[Panel-devel] ATi- Plasma- support

David Saxton david at bluehaze.org
Mon Aug 22 09:53:28 CEST 2005


On Monday 22 Aug 2005 07:19, Jessica Hall wrote:
> Forgive my ignorance, but what is actually wrong with binary drivers?
> Obviously open sourced drivers would be preferable, but the binary drivers
> I use with my graphics card are excellent. I certainly wouldn't refer to
> them as 'crippled'.  I get the same performance gaming under linux as I
> would under windows, and I'm pretty pleased with that.
>
> People who don't open source their software are not evil and don't need to
> be browbeaten. This attitude gets tiresome.
>
> Regards,
> --
> JH

The problem comes when they don't work. And with nvidia, at least, there is a 
very good example of a serious, widespread bug that has been around for the 
past couple of years:

http://www.nvnews.net/vbulletin/showthread.php?s=93e68398a13a4f50d75d6ee3191aca35&t=31858

I get this bug as well; if I turn on renderaccel in the nvidia driver (which 
presumably will be needed for drawing plasma stuff quickly, and is already 
essential for using the composite extension), then X crashes within a few 
seconds of logging into kde. I sometimes get it when using opengl.

This thread - "Trying to make headway into finding the Xid crashes source..." 
shows just how much effort nvidia users are putting into fixing the bug:

http://www.nvnews.net/vbulletin/showthread.php?s=93e68398a13a4f50d75d6ee3191aca35&t=49117

If the nvidia drivers were open source, then this effort would have fixed the 
bug a long time ago - but instead, it is currently still a very major bug. 
And it's probably not a fault of nvidia that the bug hasn't been fixed yet - 
there are simply too many hardware and software combinations for them to test 
for, given their presumably small linux development team.

It is precisely graphics drivers (which interface directly with the kernel, 
and are widely affected by the underlying hardware - and hence are so 
fragile) that would benefit the most from open sourcing. 

Regards,
David Saxton


More information about the Panel-devel mailing list