在conda安装完成cuda、cudnn、tensorflow-gpu之后,运行测试代码,出现:
NVIDIA-SMI has failed because it couldn’t communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.
问题可以看出是GPU的问题,没找到。
接着命令行输入:nvidia-smi竟然不能打印出GPU 的信息。
问题原因:2080 GPU 的显卡驱动默认是没安装的,需要手动安装,解决过程:
Automatic Install using PPA repository to install Nvidia Beta drivers
Using graphics-drivers
PPA repository allows us to install bleeding edge Nvidia beta drivers at the risk of unstable system.
- To proceed first add the
ppa:graphics-drivers/ppa
repository into your system:$ sudo add-apt-repository ppa:graphics-drivers/ppa $ sudo apt update
- Next, identify your graphic card model and recommended driver:
$ ubuntu-drivers devices == /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0 == modalias : pci:v000010DEd00001C03sv00001043sd000085ABbc03sc00i00 vendor : NVIDIA Corporation model : GP106 [GeForce GTX 1060 6GB] driver : nvidia-driver-390 - third-party free driver : nvidia-driver-410 - third-party free recommended driver : nvidia-driver-396 - third-party free driver : xserver-xorg-video-nouveau - distro free builtin
- Same as with the above standard Ubuntu repository example, either install all recommended drivers automatically:
$ sudo ubuntu-drivers autoinstall
or selectively using theapt
command. Example:$ sudo apt install nvidia-410
- Once done, reboot your system.