为Cupy提高GPU支持

Cupy安装

安装docker及nvidia-docker

在Ubuntu上安装Docker并使得Docker支持GPU

安装Cupy

在宿主机上安装GPU驱动

• 查找合适的Nvidia驱动器版本并安装

  sudo ubuntu-drivers devices
  sudo ubuntu-drivers autoinstall
  

在Docker内需且仅需安装cuda

• 查看最大支持的cuda版本

  nvidia-smi
  

• 选择指定版本的cuda，安装类型选择runfile(local)

  wget https://developer.download.nvidia.com/compute/cuda/10.1/Prod/local_installers/cuda_10.1.243_418.87.00_linux.run
  

  CUDA Toolkit Archive | NVIDIA Developer

• 安装cuda，选择仅安装cuda-toolkit，并配置环境变量

  sh cuda_10.1.243_418.87.00_linux.run
  echo "export LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/extras/CPUTI/lib64" >> ~/.bashrc
  echo "export CUDA_HOME=/usr/local/cuda/bin" >> ~/.bashrc
  echo "export PATH=$PATH:$LD_LIBRARY_PATH:$CUDA_HOME" >> ~/.bashrc
  source ~/.bashrc
  

  Ubuntu安装cuda

  • 若报错Failed to verify gcc version. See log at /var/log/cuda-installer.log for details.，则添加--override参数或安装cuda对应版本的gcc
    • 添加--override参数

      sh cuda_10.1.243_418.87.00_linux.run --override
      

      • ubuntu22.10安装cuda出错Failed to verify gcc version. See log at /var/log/cuda-installer.log for details.

    • 查看cuda对应的Versioned Online Documentation，安装合适版本的gcc

      cp /etc/apt/sources.list /etc/apt/sources.list.bak
      echo "deb https://mirrors.ustc.edu.cn/ubuntu/ focal main restricted universe multiverse" > /etc/apt/sources.list
      echo "deb https://mirrors.ustc.edu.cn/ubuntu/ focal-security main restricted universe multiverse" >> /etc/apt/sources.list
      echo "deb https://mirrors.ustc.edu.cn/ubuntu/ focal-updates main restricted universe multiverse" >> /etc/apt/sources.list
      echo "deb https://mirrors.ustc.edu.cn/ubuntu/ focal-backports main restricted universe multiverse" >> /etc/apt/sources.list
      apt update | grep NO_PUBKEY
      gpg --keyserver keyserver.ubuntu.com --recv-keys $key
      gpg --export --armor $key | apt-key add -
      apt update
      apt install gcc-7 -y
      apt install g++-7 -y
      apt upgrade -y
      apt autoremove
      update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-7 90
      update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-7 90
      update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-10 50
      update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-10 50
      update-alternatives --config gcc
      

      • CUDA Toolkit Archivee
      • apt update 报错NO_PUBKEY

在Docker内安装cupy

• 根据cuda版本选择对应版本的cupy

  pip install cupy-cuda100
  

  Installing CuPy

• 测试cupy

  python
  import cupy as cp
  import numpy as np
  data_np = np.arange(100)
  data_cp = cp.asarray(data_np)
  

cupy案例

• 矩阵乘法

  import time
  
  import numpy as np
  import cupy as cp
  
  def time_of_numpy_to_cupy_to_numpy(data_size):
      data_np = np.ones((data_size,data_size), dtype=float)
  
      time_start = time.perf_counter()
      data_cp = cp.asarray(data_np)
      time_end = time.perf_counter()
      print('cp np_to_cp duration {}ms'.format((time_end-time_start)*1000))
  
      time_start = time.perf_counter()
      data_cp = cp.matmul(data_cp, data_cp)
      time_end = time.perf_counter()
      print('cp operate duration {}ms'.format((time_end-time_start)*1000))
  
      time_start = time.perf_counter()
      data_np = np.matmul(data_np, data_np)
      time_end = time.perf_counter()
      print('np operate duration {}ms'.format((time_end-time_start)*1000))
  
      time_start = time.perf_counter()
      data_np = cp.asnumpy(data_cp)
      time_end = time.perf_counter()
      print('cp np_to_cp duration {}ms'.format((time_end-time_start)*1000))
  
  
  if __name__ == '__main__':
      data_size = 10000
      time_of_numpy_to_cupy_to_numpy(data_size)
      time_of_numpy_to_pytorch_to_numpy(data_size)