议的数据结构
 #include "caffe/syncedmem.hpp" //CPU和GPU共享内存类，用于数据同步

 const int kMaxBlobAxes = ; //Blob最大维数

 namespace caffe {

 /**
  * @brief A wrapper around SyncedMemory holders serving as the basic
  *        computational unit through which Layer%s, Net%s, and Solver%s
  *        interact.
  *
  * TODO(dox): more thorough description.
  */
 template <typename Dtype>
 class Blob {
  public:
   Blob()
        : data_(), diff_(), count_(), capacity_() {} //默认构造函数

   /// @brief Deprecated; use <code>Blob(const vector<int>& shape)</code>.
   explicit Blob(const int num, const int channels, const int height,
       const int width); //显式构造函数，避免隐式数据类型转换
   explicit Blob(const vector<int>& shape); 

   /// @brief Deprecated; use <code>Reshape(const vector<int>& shape)</code>.
   // 变形函数，根据输入参数重新设置当前Blob形状，必要时重新分配内存
   void Reshape(const int num, const int channels, const int height,
       const int width);
   /**
    * @brief Change the dimensions of the blob, allocating new memory if
    *        necessary.
    *
    * This function can be called both to create an initial allocation
    * of memory, and to adjust the dimensions of a top blob during Layer::Reshape
    * or Layer::Forward. When changing the size of blob, memory will only be
    * reallocated if sufficient memory does not already exist, and excess memory
    * will never be freed.
    *
    * Note that reshaping an input blob and immediately calling Net::Backward is
    * an error; either Net::Forward or Net::Reshape need to be called to
    * propagate the new input shape to higher layers.
    */
   void Reshape(const vector<int>& shape);
   void Reshape(const BlobShape& shape);
   void ReshapeLike(const Blob& other);
   //得到Blob形状字符串，用于打印log，见Caffe运行log
   //例如： Top Shape: 100 1 28 28 (78400)
   inline string shape_string() const {
     ostringstream stream;
     for (int i = ; i < shape_.size(); ++i) {
       stream << shape_[i] << " ";
     }
     stream << "(" << count_ << ")";
     return stream.str();
   }
   //返回blob形状
   inline const vector<int>& shape() const { return shape_; }
   /**
    * @brief Returns the dimension of the index-th axis (or the negative index-th
    *        axis from the end, if index is negative).
    *
    * @param index the axis index, which may be negative as it will be
    *        "canonicalized" using CanonicalAxisIndex.
    *        Dies on out of range index.
    */
    //返回某一维度尺寸
   inline int shape(int index) const {
     return shape_[CanonicalAxisIndex(index)];
   }
   //返回维度数目
   inline int num_axes() const { return shape_.size(); }
   //返回Blob中元素总数
   inline int count() const { return count_; }

   /**
    * @brief Compute the volume of a slice; i.e., the product of dimensions
    *        among a range of axes.
    *
    * @param start_axis The first axis to include in the slice.
    *
    * @param end_axis The first axis to exclude from the slice.
    */
    //返回Blob中某几个维度子集的元素总数
   inline int count(int start_axis, int end_axis) const {
     CHECK_LE(start_axis, end_axis);//保证start_sxis<=end_axis
     CHECK_GE(start_axis, );//保证start_sxis>=0
     CHECK_GE(end_axis, );//保证end_axis>=0
     CHECK_LE(start_axis, num_axes());//保证start_sxis小于总维度数目
     CHECK_LE(end_axis, num_axes());//保证end_axis小于总维度数目
     int count = ;
     for (int i = start_axis; i < end_axis; ++i) {
       count *= shape(i);
     }
     return count;
   }
   /**
    * @brief Compute the volume of a slice spanning from a particular first
    *        axis to the final axis.
    *
    * @param start_axis The first axis to include in the slice.
    */
    //计算从某一个维度开始的元素总数
   inline int count(int start_axis) const {
     return count(start_axis, num_axes());
   }

   /**
    * @brief Returns the 'canonical' version of a (usually) user-specified axis,
    *        allowing for negative indexing (e.g., -1 for the last axis).
    *
    * @param axis_index the axis index.
    *        If 0 <= index < num_axes(), return index.
    *        If -num_axes <= index <= -1, return (num_axes() - (-index)),
    *        e.g., the last axis index (num_axes() - 1) if index == -1,
    *        the second to last if index == -2, etc.
    *        Dies on out of range index.
    */
    //转换坐标轴索引
   inline int CanonicalAxisIndex(int axis_index) const {
     CHECK_GE(axis_index, -num_axes())//保证axis_index>=-num_axes()
         << "axis " << axis_index << " out of range for " << num_axes()
         << "-D Blob with shape " << shape_string();
     CHECK_LT(axis_index, num_axes())//保证axis_index<=num_axes()
         << "axis " << axis_index << " out of range for " << num_axes()
         << "-D Blob with shape " << shape_string();
     if (axis_index < ) {
       return axis_index + num_axes();//负索引号表示从后向前索引，例如-1为最后一个，即正索引号的N-1
     }
     return axis_index;
   }

