caffe代码阅读2：common的实现细节-2016.3.14

前言

common中给出的是一些初始化的内容，其中包括随机数生成器的内容以及google的gflags和glog的初始化，其中最主要的还是随机数生成器的内容。

重点

这里有点绕，特别是Caffe类里面有个RNG，RNG这个类里面还有个Generator类

在RNG里面会用到Caffe里面的Get()函数来获取一个新的Caffe类的实例（如果不存在的话）。

然后RNG里面用到了Generator。Generator是实际产生随机数的。

（1）Generator类

该类有两个构造函数：

Generator()//用系统的熵池或者时间来初始化随机数

explicit Generator(unsigned int seed)// 用给定的种子初始化

（2）RNG类

RNG类内部有generator_，generator_是Generator类的实例

该类有三个构造函数：

RNG(); //利用系统的熵池或者时间来初始化RNG内部的generator_

explicit RNG(unsigned int seed); // 利用给定的seed来初始化内部的generator_

explicit RNG(const RNG&);// 用其他的RNG内部的generator_设置为当前的generator_

（3）Caffe类

1)含有一个Get函数，该函数利用Boost的局部线程存储功能实现

// Make sure each thread can have different values.

// boost::thread_specific_ptr是线程局部存储机制

// 一开始的值是NULL

static boost::thread_specific_ptr<Caffe> thread_instance_;

Caffe& Caffe::Get() {

  if (!thread_instance_.get()) {// 如果当前线程没有caffe实例

    thread_instance_.reset(new Caffe());// 则新建一个caffe的实例并返回

  }

  return *(thread_instance_.get());

2)此外该类还有

SetDevice

DeviceQuery

mode

set_mode

set_random_seed

solver_count

set_solver_count

root_solver

set_root_solver

等成员函数

3)内部还有一些比较技巧性的东西比如：

// CUDA: various checks for different function calls.

//  防止重定义cudaError_t，这个以前在linux代码里面看过

// 实际上就是利用变量的局部声明

#define CUDA_CHECK(condition) \

  /* Code block avoids redefinition of cudaError_t error */ \

  do { \

    cudaError_t error = condition; \

    CHECK_EQ(error, cudaSuccess) << " " << cudaGetErrorString(error); \

  } while (0)

代码注释

下面给出common.h的详细注释：

#ifndef CAFFE_COMMON_HPP_
#define CAFFE_COMMON_HPP_

#include <boost/shared_ptr.hpp>
#include <gflags/gflags.h>
#include <glog/logging.h>

#include <climits>
#include <cmath>
#include <fstream>  // NOLINT(readability/streams)
#include <iostream>  // NOLINT(readability/streams)
#include <map>
#include <set>
#include <sstream>
#include <string>
#include <utility>  // pair
#include <vector>

#include "caffe/util/device_alternate.hpp"

// Convert macro to string
// 将宏转换为字符串
#define STRINGIFY(m) #m
#define AS_STRING(m) STRINGIFY(m)

// gflags 2.1 issue: namespace google was changed to gflags without warning.
// Luckily we will be able to use GFLAGS_GFLAGS_H_ to detect if it is version
// 2.1. If yes, we will add a temporary solution to redirect the namespace.
// TODO(Yangqing): Once gflags solves the problem in a more elegant way, let's
// remove the following hack.
// 检测gflags2.1
#ifndef GFLAGS_GFLAGS_H_
namespace gflags = google;
#endif  // GFLAGS_GFLAGS_H_

// Disable the copy and assignment operator for a class.
// 禁止某个类通过构造函数直接初始化另一个类
// 禁止某个类通过赋值来初始化另一个类
#define DISABLE_COPY_AND_ASSIGN(classname) \
private:\
  classname(const classname&);\
  classname& operator=(const classname&)

// Instantiate a class with float and double specifications.
#define INSTANTIATE_CLASS(classname) \
  char gInstantiationGuard##classname; \
  template class classname<float>; \
  template class classname<double>

// 初始化GPU的前向传播函数
#define INSTANTIATE_LAYER_GPU_FORWARD(classname) \
  template void classname<float>::Forward_gpu( \
      const std::vector<Blob<float>*>& bottom, \
      const std::vector<Blob<float>*>& top); \
  template void classname<double>::Forward_gpu( \
      const std::vector<Blob<double>*>& bottom, \
      const std::vector<Blob<double>*>& top);

