Recently The Inquirer has had some news on what they are claiming will be the next-generation NVIDIA display drivers for Vista called Forceware X. Forceware X is supposed to deliver DirectX 10 support for the G80 on Microsoft Vista, a new control panel, and more.
The 1.0-9XXX series has been relatively quick and short. Coming up in just a few days will actually be the one year anniversary of the introduction of the 1.0-8XXX series for Linux.
The latest Linux beta driver is at version 1.0-9742, so this leaves the question, what would you like to see in the next NVIDIA Linux driver series? You can recall the shared speculations here. What do you think will end up getting thrown into the next drivers?
One thing that is long overdue for revamping is the NVIDIA installer.
I don't think they could add much more performance. The driver already perform really well so I don't think spending hundreds of man hours to make it 2% faster is worth it.
The nvidia drivers are just about perfect. When I first started using them there was quite a few lockups but now they are all sorted out.
In my opinion they should work to make the drivers support AIGLX better, at the moment if there are to many windows then the new ones start going black. This is caused by it only able to use video memory. This is a real issue for me because I only have a 128mb card but two 20" monitors running at 1680x1050. So after only a few windows I start getting black ones.
Also they should try and be really quick with supporting new version of Xorg too.
I think there are some points to improve :
- the tv-out handling. An option to have a clone mode with fullscreen video on TV (like the windows driver) should be nice.
- mpeg4 hardware accelerated, especially for HD. Something like purevideo (improve xvmc?)
- maybe another clone mode on twinview with a scaling option when we cannot set the same resolution.
- fix the memory bug with tfp, to get rid of black windows bug with beryl.
I would personally like more support/documentation for running multiple cards and many multiple monitors and having accelerated GLX spanning many multiples of monitors, I have an agp and PCI-e slot and lots of big screens so running two lots of twinview and having a (good to high performance) implementation of hardware acceleration spanning 6400x1200 worth of screen estate would be sweet.
Throw in V4L support for VI/VO cards and that'd be awesome. I'm not fussed on SLI and we're not really going to benefit from hardware DX10 support, unless they use the unified shaders to convert cedega/wine D3D to OGL code on the hardware leaving the CPU free to process the original engine and physics, rather than doing both.