Unsupported AGP-bridge on VIA VT8633

Unsupported AGP-bridge on VIA VT8633

[Stian Jordet]

Hello,
I have an ASUS CUV266-DLS motherboard, with an agp-bridge which I think
is supposed to be supported. But it isn't. This is with 2.4.20-rc1, but
I've tried 2.5.47 as well, and haven't seen anything that should have
changed in 2.4.20-rc2 or 2.5.48. 

Based on agpgart interface v0.99 (c) Jeff Hartmann
agpgart: Maximum main memory to use for agp memory: 439M
agpgart: Unsupported Via chipset (device id: 3091), you might want to
try agp_try_unsupported=1.
agpgart: no supported devices found.

I have tried with agp_try_unsupported=1, but no luck. I have seen that
in pci_ids.h this device id is defined to vt8633, so I suppose it should
work. I would really like to know why it doesn't...

Regards,
Stian Jordet

Re: Unsupported AGP-bridge on VIA VT8633

[Dave Jones]

On Thu, Nov 21, 2002 at 11:01:07PM +0100, Stian Jordet wrote:
 > Based on agpgart interface v0.99 (c) Jeff Hartmann
 > agpgart: Maximum main memory to use for agp memory: 439M
 > agpgart: Unsupported Via chipset (device id: 3091), you might want to
 > try agp_try_unsupported=1.
 > agpgart: no supported devices found.
 > 
 > I have tried with agp_try_unsupported=1, but no luck.

You tried agp_try_unsupported=1 as a modprobe argument,
not as a boot time argument right ?
Quite a few people seem to fall into this trap.

When I get chance, I'll make that a boottime arg too.

		Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: Unsupported AGP-bridge on VIA VT8633

[Stian Jordet]

Quoting Dave Jones <davej@codemonkey.org.uk>:

> On Thu, Nov 21, 2002 at 11:01:07PM +0100, Stian Jordet wrote:
>  > Based on agpgart interface v0.99 (c) Jeff Hartmann
>  > agpgart: Maximum main memory to use for agp memory: 439M
>  > agpgart: Unsupported Via chipset (device id: 3091), you might want to
>  > try agp_try_unsupported=1.
>  > agpgart: no supported devices found.
>  > 
>  > I have tried with agp_try_unsupported=1, but no luck.
> 
> You tried agp_try_unsupported=1 as a modprobe argument,
> not as a boot time argument right ?
> Quite a few people seem to fall into this trap.
> 
> When I get chance, I'll make that a boottime arg too.

You were not really clear here. I tried it as a boot-time argument, because I
have agp-support compiled in. But I guess I could and should try it as a module.
I'll do that now. But why do I have to use agp_try_unsupported=1?

Regards,
Stian Jordet

Re: Unsupported AGP-bridge on VIA VT8633

[Dave Jones]

On Thu, Nov 21, 2002 at 11:20:31PM +0100, Stian Jordet wrote:

 > You were not really clear here. I tried it as a boot-time argument, because I
 > have agp-support compiled in. But I guess I could and should try it as a module.

Yup. Then do a `modprobe agpgart agp_try_unsupported=1'

 > I'll do that now. But why do I have to use agp_try_unsupported=1?

Because if it works, we can then add it to the ID table.

	Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk

Re: Unsupported AGP-bridge on VIA VT8633

[Stian Jordet]

tor, 2002-11-21 kl. 23:40 skrev Dave Jones:
> On Thu, Nov 21, 2002 at 11:20:31PM +0100, Stian Jordet wrote:
> 
>  > You were not really clear here. I tried it as a boot-time argument, because I
>  > have agp-support compiled in. But I guess I could and should try it as a module.
> 
> Yup. Then do a `modprobe agpgart agp_try_unsupported=1'
> 
>  > I'll do that now. But why do I have to use agp_try_unsupported=1?
> 
> Because if it works, we can then add it to the ID table.

It works, i think. I get this message when I load it:

Linux agpgart interface v0.99 (c) Jeff Hartmann
agpgart: Maximum main memory to use for agp memory: 439M
agpgart: Trying generic Via routines for device id: 3091
agpgart: AGP aperture is 64M @ 0xf8000000

Thank you very much. I'm very sorry if this was a lame question.

Thanks.

Regards,
Stian Jordet

Re: Unsupported AGP-bridge on VIA VT8633

[Dave Jones]

On Thu, Nov 21, 2002 at 11:56:23PM +0100, Stian Jordet wrote:
 > >  > I'll do that now. But why do I have to use agp_try_unsupported=1?
 > > Because if it works, we can then add it to the ID table.
 > It works, i think. I get this message when I load it:
 > Linux agpgart interface v0.99 (c) Jeff Hartmann
 > agpgart: Maximum main memory to use for agp memory: 439M
 > agpgart: Trying generic Via routines for device id: 3091
 > agpgart: AGP aperture is 64M @ 0xf8000000

And it survives a 3d app / testgart run ?
 
 > Thank you very much. I'm very sorry if this was a lame question.

No problem. I'll add this ID to the pending patch I have
for Linus to add various other GARTs.