// 初始化GPU的反向传播函数
#define INSTANTIATE_LAYER_GPU_BACKWARD(classname) \
  template void classname<float>::Backward_gpu( \
      const std::vector<Blob<float>*>& top, \
      const std::vector<bool>& propagate_down, \
      const std::vector<Blob<float>*>& bottom); \
  template void classname<double>::Backward_gpu( \
      const std::vector<Blob<double>*>& top, \
      const std::vector<bool>& propagate_down, \
      const std::vector<Blob<double>*>& bottom)

// 初始化GPU的前向反向传播函数
#define INSTANTIATE_LAYER_GPU_FUNCS(classname) \
  INSTANTIATE_LAYER_GPU_FORWARD(classname); \
  INSTANTIATE_LAYER_GPU_BACKWARD(classname)

// A simple macro to mark codes that are not implemented, so that when the code
// is executed we will see a fatal log.
// NOT_IMPLEMENTED实际上调用的LOG(FATAL) << "Not Implemented Yet"
#define NOT_IMPLEMENTED LOG(FATAL) << "Not Implemented Yet"

// See PR #1236
namespace cv { class Mat; }

namespace caffe {

// We will use the boost shared_ptr instead of the new C++11 one mainly
// because cuda does not work (at least now) well with C++11 features.
using boost::shared_ptr;

// Common functions and classes from std that caffe often uses.
using std::fstream;
using std::ios;
//using std::isnan;//vc++的编译器不支持这两个函数
//using std::isinf;
using std::iterator;
using std::make_pair;
using std::map;
using std::ostringstream;
using std::pair;
using std::set;
using std::string;
using std::stringstream;
using std::vector;

// A global initialization function that you should call in your main function.
// Currently it initializes google flags and google logging.
void GlobalInit(int* pargc, char*** pargv);

// A singleton class to hold common caffe stuff, such as the handler that
// caffe is going to use for cublas, curand, etc.
class Caffe {
 public:
  ~Caffe();

  // Thread local context for Caffe. Moved to common.cpp instead of
  // including boost/thread.hpp to avoid a boost/NVCC issues (#1009, #1010)
  // on OSX. Also fails on Linux with CUDA 7.0.18.
  static Caffe& Get();

  enum Brew { CPU, GPU };

  // This random number generator facade hides boost and CUDA rng
  // implementation from one another (for cross-platform compatibility).
  class RNG {
   public:
    RNG();
    explicit RNG(unsigned int seed);
    explicit RNG(const RNG&);
    RNG& operator=(const RNG&);
    void* generator();
   private:
    class Generator;
    shared_ptr<Generator> generator_;
  };

  // Getters for boost rng, curand, and cublas handles
  inline static RNG& rng_stream() {
    if (!Get().random_generator_) {
      Get().random_generator_.reset(new RNG());
    }
    return *(Get().random_generator_);
  }
#ifndef CPU_ONLY// GPU
  inline static cublasHandle_t cublas_handle() { return Get().cublas_handle_; }// cublas的句柄
  inline static curandGenerator_t curand_generator() {//curandGenerator句柄
    return Get().curand_generator_;
  }
#endif

  // Returns the mode: running on CPU or GPU.
  inline static Brew mode() { return Get().mode_; }
  // The setters for the variables
  // Sets the mode. It is recommended that you don't change the mode halfway
  // into the program since that may cause allocation of pinned memory being
  // freed in a non-pinned way, which may cause problems - I haven't verified
  // it personally but better to note it here in the header file.
  inline static void set_mode(Brew mode) { Get().mode_ = mode; }
  // Sets the random seed of both boost and curand
  static void set_random_seed(const unsigned int seed);
  // Sets the device. Since we have cublas and curand stuff, set device also
  // requires us to reset those values.
  static void SetDevice(const int device_id);
  // Prints the current GPU status.
  static void DeviceQuery();
  // Parallel training info
  inline static int solver_count() { return Get().solver_count_; }
  inline static void set_solver_count(int val) { Get().solver_count_ = val; }
  inline static bool root_solver() { return Get().root_solver_; }
  inline static void set_root_solver(bool val) { Get().root_solver_ = val; }

 protected:
#ifndef CPU_ONLY
  cublasHandle_t cublas_handle_;// cublas的句柄 
  curandGenerator_t curand_generator_;// curandGenerator句柄
#endif
  shared_ptr<RNG> random_generator_;

  Brew mode_;
  int solver_count_;
  bool root_solver_;

 private:
  // The private constructor to avoid duplicate instantiation.
  Caffe();
  // 禁止caffe这个类被复制构造函数和赋值进行构造
  DISABLE_COPY_AND_ASSIGN(Caffe);
};

