Guillaume Cottenceau wrote:
>
> Pedro Rosa <[EMAIL PROTECTED]> writes:
>
> > >
> > Yeap. Unfortunately there is something real wrong with this tdfx_dri...
> > X does not seem to have it. Rebuilding X shows several *_dri.so drivers
> > but no tdfx. Even DRI from CVS on sourceforge does not seem to compile
> > it. However DRI docs talk about its existence. And even say to turn off
> > DRI support on the kernel (why?). Could you give some hints about this?
>
> The same as before -- it's include in XF4 sources but needs some libs to
> achieve compile.
Thanks for the info. I'll try to check out.
>
>
> > > It has been successfully tested with i810 and ati-mach64, for the moment.
> > > The acceleration with i810 (with a p3-733, ok) is pretty impressive (q3
> > > with all things turned to full quality).
> > i810? How stable?
>
> pretty stable! and rock fast compared to what was expected..
Well probably there are some improvements since I last touched it... On
older sources, things went fast but I had to drop the mchine because it
was segfaulting in many things. Have you used 2.4.0 on them? Because my
main troubles were due to it.
>
> [...]
>
>
> > > What is the problem with svgalib? It works fine here for 7.0 and 7.1.
> > Right. For 7.0 & 7.1. On cooker post-7.1 (well, people are reading a new
> > distro right?) things crash badly on VESA option (comp freezes
> > sometimes) and show some funny effects on BANSHEE option (magenta screen
> > at start of SVGALIB app, console trashed). Similar things happen with
> > the lastest tarball of svgalib btw. But this time, the comp doesn't
> > freeze anymore. For info the comp is a dual Celleron 300a speeded up to
> > 450Mhz on Abit BP6 128M with Voodoo3 3000+Voodoo Graphics board
> > (hardware never showed any conflicts for half year). I tested this
> > situation on one latest build of mandrake kernel & 2.4.0-test* kernels.
>
> svgalib has not changed since 7.1, execpt a recompile by myself to add
> some %config(noreplace) stuff where needed, and to change the default
> mouse from serial-ms to ps/2.
>
> it's therefore improbable that it broke up since 7.1 ; and having some
> 300mhz comps working at 450mhz is surely a way that should be
> investigated..
I knew you may answer such thing... Well the fact is that these 300a are
full-proof. They were specially tested on overclocking (it's a religion
here, we overclock everything from bus to video, and we are very
pedantic on doing it) and have been
working with several distros since September last year (including all
Mandrakes since 6.0). Besides the machine runs non-stop. It is only
downed to insert or remove hardware. The svgalib feature was checked
even when 7.1 was still on it, along with the new "half-cooker" system
(later, due to disk space I had to scrap 7.1) There is a problem with
svgalib for sure. I would not state that is svgalib fault by itself.
Specially if it happens under two different kernels and two different
versions of svgalib. However, under 7.1, I also used 2.2 and 2.4 and
there were absolutely no problems (on 2.4 I can assure that the
configuration was exactly the same as now). Frankly I looked over every
difference between old/new and I couldn't get a hint (frame buffer,
devfs, video settings, X "features", svgalib builds&configs).
Generically the environment remains the same.
>
>
> --
> Guillaume Cottenceau -- Distribution Developer for MandrakeSoft
> http://www.mandrakesoft.com/~gc/
Ektanoor