   /// @brief Deprecated legacy shape accessor num: use shape(0) instead.
   inline int num() const { return LegacyShape(); }
   /// @brief Deprecated legacy shape accessor channels: use shape(1) instead.
   inline int channels() const { return LegacyShape(); }
   /// @brief Deprecated legacy shape accessor height: use shape(2) instead.
   inline int height() const { return LegacyShape(); }
   /// @brief Deprecated legacy shape accessor width: use shape(3) instead.
   inline int width() const { return LegacyShape(); }
   inline int LegacyShape(int index) const {
     CHECK_LE(num_axes(), )
         << "Cannot use legacy accessors on Blobs with > 4 axes.";
     CHECK_LT(index, );
     CHECK_GE(index, -);
     if (index >= num_axes() || index < -num_axes()) {
       // Axis is out of range, but still in [0, 3] (or [-4, -1] for reverse
       // indexing) -- this special case simulates the one-padding used to fill
       // extraneous axes of legacy blobs.
       return ;
     }
     return shape(index);
   }
   //offset函数用于计算偏移量
   inline int offset(const int n, const int c = , const int h = ,
       const int w = ) const {
     CHECK_GE(n, );
     CHECK_LE(n, num());
     CHECK_GE(channels(), );
     CHECK_LE(c, channels());
     CHECK_GE(height(), );
     CHECK_LE(h, height());
     CHECK_GE(width(), );
     CHECK_LE(w, width());
     return ((n * channels() + c) * height() + h) * width() + w;
   }

   inline int offset(const vector<int>& indices) const {
     CHECK_LE(indices.size(), num_axes());
     int offset = ;
     for (int i = ; i < num_axes(); ++i) {
       offset *= shape(i);
       if (indices.size() > i) {
         CHECK_GE(indices[i], );
         CHECK_LT(indices[i], shape(i));
         offset += indices[i];
       }
     }
     return offset;
   }
   /**
    * @brief Copy from a source Blob.
    *
    * @param source the Blob to copy from
    * @param copy_diff if false, copy the data; if true, copy the diff
    * @param reshape if false, require this Blob to be pre-shaped to the shape
    *        of other (and die otherwise); if true, Reshape this Blob to other's
    *        shape if necessary
    */
    //拷贝Blob到当前Blob
   void CopyFrom(const Blob<Dtype>& source, bool copy_diff = false,
       bool reshape = false);
   //下面为一系列存取器
   inline Dtype data_at(const int n, const int c, const int h,
       const int w) const {
     return cpu_data()[offset(n, c, h, w)];
   }

   inline Dtype diff_at(const int n, const int c, const int h,
       const int w) const {
     return cpu_diff()[offset(n, c, h, w)];
   }

   inline Dtype data_at(const vector<int>& index) const {
     return cpu_data()[offset(index)];
   }

   inline Dtype diff_at(const vector<int>& index) const {
     return cpu_diff()[offset(index)];
   }

   inline const shared_ptr<SyncedMemory>& data() const {
     CHECK(data_);
     return data_;
   }

   inline const shared_ptr<SyncedMemory>& diff() const {
     CHECK(diff_);
     return diff_;
   }

   const Dtype* cpu_data() const;//只读访问cpu data
   void set_cpu_data(Dtype* data);//设置cpu data
   const int* gpu_shape() const;
   const Dtype* gpu_data() const;//只读访问gpu data
   void set_gpu_data(Dtype* data);//设置gpu data
   const Dtype* cpu_diff() const;//只读访问cpu diff
   const Dtype* gpu_diff() const;//只读访问gpu diff
   Dtype* mutable_cpu_data();//读写访问
   Dtype* mutable_gpu_data();//读写访问
   Dtype* mutable_cpu_diff();//读写访问
   Dtype* mutable_gpu_diff();//读写访问
   void Update();//Blob更新运算
   void FromProto(const BlobProto& proto, bool reshape = true);//反序列化函数，从BlobProto中恢复一个Blob对象
   void ToProto(BlobProto* proto, bool write_diff = false) const;//序列化函数，将内存中的Blob对象保存到BlobProto中

