2011 Linux Graphics Survey Results

[AnonymousCoward]

What if i've moved over to the xorg.conf.d setup? Also I'd not say i was "having to rely" on editing the config, just that there are some things the graphical utility doesn't know about.

As for the compositor, I wonder how many of those not using one said that they use the nvidia blob driver? it has known full screen openGL performance issues with a compositing window manager.

I'd like to use the Nouveau driver for my card, but i need it to work with WINE. The x86-compat/* bits in gentoo don't contain a 32 bit version last time i looked. This is important because while it ran well for desktop use, it did not work with 32 bit wine.

Gentoo does have the compat libs these days ( I used it yesterday). However, it is probably too slow if you plan to use anything graphically heavy with wine. Although that was partly because my card wouldn't switch to its maximum performance state, so YMMV.

[pszilard]

Just to make things clear that there's nothing "unfortunate" on still using a Matrox gpu. I use it on a basic storage server built from old parts (and a few new ones) where even the Millenium II G200 8MB in there is more than enough to display the command line interface. I mostly just interface with it remotely anyway... I suspect that the others running equally outdated hardware are on a similar situation and don't use these systems for doing much production work.

Agreed, that was a weird comment - unless using servers and high-end workstations is not considered "use".

We have quite a few Supermicro file and compute servers and they all have the Matrox G200eW. Also, Tyan is using Aspeed graphics chips on their boards, I use it on an S7025.

[pszilard]

The use of GPU compute technologies (OpenCL, CUDA, etc) still is not widespread on Linux. In part this is due to the open-source drivers lacking the compute support.

Unless you were referring to home users only that's not entirely true.

There are quite many scientific (and some non-scientific) applications that make use of CUDA/OpenCL, a great number of them used on Linux. Also there are already a pretty large number of Linux clusters with GPUs and the majority runs Linux!