Nvidia graphics driver naming convention

[Nicolas Marie <thunderdome2@free.fr>]

Hello.

I've got three workstations using different Nvidia GPU. I upgraded them from Debian 11.6 to Debian 12:

• The first one is a Geforce GT520 (Fermi) so I used the package nvidia-graphics-drivers-legacy-390xx and it works like a charm.
• The second one is a Geforce 1050 Ti (Pascal) si I used nvidia-graphics-drivers which is the version 525 and it also works like a charm.
• The last one is a Geforce GT730 (Kepler) so I have to use the Nvidia driver version 470.

There isn't any package named nvidia-graphics-drivers-legacy-470xx, like we can find for the 390 mentioned version or the older 340 version. As a workaround, I used nvidia-graphics-drivers-tesla-470 and all the GLX stack / SMI utility / nvidia-settings utility. Basically, all seems to be working:

• The nvidia module is built through the DKMS workflow and loaded.
• The GPU is recognized.
• The basic graphics acceleration (KDE windows effects / GLXGears) works without any glitch.

My question is about the package naming: is the nvidia-graphics-drivers-tesla-470 specifically built for Tesla graphics, and the fact that I can use it for my consumer Nvidia GPU is a positive "side effect"? Or is this package totally usable with a consumer GPU and, in this case the name nvidia-graphics-drivers-legacy-470xx would be more relevant for the non-aware debian new users. As a comparison, Ubuntu uses a similar naming convention, namely nvidia-driver-390, nvidia-driver-470 and nvidia-driver-525.

P.S.: I know that this question type should be escalated during beta states but, since I used the Debian Bullseye Backports packages for nearly one year, I wrongfully believed these drivers were be maintained in the next Debian release.

Thanks for your answer.
Regards.