   /// @brief Compute the sum of absolute values (L1 norm) of the data.
   Dtype asum_data() const;//计算data的L1范数，即绝对值求和
   /// @brief Compute the sum of absolute values (L1 norm) of the diff.
   Dtype asum_diff() const;//计算diff的L1范数，即绝对值求和
   /// @brief Compute the sum of squares (L2 norm squared) of the data.
   Dtype sumsq_data() const;//计算data的平方和，用于L2范数
   /// @brief Compute the sum of squares (L2 norm squared) of the diff.
   Dtype sumsq_diff() const;//计算diff的平方和，用于L2范数

   /// @brief Scale the blob data by a constant factor.
   void scale_data(Dtype scale_factor);//data乘一个标量
   /// @brief Scale the blob diff by a constant factor.
   void scale_diff(Dtype scale_factor);//diff乘一个标量

   /**
    * @brief Set the data_ shared_ptr to point to the SyncedMemory holding the
    *        data_ of Blob other -- useful in Layer%s which simply perform a copy
    *        in their Forward pass.
    *
    * This deallocates the SyncedMemory holding this Blob's data_, as
    * shared_ptr calls its destructor when reset with the "=" operator.
    */
   void ShareData(const Blob& other);
   /**
    * @brief Set the diff_ shared_ptr to point to the SyncedMemory holding the
    *        diff_ of Blob other -- useful in Layer%s which simply perform a copy
    *        in their Forward pass.
    *
    * This deallocates the SyncedMemory holding this Blob's diff_, as
    * shared_ptr calls its destructor when reset with the "=" operator.
    */
   void ShareDiff(const Blob& other);//共享另一个Blob的diff_

   bool ShapeEquals(const BlobProto& other);

  protected:
   shared_ptr<SyncedMemory> data_;//存放指向data的指针
   shared_ptr<SyncedMemory> diff_;//存放指向diff的指针
   shared_ptr<SyncedMemory> shape_data_;
   vector<int> shape_;
   int count_;//存放有效元素数目信息
   int capacity_;//存放Blob容器的容量信息

   DISABLE_COPY_AND_ASSIGN(Blob);//禁止拷贝构造函数、赋值运算符重载
 };  // class Blob

 }  // namespace caffe

 #endif  // CAFFE_BLOB_HPP_

include/caffe/syncedmem.hpp

 #ifndef CAFFE_SYNCEDMEM_HPP_
 #define CAFFE_SYNCEDMEM_HPP_

 #include <cstdlib>

 #ifdef USE_MKL
   #include "mkl.h"
 #endif

 #include "caffe/common.hpp"

 namespace caffe {

 // If CUDA is available and in GPU mode, host memory will be allocated pinned,
 // using cudaMallocHost. It avoids dynamic pinning for transfers (DMA).
 // The improvement in performance seems negligible in the single GPU case,
 // but might be more significant for parallel training. Most importantly,
 // it improved stability for large models on many GPUs.
 // 如果是GPU模式且CUDA使能，则主机内存会以页锁定内存方式分配（使用cudaMallocHost（）函数）
 // 对单GPU提升并不明显，对多GPU提升非常明显
 inline void CaffeMallocHost(void** ptr, size_t size, bool* use_cuda) {
 #ifndef CPU_ONLY
   if (Caffe::mode() == Caffe::GPU) {
     CUDA_CHECK(cudaMallocHost(ptr, size));
     *use_cuda = true;
     return;
   }
 #endif
 #ifdef USE_MKL
   *ptr = mkl_malloc(size ? size:, );
 #else
   *ptr = malloc(size);
 #endif
   *use_cuda = false;
   CHECK(*ptr) << "host allocation of size " << size << " failed";
 }
 // 与CaffeMallocHost对应
 inline void CaffeFreeHost(void* ptr, bool use_cuda) {
 #ifndef CPU_ONLY
   if (use_cuda) {
     CUDA_CHECK(cudaFreeHost(ptr));
     return;
   }
 #endif
 #ifdef USE_MKL
   mkl_free(ptr);
 #else
   free(ptr);
 #endif
 }