}  // namespace caffe

#endif  // CAFFE_COMMON_HPP_

下面给出common.cpp的详细注释

#include <boost/thread.hpp>
#include <glog/logging.h>
#include <cmath>
#include <cstdio>
#include <ctime>

#include "caffe/common.hpp"
#include "caffe/util/rng.hpp"

namespace caffe {

// Make sure each thread can have different values.
// boost::thread_specific_ptr是线程局部存储机制
// 一开始的值是NULL
static boost::thread_specific_ptr<Caffe> thread_instance_;

Caffe& Caffe::Get() {
  if (!thread_instance_.get()) {// 如果当前线程没有caffe实例
    thread_instance_.reset(new Caffe());// 则新建一个caffe的实例并返回
  }
  return *(thread_instance_.get());
}

// random seeding
// linux下的熵池下获取随机数的种子
int64_t cluster_seedgen(void) {
  int64_t s, seed, pid;
  FILE* f = fopen("/dev/urandom", "rb");
  if (f && fread(&seed, 1, sizeof(seed), f) == sizeof(seed)) {
    fclose(f);
    return seed;
  }

  LOG(INFO) << "System entropy source not available, "
              "using fallback algorithm to generate seed instead.";
  if (f)
    fclose(f);
  // 采用传统的基于时间来生成随机数种子
  pid = getpid();
  s = time(NULL);
  seed = std::abs(((s * 181) * ((pid - 83) * 359)) % 104729);
  return seed;
}

// 初始化gflags和glog
void GlobalInit(int* pargc, char*** pargv) {
  // Google flags.
  ::gflags::ParseCommandLineFlags(pargc, pargv, true);
  // Google logging.
  ::google::InitGoogleLogging(*(pargv)[0]);
  // Provide a backtrace on segfault.
  ::google::InstallFailureSignalHandler();
}

#ifdef CPU_ONLY  // CPU-only Caffe.

Caffe::Caffe()
    : random_generator_(), mode_(Caffe::CPU),// shared_ptr<RNG> random_generator_;   Brew mode_;
      solver_count_(1), root_solver_(true) { }// int solver_count_;   bool root_solver_;

Caffe::~Caffe() { }
//  手动设定随机数生成器的种子
void Caffe::set_random_seed(const unsigned int seed) {
  // RNG seed
  Get().random_generator_.reset(new RNG(seed));
}

void Caffe::SetDevice(const int device_id) {
  NO_GPU;
}

void Caffe::DeviceQuery() {
  NO_GPU;
}

// 定义RNG内部的Generator类
class Caffe::RNG::Generator {
 public:
  Generator() : rng_(new caffe::rng_t(cluster_seedgen())) {}// linux下的熵池生成随机数种子，注意typedef boost::mt19937 rng_t;这个在utils/rng.hpp头文件里面
  explicit Generator(unsigned int seed) : rng_(new caffe::rng_t(seed)) {}// 采用给定的种子初始化
  caffe::rng_t* rng() { return rng_.get(); }// 属性
 private:
  shared_ptr<caffe::rng_t> rng_;// 内部变量
};

// 实现RNG内部的构造函数
Caffe::RNG::RNG() : generator_(new Generator()) { }

Caffe::RNG::RNG(unsigned int seed) : generator_(new Generator(seed)) { }

// 实现RNG内部的运算符重载
Caffe::RNG& Caffe::RNG::operator=(const RNG& other) {
  generator_ = other.generator_;
  return *this;
}

void* Caffe::RNG::generator() {
  return static_cast<void*>(generator_->rng());
}

#else  // Normal GPU + CPU Caffe.
// 构造函数，初始化cublas和curand库的句柄
Caffe::Caffe()
    : cublas_handle_(NULL), curand_generator_(NULL), random_generator_(),
    mode_(Caffe::CPU), solver_count_(1), root_solver_(true) {
  // Try to create a cublas handler, and report an error if failed (but we will
  // keep the program running as one might just want to run CPU code).
  // 初始化cublas并获得句柄
  if (cublasCreate(&cublas_handle_) != CUBLAS_STATUS_SUCCESS) {
    LOG(ERROR) << "Cannot create Cublas handle. Cublas won't be available.";
  }
  // Try to create a curand handler.
  if (curandCreateGenerator(&curand_generator_, CURAND_RNG_PSEUDO_DEFAULT)
      != CURAND_STATUS_SUCCESS ||
      curandSetPseudoRandomGeneratorSeed(curand_generator_, cluster_seedgen())
      != CURAND_STATUS_SUCCESS) {
    LOG(ERROR) << "Cannot create Curand generator. Curand won't be available.";
  }
}

