For people wanting to give the 450.36.06 cuda driver a try, run the following from nvidia-all dir:
```
wget http://developer.download.nvidia.com/compute/cuda/11.0.1/local_installers/cuda_11.0.1_450.36.06_linux.run
sudo chmod +x cuda_11.0.1_450.36.06_linux.run
./cuda_11.0.1_450.36.06_linux.run --extract=$HOME/nv-cuda (or any absolute path of your choice)
mv $HOME/nv-cuda/NVIDIA-Linux-x86_64-450.36.06.run ./
makepkg -si
```
Then select option 9 (custom), followed by 2 (Vulkan dev), then input 450.36.06
Please keep in mind this doesn't reflect official support from nvidia, since we're using custom patches to extend compat to newer kernels than the drivers actually support.
Following https://github.com/Tk-Glitch/PKGBUILDS/issues/567
- Remove the PRIME_render_offload option that seemingly made little sense anyway.
- Remove the coolbits option that somehow didn't work correctly for multiple users, that will at least not interfere with user-set coolbits.
- Make a separate nvidia-settings package (optional) for nvidia-settings.
- Get rid of legacy/unneeded options and 20-nvidia.conf Xorg file.
- Rollback to the usual nouveau blacklisting way, and remove nvidiafb blacklist. If your setup doesn't work correctly with this, it's a setup issue as virtually all other linux nvidia packages are doing it that way 🐸
- Remove the prime-oriented intel outputclass from 10-nvidia-drm-outputclass.conf. Nobody does this, and I suspect it creates some issues for some. I'll be waiting for some feedback to offer it optionally if needed.
- Double = signs cleanups.
Should fix https://github.com/Frogging-Family/nvidia-all/issues/2 but will now require the user to enable coolbits by themselves.