 /**
  * @brief Manages memory allocation and synchronization between the host (CPU)
  *        and device (GPU).
  *
  * TODO(dox): more thorough description.
  */
  // 该类负责存储分配以及主机设备间同步
 class SyncedMemory {
  public:
   SyncedMemory();
   explicit SyncedMemory(size_t size);
   ~SyncedMemory();
   const void* cpu_data();       //只读
   void set_cpu_data(void* data);//设置
   const void* gpu_data();       //只读
   void set_gpu_data(void* data);//设置
   void* mutable_cpu_data();     //读写
   void* mutable_gpu_data();     //读写
   //状态机变量，4种状态：未初始化 CPU数据有效 GPU数据有效 已同步
   enum SyncedHead { UNINITIALIZED, HEAD_AT_CPU, HEAD_AT_GPU, SYNCED }；
   SyncedHead head() { return head_; }//获取当前状态机变量值
   size_t size() { return size_; }    //获取当前存储空间尺寸

 #ifndef CPU_ONLY
   void async_gpu_push(const cudaStream_t& stream);
 #endif

  private:
   void check_device();

   void to_cpu();  //数据同步到CPU
   void to_gpu();  //数据同步到GPU
   void* cpu_ptr_; //位于CPU的数据指针
   void* gpu_ptr_; //位于GPU的数据指针
   size_t size_;   //存储空间大小
   SyncedHead head_;  //状态机变量
   bool own_cpu_data_;//标志是否拥有CPU数据权（否，即从别的对象共享）
   bool cpu_malloc_use_cuda_;
   bool own_gpu_data_;//标志是否拥有GPU数据权
   int device_;       

   DISABLE_COPY_AND_ASSIGN(SyncedMemory);
 };  // class SyncedMemory

 }  // namespace caffe

 #endif  // CAFFE_SYNCEDMEM_HPP_

src/caffe/blob.cpp

 #include <climits>
 #include <vector>

 #include "caffe/blob.hpp"
 #include "caffe/common.hpp"
 #include "caffe/syncedmem.hpp"
 #include "caffe/util/math_functions.hpp"

 namespace caffe {
 //变维函数，将（num,channels,height,width）转为vector<int>并调用重载函数
 template <typename Dtype>
 void Blob<Dtype>::Reshape(const int num, const int channels, const int height,
     const int width) {
   vector<int> shape();
   shape[] = num;
   shape[] = channels;
   shape[] = height;
   shape[] = width;
   Reshape(shape);
 }
 //真正的变维函数
 template <typename Dtype>
 void Blob<Dtype>::Reshape(const vector<int>& shape) {
   CHECK_LE(shape.size(), kMaxBlobAxes);//保证vector维度小于kMaxBlobAxes
   count_ = ;//用于计算元素总数=num*channels*height*width
   shape_.resize(shape.size());//成员变量维度重置
   if (!shape_data_ || shape_data_->size() < shape.size() * sizeof(int)) {
     shape_data_.reset(new SyncedMemory(shape.size() * sizeof(int)));
   }
   int* shape_data = static_cast<int*>(shape_data_->mutable_cpu_data());
   for (int i = ; i < shape.size(); ++i) {
     CHECK_GE(shape[i], );
     if (count_ != ) {//保证count不溢出
       CHECK_LE(shape[i], INT_MAX / count_) << "blob size exceeds INT_MAX";
     }
     count_ *= shape[i];
     shape_[i] = shape[i];//成员变量赋值
     shape_data[i] = shape[i];
   }
   if (count_ > capacity_) {//如果新的count_大于当前分配的空间容量
     capacity_ = count_;    //扩容，重新分配data_和diff_空间
     data_.reset(new SyncedMemory(capacity_ * sizeof(Dtype)));
     diff_.reset(new SyncedMemory(capacity_ * sizeof(Dtype)));
   }
 }

 template <typename Dtype>
 void Blob<Dtype>::Reshape(const BlobShape& shape) {
   CHECK_LE(shape.dim_size(), kMaxBlobAxes);
   vector<int> shape_vec(shape.dim_size());
   for (int i = ; i < shape.dim_size(); ++i) {
     shape_vec[i] = shape.dim(i);
   }
   Reshape(shape_vec);
 }

 template <typename Dtype>
 void Blob<Dtype>::ReshapeLike(const Blob<Dtype>& other) {
   Reshape(other.shape());
 }
 //构造函数
 template <typename Dtype>
 Blob<Dtype>::Blob(const int num, const int channels, const int height,
     const int width)
   // capacity_ must be initialized before calling Reshape
   : capacity_() {
   //调用Reshape之前必须初始化capacity，否则会导致不可预期的后果
   Reshape(num, channels, height, width);
 }

