Addendum - Re: XF86 4.0.1 (V3 3000) & DRI working? (I think, maybe...read on, kinda long)

Addendum - Re: XF86 4.0.1 (V3 3000) & DRI working? (I think, maybe...read on, kinda long)

Post by Chris Rip » Sat, 04 Nov 2000 06:52:50



Alright, I'm confused. :) (Crossposted to a couple more groups for some more
insight?)

I used alien on the tdfx_dri rpm (3dfx's X server package), used
dpkg --extract to open it in a separate location, and copied *their* libGL,
GLU, glut, etc. etc. into /usr/lib along with the symlinks (and ldconfig'd
it).

If I run 'gears' it seems to run OK, and then when I quit gears I get this
output:

chris:/home/chris# (0): [drm] removed 1 reserved context for kernel
(0): [drm] unmapping 4096 bytes of SAREA 0xcfaf6000 at 0x40017000
(==) TDFX(0): Write-combining range (0xe8000000,0x2000000)
(II) TDFX(0): Textures Memory 3.96 MB
(0): [drm] created "tdfx" driver at busid "PCI:1:0:0"
(0): [drm] added 4096 byte SAREA at 0xcacc9000
(0): [drm] mapped SAREA 0xcacc9000 to 0x40017000
(0): [drm] framebuffer handle = 0xe8000000
(0): [drm] added 1 reserved context for kernel
(II) TDFX(0): [drm] Registers = 0xe4000000
(II) TDFX(0): visual configs initialized
(0): X context handle = 0x00000001
(0): [drm] installed DRM signal handler
(0): [DRI] installation complete
(II) TDFX(0): direct rendering enabled

So what's going on? Is it running hardware accelerated or not?! glxinfo
reports 'Mesa Indirect' but this looks like it might *really* be running in
hardware (glxinfo somehow in error)?  As in when gears quits it re-inits the
hardware after using it, or is it re-init-ing the hardware because it didn't
use it....

RRRRRRrrrrrrrrrrr.  Confusing.

 
 
 

Addendum - Re: XF86 4.0.1 (V3 3000) & DRI working? (I think, maybe...read on, kinda long)

Post by Chris Rip » Wed, 08 Nov 2000 04:00:00


OK, I figured this out by trial and error, and from reading lots of messages
in 3dfx.glide.linux

install your XF 4.0.1 from the debs
install rpm, you'll have to symlink /usr/lib/rpm to /var/lib/rpm manually
(rpm wants it there)
stop X
remove *all* your /usr/lib/libGL* and /usr/lib/libglut* files and symlinks
manually
un-install any previous glide versions (1 and 2) with apt-get remove or
dpkg --purge.
you should leave xlib[os]mesa3[-dev] intact along with glutg3 (I think, at
least I did.)
rmmod tdfx if it shows up in lsmod
put 'DefaultColorDepth 16' in the 'Screen' section of your XF86Config[-4]
file. Not sure if this is needed but it was hinted at by others in
3dfx.glide.linux

Now, go to 3dfx.linux.com and download the rpms....*and use rpm to
install/build these!!!*
Previously I had used alien to convert these into .debs, and apparently
something was lost in the translation. *shrug, who knows*.  You'll have to
use --nodeps on them.  I'm assuming that at the next iteration of the
XF4.0.1 debs, I'll have to re-install these using --force as well, we'll
have to see what happens there.

Run 'ldconfig' and it should happen!

start up X and run 'glxinfo' you should *not* see 'Mesa GLX Indirect'
anywhere in there, but rather something along the lines of 'Voodoo DRI'
where it used to be.  Mesa will show as 3.3 beta instead of 3.4, but who the
hell cares, it works.

Note: I just did all this from a console window via ssh, so I wasn't
actually present to see my monitor when I did all this, but previously when
running a GL app, X would show up in top as using a lot of processor time,
and now the bulk of the processor time is used by the GL app instead.  Just
for kicks I ran quake-gl from the console and I was getting quake-console
messages in a decent time (You got the nails, etc.) so I can only assume it
was working.

Brilliant! Glad I tried this as I was ready to give up and install Red Hat
again.....

I'll follow up when I get home and can actually see the *y thing in
action.

Now just have to get the USB Wacom Graphire working in X and I'll be set!
*grin*


Quote:> Alright, I'm confused. :) (Crossposted to a couple more groups for some
more
> insight?)

