Process:
1. Install the graphics driver
X Disable Service (GUI closed, preferably a computer backup) - can not be achieved by remote control operation later. Therefore, this section also needs to operate on the local. Alternate reason a computer can be seen how the next steps in another computer.
Black screen, solution:
2. Install CUDA
3.CUDNN
1.
ubuntu16.04 system run graphics driver installed nvidia
来自 <https://blog.csdn.net/xunan003/article/details/81665835>
2.
ubuntu16.04 installation tensorflow-gpu and cuda8.0 accelerated training
来自< https://www.cnblogs.com/buxizhizhoum/p/8086230.html >
3.
Ubuntu16.04 install NVIDIA driver, to achieve accelerated GPU
来自 <https://blog.csdn.net/zhang970187013/article/details/81012845>
4.Ubuntu16.04 use apt get command to install Nvidia graphics driver
来自 <https://blog.csdn.net/breeze5428/article/details/80013753>
Installation under WINDOWS environment:
reference:
https://blog.csdn.net/weixin_44095128/article/details/88664928
cuda installation problems, despite the vs2015 installed;
reference:
https://blog.csdn.net/zzpong/article/details/80282814
This effective
Win10 NVDIA CUDA solutions for the installation fails
来自 <https://blog.csdn.net/w670165403/article/details/81737139>
cudnn (essentially a cuda a patch of it, no need to install)
Copy the folder to the cuda file cudnn
“C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0”
About tensorfolw-gpu download, following instructions :( not in the system command line, but the anaconda prompt)
pip install D:\tensorflow_gpu-1.13.1-cp37-cp37m-win_amd64.whl
Note that without the quotes
If not, then the above command: select the anaconda prompt input
pip install tensorflow-gpu==1.13.1
other:
Be sure to look at the inside of his anaconda python is the number of reference:
(Solution) tensorflow ImportError: DLL load failed: The specified module could not be found
来自 <https://blog.csdn.net/Halsey_/article/details/88411095>