Caffe::~Caffe() {
  // 销毁句柄
  if (cublas_handle_) CUBLAS_CHECK(cublasDestroy(cublas_handle_));
  if (curand_generator_) {
    CURAND_CHECK(curandDestroyGenerator(curand_generator_));
  }
}
// 初始化CUDA的随机数种子以及cpu的随机数种子
void Caffe::set_random_seed(const unsigned int seed) {
  // Curand seed
  static bool g_curand_availability_logged = false;// 判断是否log了curand的可用性，如果没有则log一次，log之后则再也不log，用的是静态变量
  if (Get().curand_generator_) {
    // CURAND_CHECK见/utils/device_alternate.hpp中的宏定义
    CURAND_CHECK(curandSetPseudoRandomGeneratorSeed(curand_generator(),
        seed));
    CURAND_CHECK(curandSetGeneratorOffset(curand_generator(), 0));
  } else {
    if (!g_curand_availability_logged) {
        LOG(ERROR) <<
            "Curand not available. Skipping setting the curand seed.";
        g_curand_availability_logged = true;
    }
  }
  // RNG seed
  // CPU code
  Get().random_generator_.reset(new RNG(seed));
}

// 设置GPU设备并初始化句柄以及随机数种子
void Caffe::SetDevice(const int device_id) {
  int current_device;
  CUDA_CHECK(cudaGetDevice(¤t_device));// 获取当前设备id
  if (current_device == device_id) {
    return;
  }
  // The call to cudaSetDevice must come before any calls to Get, which
  // may perform initialization using the GPU.
  // 在Get之前必须先执行cudasetDevice函数
  CUDA_CHECK(cudaSetDevice(device_id));
  // 清理以前的句柄
  if (Get().cublas_handle_) CUBLAS_CHECK(cublasDestroy(Get().cublas_handle_));
  if (Get().curand_generator_) {
    CURAND_CHECK(curandDestroyGenerator(Get().curand_generator_));
  }
  // 创建新句柄
  CUBLAS_CHECK(cublasCreate(&Get().cublas_handle_));
  CURAND_CHECK(curandCreateGenerator(&Get().curand_generator_,
      CURAND_RNG_PSEUDO_DEFAULT));
  // 设置随机数种子
  CURAND_CHECK(curandSetPseudoRandomGeneratorSeed(Get().curand_generator_,
      cluster_seedgen()));
}

// 获取设备信息
void Caffe::DeviceQuery() {
  cudaDeviceProp prop;
  int device;
  if (cudaSuccess != cudaGetDevice(&device)) {
    printf("No cuda device present.\n");
    return;
  }
  // #define CUDA_CHECK(condition) \
  /* Code block avoids redefinition of cudaError_t error */ \
  //do { \
  //  cudaError_t error = condition; \
  //  CHECK_EQ(error, cudaSuccess) << " " << cudaGetErrorString(error); \
  //} while (0)
  CUDA_CHECK(cudaGetDeviceProperties(&prop, device));
  LOG(INFO) << "Device id:                     " << device;
  LOG(INFO) << "Major revision number:         " << prop.major;
  LOG(INFO) << "Minor revision number:         " << prop.minor;
  LOG(INFO) << "Name:                          " << prop.name;
  LOG(INFO) << "Total global memory:           " << prop.totalGlobalMem;
  LOG(INFO) << "Total shared memory per block: " << prop.sharedMemPerBlock;
  LOG(INFO) << "Total registers per block:     " << prop.regsPerBlock;
  LOG(INFO) << "Warp size:                     " << prop.warpSize;
  LOG(INFO) << "Maximum memory pitch:          " << prop.memPitch;
  LOG(INFO) << "Maximum threads per block:     " << prop.maxThreadsPerBlock;
  LOG(INFO) << "Maximum dimension of block:    "
      << prop.maxThreadsDim[0] << ", " << prop.maxThreadsDim[1] << ", "
      << prop.maxThreadsDim[2];
  LOG(INFO) << "Maximum dimension of grid:     "
      << prop.maxGridSize[0] << ", " << prop.maxGridSize[1] << ", "
      << prop.maxGridSize[2];
  LOG(INFO) << "Clock rate:                    " << prop.clockRate;
  LOG(INFO) << "Total constant memory:         " << prop.totalConstMem;
  LOG(INFO) << "Texture alignment:             " << prop.textureAlignment;
  LOG(INFO) << "Concurrent copy and execution: "
      << (prop.deviceOverlap ? "Yes" : "No");
  LOG(INFO) << "Number of multiprocessors:     " << prop.multiProcessorCount;
  LOG(INFO) << "Kernel execution timeout:      "
      << (prop.kernelExecTimeoutEnabled ? "Yes" : "No");
  return;
}


class Caffe::RNG::Generator {
 public:
  Generator() : rng_(new caffe::rng_t(cluster_seedgen())) {}
  explicit Generator(unsigned int seed) : rng_(new caffe::rng_t(seed)) {}
  caffe::rng_t* rng() { return rng_.get(); }
 private:
  shared_ptr<caffe::rng_t> rng_;
};