> I used alien on the tdfx_dri rpm (3dfx's X server package), used
> dpkg --extract to open it in a separate location, and copied *their*
libGL,
> GLU, glut, etc. etc. into /usr/lib along with the symlinks (and ldconfig'd
> it).

> If I run 'gears' it seems to run OK, and then when I quit gears I get this
> output:

> chris:/home/chris# (0): [drm] removed 1 reserved context for kernel
> (0): [drm] unmapping 4096 bytes of SAREA 0xcfaf6000 at 0x40017000
> (==) TDFX(0): Write-combining range (0xe8000000,0x2000000)
> (II) TDFX(0): Textures Memory 3.96 MB
> (0): [drm] created "tdfx" driver at busid "PCI:1:0:0"
> (0): [drm] added 4096 byte SAREA at 0xcacc9000
> (0): [drm] mapped SAREA 0xcacc9000 to 0x40017000
> (0): [drm] framebuffer handle = 0xe8000000
> (0): [drm] added 1 reserved context for kernel
> (II) TDFX(0): [drm] Registers = 0xe4000000
> (II) TDFX(0): visual configs initialized
> (0): X context handle = 0x00000001
> (0): [drm] installed DRM signal handler
> (0): [DRI] installation complete
> (II) TDFX(0): direct rendering enabled

> So what's going on? Is it running hardware accelerated or not?! glxinfo
> reports 'Mesa Indirect' but this looks like it might *really* be running
in
> hardware (glxinfo somehow in error)?  As in when gears quits it re-inits
the
> hardware after using it, or is it re-init-ing the hardware because it
didn't
> use it....

> RRRRRRrrrrrrrrrrr.  Confusing.


 
 
 

Addendum - Re: XF86 4.0.1 (V3 3000) & DRI working? (I think, maybe...read on, kinda long)

Post by Edward W. Champion Jr » Mon, 13 Nov 2000 16:39:45




> OK, I figured this out by trial and error, and from reading lots of
> messages in 3dfx.glide.linux
> .....
> start up X and run 'glxinfo' you should *not* see 'Mesa GLX Indirect'
> anywhere in there, but rather something along the lines of 'Voodoo DRI'
> where it used to be.  Mesa will show as 3.3 beta instead of 3.4, but who
> the hell cares, it works....

One other thing I haven't seen posted, is the max 3D resolution, if you
start X above 1280x1024 acceleration won't work, you can switch up after
startup, or I least I can, and stiil run acceralted 3d in a window.  I
like to use glcthugha to test these things because it puts a huge strain
on all system resources.  Anyway, for the first time I key press a key in
glcthugha and it responds just shy of immediatlely, as opposed to several
seconds.

/edc

 
 
 

1. Ready - Sorta, kinda, I think, maybe

Ready to take another crack at installing Linux on this bastardized Compaq.  
The big issue this time around is PGP.  Gotta have it.  According to the
info at MIT, I have to use either Red Hat or Slackware non-glibc versions.

What are the issues?  Does the kernel have to be compiled with libc5, or
does libc5 just have to be available for PGP?  Is there something going on
with the other distros that keeps PGP from working?

All clues appreciated.
--
Ed Waldron
Naples, FL
Personal Home Page:  http://ed.waldron.home.att.net/
**********
Blue Plate Special:  T-Bone $.50 -- With meat $8.99

PGP Keys at ldap://certserver.pgp.com
ID: 0xFCD0815D
FP: FD06 4B59 786F B5BC E9DD ADEC F863 568F FCD0 815D
**********

2. help: can not compile hello.C using g++

3. v3 3000 + RH6.2 not working

4. 2.5.46 compile error

5. Trying to get my optiquest v73 and my v3-3000 to work

6. History editing in csh (man discrepancy?)

7. Linksys Wireless Card V3 in Redhat 7.3 laptop (it's kinda working)

8. Biostar M5ATA clock problem solved

9. Question on Apache 1.3.19/FilesMatch/Content Handler behavior - kinda long but please read

10. DRI very slow with Voodoo3 3000

11. DRI and Voodoo3 3000, direct rendering=yes

12. Corrupt Graphics, Voodoo 3000, DRI

13. FBSD4.5 + xf86 4.2.0 + ATI Radeon 7500 = DRI does not work