This is the kernel-module needed by the proprietary binary nvidia
driver. You also need the nvidia-driver package from SlackBuilds.org.
To build the package for a kernel different from the running one,
start the script setting the KERNEL variable as in
KERNEL=4.6.3 ./nvidia-kernel.SlackBuild
This script now includes the option to build the open kernel module
instead of the default proprietary module. To build the open module,
pass "OPEN=yes" to the script. (See notes below.)
A default config file is placed at
/usr/share/X11/xorg.conf.d/10-nvidia.conf
to make sure that X loads the nvidia module. If you need to make
changes, move that file to /etc/X11/xorg.conf.d/ and edit the copy.
You do not need this file at all if you have a proper and complete
xorg.conf.
The xf86-video-nouveau-blacklist package from /extra is required.
After installation, you will need to reboot your computer for the
changes to take effect.
NOTES ON THE OPEN KERNEL MODULE (quoted from the Nvidia driver README)
"The proprietary flavor supports the GPU architectures Maxwell, Pascal,
Volta, Turing, and later GPUs.
"The open flavor of kernel modules supports Turing and later GPUs. The
open kernel modules cannot support GPUs before Turing, because the open
kernel modules depend on the GPU System Processor (GSP) first
introduced in Turing.
"The following features will only work with the open kernel modules
flavor of the driver:
o NVIDIA Confidential Computing
o Magnum IO GPUDirect Storage (GDS)
o Heterogeneous Memory Management (HMM)
o CPU affinity for GPU fault handlers
o DMABUF support for CUDA allocations"
be51d1869d