Caffe::RNG::RNG() : generator_(new Generator()) { }

Caffe::RNG::RNG(unsigned int seed) : generator_(new Generator(seed)) { }

Caffe::RNG& Caffe::RNG::operator=(const RNG& other) {
  generator_.reset(other.generator_.get());
  return *this;
}

void* Caffe::RNG::generator() {
  return static_cast<void*>(generator_->rng());
}
// cublas的geterrorstring
const char* cublasGetErrorString(cublasStatus_t error) {
  switch (error) {
  case CUBLAS_STATUS_SUCCESS:
    return "CUBLAS_STATUS_SUCCESS";
  case CUBLAS_STATUS_NOT_INITIALIZED:
    return "CUBLAS_STATUS_NOT_INITIALIZED";
  case CUBLAS_STATUS_ALLOC_FAILED:
    return "CUBLAS_STATUS_ALLOC_FAILED";
  case CUBLAS_STATUS_INVALID_VALUE:
    return "CUBLAS_STATUS_INVALID_VALUE";
  case CUBLAS_STATUS_ARCH_MISMATCH:
    return "CUBLAS_STATUS_ARCH_MISMATCH";
  case CUBLAS_STATUS_MAPPING_ERROR:
    return "CUBLAS_STATUS_MAPPING_ERROR";
  case CUBLAS_STATUS_EXECUTION_FAILED:
    return "CUBLAS_STATUS_EXECUTION_FAILED";
  case CUBLAS_STATUS_INTERNAL_ERROR:
    return "CUBLAS_STATUS_INTERNAL_ERROR";
#if CUDA_VERSION >= 6000
  case CUBLAS_STATUS_NOT_SUPPORTED:
    return "CUBLAS_STATUS_NOT_SUPPORTED";
#endif
#if CUDA_VERSION >= 6050
  case CUBLAS_STATUS_LICENSE_ERROR:
    return "CUBLAS_STATUS_LICENSE_ERROR";
#endif
  }
  return "Unknown cublas status";
}
// curand的getlasterrorstring
const char* curandGetErrorString(curandStatus_t error) {
  switch (error) {
  case CURAND_STATUS_SUCCESS:
    return "CURAND_STATUS_SUCCESS";
  case CURAND_STATUS_VERSION_MISMATCH:
    return "CURAND_STATUS_VERSION_MISMATCH";
  case CURAND_STATUS_NOT_INITIALIZED:
    return "CURAND_STATUS_NOT_INITIALIZED";
  case CURAND_STATUS_ALLOCATION_FAILED:
    return "CURAND_STATUS_ALLOCATION_FAILED";
  case CURAND_STATUS_TYPE_ERROR:
    return "CURAND_STATUS_TYPE_ERROR";
  case CURAND_STATUS_OUT_OF_RANGE:
    return "CURAND_STATUS_OUT_OF_RANGE";
  case CURAND_STATUS_LENGTH_NOT_MULTIPLE:
    return "CURAND_STATUS_LENGTH_NOT_MULTIPLE";
  case CURAND_STATUS_DOUBLE_PRECISION_REQUIRED:
    return "CURAND_STATUS_DOUBLE_PRECISION_REQUIRED";
  case CURAND_STATUS_LAUNCH_FAILURE:
    return "CURAND_STATUS_LAUNCH_FAILURE";
  case CURAND_STATUS_PREEXISTING_FAILURE:
    return "CURAND_STATUS_PREEXISTING_FAILURE";
  case CURAND_STATUS_INITIALIZATION_FAILED:
    return "CURAND_STATUS_INITIALIZATION_FAILED";
  case CURAND_STATUS_ARCH_MISMATCH:
    return "CURAND_STATUS_ARCH_MISMATCH";
  case CURAND_STATUS_INTERNAL_ERROR:
    return "CURAND_STATUS_INTERNAL_ERROR";
  }
  return "Unknown curand status";
}

#endif  // CPU_ONLY

}  // namespace caffe

参考：

[1]CUDA Toolkit文档

http://docs.nvidia.com/cuda/index.html