 template <typename Dtype>
 Blob<Dtype>::Blob(const vector<int>& shape)
   // capacity_ must be initialized before calling Reshape
   : capacity_() {
   Reshape(shape);
 }

 template <typename Dtype>
 const int* Blob<Dtype>::gpu_shape() const {
   CHECK(shape_data_);
   return (const int*)shape_data_->gpu_data();
 }
 //只读获得cpu data指针
 template <typename Dtype>
 const Dtype* Blob<Dtype>::cpu_data() const {
   CHECK(data_);//保证data_不为空
   return (const Dtype*)data_->cpu_data();
 }
 //修改cpu data指针
 template <typename Dtype>
 void Blob<Dtype>::set_cpu_data(Dtype* data) {
   CHECK(data);
   // Make sure CPU and GPU sizes remain equal
   size_t size = count_ * sizeof(Dtype);
   if (data_->size() != size) {
     data_.reset(new SyncedMemory(size));
     diff_.reset(new SyncedMemory(size));
   }
   data_->set_cpu_data(data);//设置成员变量值为传入参数值
 }
 //只读获得gpu data指针
 template <typename Dtype>
 const Dtype* Blob<Dtype>::gpu_data() const {
   CHECK(data_);//保证不为空
   return (const Dtype*)data_->gpu_data();
 }
 //修改gpu data指针
 template <typename Dtype>
 void Blob<Dtype>::set_gpu_data(Dtype* data) {
   CHECK(data);
   // Make sure CPU and GPU sizes remain equal
   size_t size = count_ * sizeof(Dtype);
   if (data_->size() != size) {
     data_.reset(new SyncedMemory(size));
     diff_.reset(new SyncedMemory(size));
   }
   data_->set_gpu_data(data);
 }
 //只读获得cpu diff指针
 template <typename Dtype>
 const Dtype* Blob<Dtype>::cpu_diff() const {
   CHECK(diff_);
   return (const Dtype*)diff_->cpu_data();
 }
 //只读获得gpu diff指针
 template <typename Dtype>
 const Dtype* Blob<Dtype>::gpu_diff() const {
   CHECK(diff_);
   return (const Dtype*)diff_->gpu_data();
 }
 //读写访问cpu_data
 template <typename Dtype>
 Dtype* Blob<Dtype>::mutable_cpu_data() {
   CHECK(data_);
   return static_cast<Dtype*>(data_->mutable_cpu_data());
 }
 //读写访问gpu_data
 template <typename Dtype>
 Dtype* Blob<Dtype>::mutable_gpu_data() {
   CHECK(data_);
   return static_cast<Dtype*>(data_->mutable_gpu_data());
 }
 //读写访问cpu_diff
 template <typename Dtype>
 Dtype* Blob<Dtype>::mutable_cpu_diff() {
   CHECK(diff_);
   return static_cast<Dtype*>(diff_->mutable_cpu_data());
 }
 //读写访问gpu_diff
 template <typename Dtype>
 Dtype* Blob<Dtype>::mutable_gpu_diff() {
   CHECK(diff_);
   return static_cast<Dtype*>(diff_->mutable_gpu_data());
 }
 //共享另一个Blob的data指针
 template <typename Dtype>
 void Blob<Dtype>::ShareData(const Blob& other) {
   CHECK_EQ(count_, other.count());
   data_ = other.data();
 }
 //共享另一个Blob的diff指针
 template <typename Dtype>
 void Blob<Dtype>::ShareDiff(const Blob& other) {
   CHECK_EQ(count_, other.count());
   diff_ = other.diff();
 }

 // The "update" method is used for parameter blobs in a Net, which are stored
 // as Blob<float> or Blob<double> -- hence we do not define it for
 // Blob<int> or Blob<unsigned int>.
 //更新函数用于网络参数Blob的更新。其中int和unsigned int类型的处理并未实现
 //实现的类型为Blob<float>和Blob<double>
 template <> void Blob<unsigned int>::Update() { NOT_IMPLEMENTED; }
 template <> void Blob<int>::Update() { NOT_IMPLEMENTED; }

