NVIDIA Display Driver 1.0-9755
March 8th, 2007 by CrusaderNVIDIA has released 1.0-9755 of their Linux display driver for 32-bit and 64-bit architectures. Changes:
- Added support for Quadro FX 4600 and Quadro FX 5600
- Added initial support for NVIDIA SLI with GeForce 8800, Quadro FX 4600, and Quadro FX 5600
If you have questions or issues, NVIDIA directs you to check their Linux discussion forum, or e-mail linux-bugs@nvidia.com.
Download: [ nvidia.com ]




March 12th, 2007 at 6:11 am
The Nvidia driver just works, as we all know. And for the proprietary, that’s pretty damn good. ATI can’t come close in LinuxLand. But there is one final step for Nvidia to take. One thing that I have only seen HP do in the proprietary driver market.
HP’s new Proliant Service Pack, 7.70 for linux, has a userland shim that checks at boot to see if it’s kernel modules are loaded- aka if the kernel version has changed. It then recompiles/reinserts the modules into the currently running kernels /usr/lib/modules, etc. I had never seen this done before with proprietary drivers, but I thought it was very cool. Until you properly uninstall the PSP, it will keep checking at boot and recompiling itself for the running kernel.
The script they use to do this seems fairly simple, and it’s freely available, so I was wondering if Nvidia might have a look at what can be done with their driver. I think their Linux driver is pretty much neck and neck with their windows one, if not beyond it in some cases, but this one last mile would make a huge difference.
Recompiling proprietary drivers between kernel versions is what makes proprietary drivers suck for the user. This would be a big boon for Nvidia and make it irritation and/or dumbass proof. Just a thought.
March 16th, 2007 at 12:19 pm
I have never heard of that little gem from HP. This definitely would be a great addition to any proprietary driver for *nix. The NV kernel module is small so a recompile only takes a few moments.
One caveat though: some kernel revisions (major version number changes IE 2.6.18 – 2.6.20) _typically_ require some form of patch to get working again. This would require some more tomfoolery but is definitely doable on a system with a net connection. Typically this is fixed through user patches, or when NV gets around to another driver version bump/release. But, with a magical little script, nv could hotfix patches for these kernel bumps between releases.
March 16th, 2007 at 3:11 pm
Sounds like they’re doing DKMS which Mandriva, for example, does by default.