Nvidia drivers: RPMfusion vs Nvidia?

[NitroPress]

I'm not clear on the difference between the RPMFusion drivers and Nvidia's own drivers.

My new system uses the GeForce 8200 chipset, which is not well recognized by the installer. Basically, I couldn't get a GUI with the built-in video system. So I stuck in an old FX5500 card, which let me do a full GUI install. Then I tried installing Nvidia's driver package, but (after a warning about the older card) it said it could not find a compatible card and quit.

So, with some hassles, I ran the RPMfusion installation/update listed in the sticky thread, and it worked fine (the 5500 stopped working, but the built-in 8200 started working just fine).

What's the difference between the RPMFusion driver set and Nvidia's? Can/could/should I install Nvidia's package now that the community drivers have things working?

It looks as if I will have to do a completely fresh install for other reasons, so some details about the differences and non/compatibility of these two packages would be helpful.

Thanks!

[hephasteus]

Originally Posted by NitroPress

I'm not clear on the difference between the RPMFusion drivers and Nvidia's own drivers.

My new system uses the GeForce 8200 chipset, which is not well recognized by the installer. Basically, I couldn't get a GUI with the built-in video system. So I stuck in an old FX5500 card, which let me do a full GUI install. Then I tried installing Nvidia's driver package, but (after a warning about the older card) it said it could not find a compatible card and quit.

So, with some hassles, I ran the RPMfusion installation/update listed in the sticky thread, and it worked fine (the 5500 stopped working, but the built-in 8200 started working just fine).

What's the difference between the RPMFusion driver set and Nvidia's? Can/could/should I install Nvidia's package now that the community drivers have things working?

It looks as if I will have to do a completely fresh install for other reasons, so some details about the differences and non/compatibility of these two packages would be helpful.

Thanks!

Install using linux vesa. Don't use nvidia drivers from nvidia page. Have all your questions answered on a sticky thread about how to get nvidia drivers working.

[PabloTwo]

I'll attempt to actually answer some of your questions, primarily, what is the difference between the Nvidia driver from Nvidia and the Nvidia driver as provided by the kmod-nvidia package from the RPMFusion repo.

The actual video driver in either is actually the same. The primary difference is in how it's installed and how it's maintained. In either case, the driver requires a kernel module, which is specific to the version of kernel you are using, and with each kernel upgrade, a new nvidia kernel module must be provided to match the new kernel.

To install the driver as provided by Nvidia, you are required to compile the driver on your system. All that's done by the install script, but it requires you to have a C compiler, gcc and the kernel-devel package for the kernel your compiling it against. If you were to use the the package from Nvidia, then the next kernel update would 'break' your video driver (the one for the previous version kernel won't work with the new kernel version) and you have to repeat the process..... basically, reinstall the driver going through the whole compile/install proceedure again.

The kmod-nvidia packages make life a lot easier, as they provide a pre-compiled driver for your kenel version. And when a new kernel version comes out, it is quickly followed by a matching kmod-nividia package to match, so that when you do a a general 'update' which might include a new version of the kernel, the kmod-nvidia package will usually be right there with it. No fuss, no muss. On occassion though, the kmod-nvidia package might lag a bit behind the release of the kernel, so one has to be careful not to update the kernel until checking to see that the kmod-nvidia package to match is also available from the RPMFusion repo. To get around the late kmod-nvida problem, there is the akmod-nvidia solution. With that, if you update to a new kernel version without having the kmod-nvidia package to go with it, it will 'build' the new driver for you at boot time (requires gcc, kernel-devel just like the Nvidia package).

Hope this helps shed some light on the answers you seek.

[JohnVV]

to add
The nvidia.run installer will install different meas GL files than fedora uses . 99.9% of the time this dose not matter much , some times it dose .i build a bunch of my own code and having the gl *.h, *.so files change on me breaks things .
i get around this by reinstalling "mesa-libGL mesa-libGLU mesa-libGL-devel mesa-libGLU-devel"
after i run the .run file in my sig. for EVERY and ALL kernel, X11, and xorg updates

[brebs]

Originally Posted by PabloTwo

and you have to repeat the process

So automate it - let Nvidia's own driver recompile automatically as-needed on a reboot

With newbie confusion over kmods, akmods, and then nvidia driver incompatibilities and xorg.conf custom settings that might be needed, nvidia questions seem to be the number 1 question asked on this forum

Most of them should be asked in Nvidia's own forum.

JohnVV, fix the Mesa file overwrite problem with --no-opengl-headers, as in my URL.

[NitroPress]

Originally Posted by PabloTwo

Hope this helps shed some light on the answers you seek.

Much, and thanks again.

[JohnVV]

i know i should have reread "sh ./NVIDIA-Linux-x86-96.43.09-pkg1.run -A"