 template <typename Dtype>
 void Blob<Dtype>::Update() {
   // We will perform update based on where the data is located.
   //data在哪儿就在哪儿更新
   switch (data_->head()) {
   case SyncedMemory::HEAD_AT_CPU:
     // perform computation on CPU
     //执行CPU运算，data[i]=data[i]-diff[i]，i=0~count-1
     caffe_axpy<Dtype>(count_, Dtype(-),
         static_cast<const Dtype*>(diff_->cpu_data()),
         static_cast<Dtype*>(data_->mutable_cpu_data()));
     break;
   case SyncedMemory::HEAD_AT_GPU://data位于GPU或CPU/GPU已同步
   case SyncedMemory::SYNCED:
 #ifndef CPU_ONLY
     // perform computation on GPU
     //执行GPU运算,data[i]=data[i]-diff[i]，i=0~count-1
     caffe_gpu_axpy<Dtype>(count_, Dtype(-),
         static_cast<const Dtype*>(diff_->gpu_data()),
         static_cast<Dtype*>(data_->mutable_gpu_data()));
 #else
     NO_GPU;//编译时如果打开CPU_ONLY，则GPU禁用
 #endif
     break;
   default:
     LOG(FATAL) << "Syncedmem not initialized.";
   }
 }

 template <> unsigned int Blob<unsigned int>::asum_data() const {
   NOT_IMPLEMENTED;
   return ;
 }

 template <> int Blob<int>::asum_data() const {
   NOT_IMPLEMENTED;
   return ;
 }
 //计算L1范数，即绝对值和
 template <typename Dtype>
 Dtype Blob<Dtype>::asum_data() const {
   if (!data_) { return ; }
   switch (data_->head()) {
   case SyncedMemory::HEAD_AT_CPU:
     return caffe_cpu_asum(count_, cpu_data());//执行CPU上的asum计算
   case SyncedMemory::HEAD_AT_GPU:
   case SyncedMemory::SYNCED:
 #ifndef CPU_ONLY
   {
     Dtype asum;
     caffe_gpu_asum(count_, gpu_data(), &asum);//执行GPU上的asum计算
     return asum;
   }
 #else
     NO_GPU;
 #endif
   case SyncedMemory::UNINITIALIZED:
     return ;
   default:
     LOG(FATAL) << "Unknown SyncedMemory head state: " << data_->head();
   }
   return ;
 }

 template <> unsigned int Blob<unsigned int>::asum_diff() const {
   NOT_IMPLEMENTED;
   return ;
 }

 template <> int Blob<int>::asum_diff() const {
   NOT_IMPLEMENTED;
   return ;
 }

 template <typename Dtype>
 Dtype Blob<Dtype>::asum_diff() const {
   if (!diff_) { return ; }
   switch (diff_->head()) {
   case SyncedMemory::HEAD_AT_CPU:
     return caffe_cpu_asum(count_, cpu_diff());
   case SyncedMemory::HEAD_AT_GPU:
   case SyncedMemory::SYNCED:
 #ifndef CPU_ONLY
   {
     Dtype asum;
     caffe_gpu_asum(count_, gpu_diff(), &asum);
     return asum;
   }
 #else
     NO_GPU;
 #endif
   case SyncedMemory::UNINITIALIZED:
     return ;
   default:
     LOG(FATAL) << "Unknown SyncedMemory head state: " << diff_->head();
   }
   return ;
 }

 template <> unsigned int Blob<unsigned int>::sumsq_data() const {
   NOT_IMPLEMENTED;
   return ;
 }

 template <> int Blob<int>::sumsq_data() const {
   NOT_IMPLEMENTED;
   return ;
 }
 //用于L2范数，平方和
 template <typename Dtype>
 Dtype Blob<Dtype>::sumsq_data() const {
   Dtype sumsq;
   const Dtype* data;
   if (!data_) { return ; }
   switch (data_->head()) {
   case SyncedMemory::HEAD_AT_CPU:
     data = cpu_data();
     sumsq = caffe_cpu_dot(count_, data, data);
     break;
   case SyncedMemory::HEAD_AT_GPU:
   case SyncedMemory::SYNCED:
 #ifndef CPU_ONLY
     data = gpu_data();
     caffe_gpu_dot(count_, data, data, &sumsq);
 #else
     NO_GPU;
 #endif
     break;
   case SyncedMemory::UNINITIALIZED:
     return ;
   default:
     LOG(FATAL) << "Unknown SyncedMemory head state: " << data_->head();
   }
   return sumsq;
 }

 template <> unsigned int Blob<unsigned int>::sumsq_diff() const {
   NOT_IMPLEMENTED;
   return ;
 }

 template <> int Blob<int>::sumsq_diff() const {
   NOT_IMPLEMENTED;
   return ;
 }

