Nvidia driver & Cuda
- ubuntu clean install, with "download additional drivers" option marked
(otherwise we get a black screen right after installation, even though we use onboard vga in case of a Supermicro MB);
- update ubuntu, reboot;
$ sudo apt update $ sudo reboot
- install needed packages;
$ sudo apt install mc, ssh, make, gcc, g++, build-essential
- strongly recommended - switch from console to ssh via another pc, to avoid any black screen issues and loss of display;
- exit graphical interface, otherwise installer will fail;
$ sudo init 3
- remove any nvidia drivers and/or cuda version that were/if were installed;
$ sudo apt purge nvidia* $ sudo apt remove nvidia-* $ sudo apt autoremove && sudo apt clean
$ sudo rm /etc/apt/sources.list.d/cuda* $ sudo rm -rf /usr/local/cuda*
source:
https://gist.github.com/MihailCosmin/affa6b1b71b43787e9228c25fe15aeba
- (in some cases) remove the nouveau driver
$ sudo bash -c "echo blacklist nouveau > /etc/modprobe.d/blacklist-nvidia-nouveau.conf" $ sudo bash -c "echo options nouveau modeset=0 >> /etc/modprobe.d/blacklist-nvidia-nouveau.conf"
$ cat /etc/modprobe.d/blacklist-nvidia-nouveau.conf
blacklist nouveau options nouveau modeset=0
$ sudo update-initramfs -u
$ sudo reboot
- download and install Nvidia driver (run file)
$ wget https://us.download.nvidia.com/XFree86/Linux-x86_64/535.104.05/NVIDIA-Linux-x86_64-535.104.05.run
$ sudo sh NVIDIA-Linux-x86_64-535.104.05.run
- download and install Cuda toolkit 11.8 (run file)
$ wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda_11.8.0_520.61.05_linux.run
$ sudo sh cuda_11.8.0_520.61.05_linux.run
- options to install - unmark the nvidia driver, in case you have another version installed;
- You may get a warning at the end about the nvidia driver version, but the installation should succeed;
- quoted from Cuda postinstall instructions;
Please make sure that
- PATH includes /usr/local/cuda-11.8/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-11.8/lib64, or, add /usr/local/cuda-11.8/lib64 to /etc/ld.so.conf and run ldconfig as root
- follow the postinstall instructions above and add the path lines to "/home/user/.bashrc"; (replace user with your username);
$ nano /home/user/.bashrc
- add the lines at the end of the file and save the changes;
export PATH="/usr/local/cuda-11.8/bin:$PATH" export LD_LIBRARY_PATH="/usr/local/cuda-11.8/lib64:$LD_LIBRARY_PATH"
- reboot;
- check Cuda version;
$ nvcc --version
- remove the ineffective driver:
sudo dkms remove rtl88x2ce/35403 --all
- install the correct driver:
sudo apt install --reinstall git bc
git clone https://github.com/lwfinger/rtw89.git
cd rtw89
make
sudo make install
- You will probably need to disable Secure Boot. Reboot. (didn't check, it worked without it)
When your kernel version changes, then you need to do the following:
cd ~/rtw89
git pull
make clean
make
sudo make install
EDIT: Let’s load the required firmware:
cd /usr/lib/firmware/
sudo mkdir rtw89
sudo wget https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/plain/rtw89/rtw8852c_fw.bin