From: Jon M. Taylor <taylorj@gaia.ecs.csus.edu>
  To  : ggi-develop@eskimo.com
  Date: Thu, 29 Jul 1999 12:04:00 -0700 (PDT)

Re: Accellerated drivers (once more :( )

On Thu, 29 Jul 1999, Jos Hulzink wrote:

> Hi,
> 
> As the ViRGE driver is devellopping very fast now (I got serious help from
> many sides), accelleration is really getting a hot topic for me. I know,
> accelleration must be done in a library as much as possible, to keep the
> kernel module small. (my opinion is still that this is bullshit, but
> okay).

	The kernel module should be as small as it can be without 
compromising system stability (by exposing dangerous registers, etc) or 
losing too much functionality.
 
> So, I come up with a mechanism once (soon I hope) that can do
> accelleration in libraries. Question one: Has someone already done work on
> this and documented it ? I found some pingpong buffer code in the matrox
> driver, but this driver is rather outdated and undocumented.

	Well, I started out with ioctl-per-accel, and until you have good, 
solid acceleration support working both in kernelspace and userspace I 
would recommend staying with ioctls.  They are quite slow, but also quite 
simple to code, and it should be easy enough to switch to using pingpong 
buffers once everything else is stable.
 
> Question two: Say, I got a library. How does GGI know it has to use that
> library ? I mean: Does GGI know what chipset is behind the framebuffer ?
> Trying to tell a ViRGE it has to accellerate won't work in NVidia
> language...

	For GGIMesa, what I did was to have the genkgi-helper library
autodetect which KGI driver was present.  I took the cheezy way out and just
checked for the existence of /proc/kgi/savage4, but the proper way is to
query the KGI driver for suggest-strings and parse those.

> I'm not very good in the GGI related stuff, so magbe the answer is just
> "look there...". Sorry if it is, but I don't have enough time to get the
> complete GGI project clear to me.

	No worries.  Feel free to ask me about this stuff, I have done a lot 
of it over the past six months....

Jon

---
'Cloning and the reprogramming of DNA is the first serious step in 
becoming one with God.'
	- Scientist G. Richard Seed