hybrid-graphics-linux team mailing list archive

[Eric Appleman <erappleman@xxxxxxxxx>]

Here's what we know.

* Nvidia is joining the Linux Foundation. Why? No clue. Maybe just for 
improving Tegra.

* For the moment, they are locked out of the GPL-only driver hooks of 
dma-buf for kernel 3.3 and probably 3.4.

* Per the dri-devel mailing list, LKML, and Mr. Plagman, there is no 
other option if they can't use dma-buf.

* According to leaked specifications and insider information on Notebook 
Reviews, the 600 series notebooks will be a mix of Fermi and Kepler.

Up to GT 635M - Fermi
GT 640M/GT 650M - Kepler
GTX 660M - Most likely Kepler
GTX 670M - Fermi
GTX 675M - Fermi
GTX 680M - Kepler

* GTX laptops (except perhaps the 680M and those with 3D) will have 
Optimus. Sager is definitely doing it for their 670M laptop and Dell is 
expected to use it for their Alienware line like they've been doing 
since the 460M.

Knowing all of this, I wouldn't think very much of Nvidia's situation. 
It sucks right now and is going to get worse when their GTX customers 
who use Linux find that they have no drivers and barely any CUDA 
documentation in the binary.

How can Nvidia fix all of this? Open specifications or authoring code 
for the Nouveau and Intel driver stacks would be a start. They could 
lobby harder for the ability to use dma-buf, but it looks like they've 
given up at least until the next major Linux dev conference.

- Eric