 template <typename Dtype>
 Dtype Blob<Dtype>::sumsq_diff() const {
   Dtype sumsq;
   const Dtype* diff;
   if (!diff_) { return ; }
   switch (diff_->head()) {
   case SyncedMemory::HEAD_AT_CPU:
     diff = cpu_diff();
     sumsq = caffe_cpu_dot(count_, diff, diff);
     break;
   case SyncedMemory::HEAD_AT_GPU:
   case SyncedMemory::SYNCED:
 #ifndef CPU_ONLY
     diff = gpu_diff();
     caffe_gpu_dot(count_, diff, diff, &sumsq);
     break;
 #else
     NO_GPU;
 #endif
   case SyncedMemory::UNINITIALIZED:
     return ;
   default:
     LOG(FATAL) << "Unknown SyncedMemory head state: " << data_->head();
   }
   return sumsq;
 }

 template <> void Blob<unsigned int>::scale_data(unsigned int scale_factor) {
   NOT_IMPLEMENTED;
 }

 template <> void Blob<int>::scale_data(int scale_factor) {
   NOT_IMPLEMENTED;
 }
 //对data_进行幅度缩放
 template <typename Dtype>
 void Blob<Dtype>::scale_data(Dtype scale_factor) {
   Dtype* data;
   if (!data_) { return; }
   switch (data_->head()) {
   case SyncedMemory::HEAD_AT_CPU:
     data = mutable_cpu_data();
     caffe_scal(count_, scale_factor, data);
     return;
   case SyncedMemory::HEAD_AT_GPU:
   case SyncedMemory::SYNCED:
 #ifndef CPU_ONLY
     data = mutable_gpu_data();
     caffe_gpu_scal(count_, scale_factor, data);
     return;
 #else
     NO_GPU;
 #endif
   case SyncedMemory::UNINITIALIZED:
     return;
   default:
     LOG(FATAL) << "Unknown SyncedMemory head state: " << data_->head();
   }
 }

 template <> void Blob<unsigned int>::scale_diff(unsigned int scale_factor) {
   NOT_IMPLEMENTED;
 }

 template <> void Blob<int>::scale_diff(int scale_factor) {
   NOT_IMPLEMENTED;
 }

