Tue Jan 19 14:04:29 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 418.56 Driver Version: 418.56 CUDA Version: 10.1 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce RTX 208... Off | 00000000:65:00.0 On | N/A |
| 41% 35C P8 20W / 250W | 454MiB / 10987MiB | 2% Default |
+-------------------------------+----------------------+----------------------+
| 1 GeForce RTX 208... Off | 00000000:B6:00.0 Off | N/A |
| 41% 32C P8 3W / 250W | 1MiB / 10989MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 1337 G /usr/lib/xorg/Xorg 18MiB |
| 0 1392 G /usr/bin/gnome-shell 58MiB |
| 0 1595 G /usr/lib/xorg/Xorg 137MiB |
| 0 1707 G /usr/bin/gnome-shell 140MiB |
| 0 2248 G ...AAAAAAAAAAAIAAAAAAAAAA== --shared-files 98MiB |
+-----------------------------------------------------------------------------+
$ nvcc --version
Command 'nvcc' not found, but can be installed with:
sudo apt install nvidia-cuda-toolkit
위 처럼 driver version과 CUDA버전을 확인 할 수 있는데 이 경우 nvcc -V 명령이 안먹는 경우가 있다.
CUDA has 2 primary APIs, the runtime and the driver API. Both have a corresponding version (e.g. 8.0, 9.0, etc.)
The necessary support for the driver API (e.g. libcuda.so on linux) is installed by the GPU driver installer.
The necessary support for the runtime API (e.g. libcudart.so on linux, and also nvcc) is installed by the CUDA toolkit installer (which may also have a GPU driver installer bundled in it).
In any event, the (installed) driver API version may not always match the (installed) runtime API version, especially if you install a GPU driver independently from installing CUDA (i.e. the CUDA toolkit).
(발췌) https://stackoverflow.com/questions/53422407/different-cuda-versions-shown-by-nvcc-and-nvidia-smi
sudo apt install nvidia-cuda-toolkit
이 경우는 PATH가 빠진 것으로 /usr/local/폴더에 cuda가 있는지 확인해 보자. SDK Manager를 통해 CUDA 10.0을 설치 했으므로 /usr/local/cuda-10.0 폴더가 존재 한다. /usr/local/cuda-10.0/bin 폴더에 nvcc명령도 있는 것을 확인 할 수 있다. 즉 .bashrc에 PATH환경 설정을 해주면 된다.
export PATH=/usr/local/cuda-10.0/bin:/usr/local/cuda-10.0/NsightCompute-2019.1{PATH:+:{PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64{LD_LIBRARY_PATH:+:{LD_LIBRARY_PATH}}
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-ubuntu1804.pin
sudo mv cuda-ubuntu1804.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/10.1/Prod/local_installers/cuda-repo-ubuntu1804-10-1-local-10.1.243-418.87.00_1.0-1_amd64.deb
(nvidia install 가이드 순서가 바뀌어 있음)
sudo apt-key add /var/cuda-repo-10-1-local-10.1.243-418.87.00/7fa2af80.pub
sudo dpkg -i cuda-repo-ubuntu1804-10-1-local-10.1.243-418.87.00_1.0-1_amd64.deb
sudo apt-get update
sudo apt-get -y install cuda
Errors were encountered while processing:
/tmp/apt-dpkg-install-hiPrxn/30-libcublas-dev_10.2.1.243-1_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)