		Dave
-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: Unsupported AGP-bridge on VIA VT8633

[Stian Jordet]

fre, 2002-11-22 kl. 00:14 skrev Dave Jones:
> On Thu, Nov 21, 2002 at 11:56:23PM +0100, Stian Jordet wrote:
>  > >  > I'll do that now. But why do I have to use agp_try_unsupported=1?
>  > > Because if it works, we can then add it to the ID table.
>  > It works, i think. I get this message when I load it:
>  > Linux agpgart interface v0.99 (c) Jeff Hartmann
>  > agpgart: Maximum main memory to use for agp memory: 439M
>  > agpgart: Trying generic Via routines for device id: 3091
>  > agpgart: AGP aperture is 64M @ 0xf8000000
> 
> And it survives a 3d app / testgart run ?

Yes, it does. I have a Radeon 9700 Pro, and I have been running XFree86
cvs to get 2d support. This has worked perfect for me. But yesterday ATI
launched their binary drivers, and I wanted to give them a try. Then I
had to get agp working. And after I got your mail, I thought everything
was going just fine. Everything loaded fine, but the computer locked up
everytime something 3d appeared. At this time I wasn't sure what I
should blame. The driver, og the agpgart. So I found my old Radeon
All-in-Wonder, with good DRI-support, and now it works perfect :)

So to make a long story short (too late), the agpgart seems to work
perfect, altough my Radeon 9700 Pro didn't.

>  > Thank you very much. I'm very sorry if this was a lame question.
> 
> No problem. I'll add this ID to the pending patch I have
> for Linus to add various other GARTs.

Great! Thank you :)

Regards,
Stian Jordet

(Off to play more tuxracer before I switch cards again)

Re: Unsupported AGP-bridge on VIA VT8633

[Stian Jordet]

fre, 2002-11-22 kl. 00:14 skrev Dave Jones:
> On Thu, Nov 21, 2002 at 11:56:23PM +0100, Stian Jordet wrote:
>  > >  > I'll do that now. But why do I have to use agp_try_unsupported=1?
>  > > Because if it works, we can then add it to the ID table.
>  > It works, i think. I get this message when I load it:
>  > Linux agpgart interface v0.99 (c) Jeff Hartmann
>  > agpgart: Maximum main memory to use for agp memory: 439M
>  > agpgart: Trying generic Via routines for device id: 3091
>  > agpgart: AGP aperture is 64M @ 0xf8000000
> 
> And it survives a 3d app / testgart run ?

Bah, I'm not sure anymore. I have two motherboards. Asus CUV-266DLS and
Rioworks SDVIA. Both are dual cpu P3 motherboards. The first has a Via
Apollo Pro 266 chipset, while the Rioworks has a Via Apollo Pro 133A. I
have two graphics-adapters. ATI Radeon All-in-Wonder, and one ATI Radeon
9700 Pro.

With the SDVIA board, I get dri to work with both vga-cards. (On the
Radeon 9700 using the ATI-binary drivers.)

On the Asus, I have to load agpgart with agp_try_unsupported=1. Then I
get the old Radeon AIW to work perfectly with dri, but with the 9700, my
box crashes hard everytime I try anything 3d related.

But this can probably be blamed on the ATI-binary drivers? I just wanted
to let you know.

Best regards,
Stian Jordet

Re: Unsupported AGP-bridge on VIA VT8633

[Rogier Wolff]

On Thu, Nov 21, 2002 at 11:56:23PM +0100, Stian Jordet wrote:
> tor, 2002-11-21 kl. 23:40 skrev Dave Jones:
> > On Thu, Nov 21, 2002 at 11:20:31PM +0100, Stian Jordet wrote:
> > 
> >  > You were not really clear here. I tried it as a boot-time argument, because I
> >  > have agp-support compiled in. But I guess I could and should try it as a module.
> > 
> > Yup. Then do a `modprobe agpgart agp_try_unsupported=1'
> > 
> >  > I'll do that now. But why do I have to use agp_try_unsupported=1?
> > 
> > Because if it works, we can then add it to the ID table.
> 
> It works, i think. I get this message when I load it:
> 
> Linux agpgart interface v0.99 (c) Jeff Hartmann
> agpgart: Maximum main memory to use for agp memory: 439M
> agpgart: Trying generic Via routines for device id: 3091
> agpgart: AGP aperture is 64M @ 0xf8000000

Hi Dave, Stian,

"Same here": The module loads...

However, I have the impression that it doesn't work.  I first noticed 
that my Xserver was taking 50% of my CPU while playing videos. This
is much less on all other computers that I tried that on. 

Turns out that "write bandwidth" to video memory is less than 40Mb
per second. On another system with the same video card, but with a 
supported AGP bridge, I get almost 200Mbyte per second. I'll go and
dig for "testgart" and see what that proves.... 

			Roger. 

-- 
** R.E.Wolff@BitWizard.nl ** http://www.BitWizard.nl/ ** +31-15-2600998 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
* The Worlds Ecosystem is a stable system. Stable systems may experience *
* excursions from the stable situation. We are currently in such an      * 
* excursion: The stable situation does not include humans. ***************