 template <typename Dtype>
 void Blob<Dtype>::scale_diff(Dtype scale_factor) {
   Dtype* diff;
   if (!diff_) { return; }
   switch (diff_->head()) {
   case SyncedMemory::HEAD_AT_CPU:
     diff = mutable_cpu_diff();
     caffe_scal(count_, scale_factor, diff);
     return;
   case SyncedMemory::HEAD_AT_GPU:
   case SyncedMemory::SYNCED:
 #ifndef CPU_ONLY
     diff = mutable_gpu_diff();
     caffe_gpu_scal(count_, scale_factor, diff);
     return;
 #else
     NO_GPU;
 #endif
   case SyncedMemory::UNINITIALIZED:
     return;
   default:
     LOG(FATAL) << "Unknown SyncedMemory head state: " << diff_->head();
   }
 }
 //判断形状是否相同
 template <typename Dtype>
 bool Blob<Dtype>::ShapeEquals(const BlobProto& other) {
   if (other.has_num() || other.has_channels() ||
       other.has_height() || other.has_width()) {
     // Using deprecated 4D Blob dimensions --每个维度对比
     // shape is (num, channels, height, width).
     // Note: we do not use the normal Blob::num(), Blob::channels(), etc.
     // methods as these index from the beginning of the blob shape, where legacy
     // parameter blobs were indexed from the end of the blob shape (e.g., bias
     // Blob shape (1 x 1 x 1 x N), IP layer weight Blob shape (1 x 1 x M x N)).
     return shape_.size() <=  &&
            LegacyShape(-) == other.num() &&
            LegacyShape(-) == other.channels() &&
            LegacyShape(-) == other.height() &&
            LegacyShape(-) == other.width();
   }
   //直接对比
   vector<int> other_shape(other.shape().dim_size());
   for (int i = ; i < other.shape().dim_size(); ++i) {
     other_shape[i] = other.shape().dim(i);
   }
   return shape_ == other_shape;
 }
 //从另一个Blob拷贝data（可选diff），必要时进行变维
 template <typename Dtype>
 void Blob<Dtype>::CopyFrom(const Blob& source, bool copy_diff, bool reshape) {
   if (source.count() != count_ || source.shape() != shape_) {
     if (reshape) {
       ReshapeLike(source);
     } else {
       LOG(FATAL) << "Trying to copy blobs of different sizes.";
     }
   }
   switch (Caffe::mode()) {
   case Caffe::GPU:
     if (copy_diff) {
       caffe_copy(count_, source.gpu_diff(),
           static_cast<Dtype*>(diff_->mutable_gpu_data()));
     } else {
       caffe_copy(count_, source.gpu_data(),
           static_cast<Dtype*>(data_->mutable_gpu_data()));
     }
     break;
   case Caffe::CPU:
     if (copy_diff) {
       caffe_copy(count_, source.cpu_diff(),
           static_cast<Dtype*>(diff_->mutable_cpu_data()));
     } else {
       caffe_copy(count_, source.cpu_data(),
           static_cast<Dtype*>(data_->mutable_cpu_data()));
     }
     break;
   default:
     LOG(FATAL) << "Unknown caffe mode.";
   }
 }
 //从BlobProto中加载一个Blob
 template <typename Dtype>
 void Blob<Dtype>::FromProto(const BlobProto& proto, bool reshape) {
   if (reshape) {//获取维度信息
     vector<int> shape;
     if (proto.has_num() || proto.has_channels() ||
         proto.has_height() || proto.has_width()) {
       // Using deprecated 4D Blob dimensions --
       // shape is (num, channels, height, width).
       shape.resize();
       shape[] = proto.num();
       shape[] = proto.channels();
       shape[] = proto.height();
       shape[] = proto.width();
     } else {
       shape.resize(proto.shape().dim_size());
       for (int i = ; i < proto.shape().dim_size(); ++i) {
         shape[i] = proto.shape().dim(i);
       }
     }
     Reshape(shape);//变维
   } else {
     CHECK(ShapeEquals(proto)) << "shape mismatch (reshape not set)";
   }
   // copy data 加载数据
   Dtype* data_vec = mutable_cpu_data();
   if (proto.double_data_size() > ) {//若之前保存的是double类型的data
     CHECK_EQ(count_, proto.double_data_size());
     for (int i = ; i < count_; ++i) {
       data_vec[i] = proto.double_data(i);//加载double data
     }
   } else {
     CHECK_EQ(count_, proto.data_size());
     for (int i = ; i < count_; ++i) {
       data_vec[i] = proto.data(i);//否则加载float data
     }
   }
   if (proto.double_diff_size() > ) {//若之前保存的是double类型的diff
     CHECK_EQ(count_, proto.double_diff_size());
     Dtype* diff_vec = mutable_cpu_diff();
     for (int i = ; i < count_; ++i) {
       diff_vec[i] = proto.double_diff(i);//加载double diff
     }
   } else if (proto.diff_size() > ) {
     CHECK_EQ(count_, proto.diff_size());
     Dtype* diff_vec = mutable_cpu_diff();
     for (int i = ; i < count_; ++i) {
       diff_vec[i] = proto.diff(i);//否则加载float diff
     }
   }
 }

 template <>
 void Blob<double>::ToProto(BlobProto* proto, bool write_diff) const {
   proto->clear_shape();
   for (int i = ; i < shape_.size(); ++i) {
     proto->mutable_shape()->add_dim(shape_[i]);
   }
   proto->clear_double_data();
   proto->clear_double_diff();
   const double* data_vec = cpu_data();
   for (int i = ; i < count_; ++i) {
     proto->add_double_data(data_vec[i]);
   }
   if (write_diff) {
     const double* diff_vec = cpu_diff();
     for (int i = ; i < count_; ++i) {
       proto->add_double_diff(diff_vec[i]);
     }
   }
 }
 //将Blob中的data（可选diff）导出到BlobProto结构体，便于存储到磁盘文件中
 template <>
 void Blob<float>::ToProto(BlobProto* proto, bool write_diff) const {
   proto->clear_shape();//重置proto维度，保证与Blob相同
   for (int i = ; i < shape_.size(); ++i) {
     proto->mutable_shape()->add_dim(shape_[i]);
   }
   proto->clear_data();//清楚data
   proto->clear_diff();//清楚diff
   const float* data_vec = cpu_data();//将data导出到proto
   for (int i = ; i < count_; ++i) {
     proto->add_data(data_vec[i]);
   }
   if (write_diff) {//如果要求导出diff
     const float* diff_vec = cpu_diff();//导出diff到proto
     for (int i = ; i < count_; ++i) {
       proto->add_diff(diff_vec[i]);
     }
   }
 }

 INSTANTIATE_CLASS(Blob);//实例化Blob类模板（float,double ）
 template class Blob<int>;
 template class Blob<unsigned int>;

 }  // namespace caffe

初学者，慢慢看，待更新……

摘抄参考赵永科《深度学习 21天实战caffe》