วันจันทร์ที่ 27 มิถุนายน พ.ศ. 2559

การติดตั้ง cuDNN

#################### cudnn #####################

1.https://developer.nvidia.com/rdp/cudnn-download
  #Download cudnn

  mv test_Old test_New
  #Rename folder

2.sudo cp include/cudnn.h /usr/include

3.sudo cp lib64/libcudnn* /usr/lib/x86_64-linux-gnu/

4.sudo chmod a+r /usr/lib/x86_64-linux-gnu/libcudnn*

5.sudo apt-get install nvidia-cuda-toolkit

6.sudo add-apt-repository ppa:graphics-drivers/ppa
  #Add the ppa repo

7.sudo apt-get update

8.sudo ubuntu-drivers autoinstall
  #Install the recommended driver (currently nvidia-364)

9.sudo reboot

10.sudo apt-get update

11.sudo apt-get upgrade

12.nvcc -V

13.sudo apt-get install nvidia-nsight

14.sudo apt-get install nvidia-profiler

15.sudo apt-get install libcupti-dev zlib1g-dev

########## Put symlinks in /usr/local/cuda ##########################

16.cd /usr/local/cuda

17.sudo ln -s  /usr/lib/x86_64-linux-gnu/ lib64

18.sudo ln -s  /usr/include/ include

19.sudo ln -s  /usr/bin/ bin

20.sudo ln -s  /usr/lib/x86_64-linux-gnu/ nvvm

21.cd /usr/local/cuda/extras/CUPTI

22.sudo ln -s  /usr/lib/x86_64-linux-gnu/ lib64


Ref: https://devtalk.nvidia.com/default/topic/936212/tensorflow-cannot-find-cudnn-ubuntu-16-04-cuda7-5-/

ไม่มีความคิดเห็น:

แสดงความคิดเห็น