Not all Unhappy campers on ati

[deanjo]

Then show me an ati agp card working with it.

My bad I was thinking of a nvidia card (which does work with the gart driver).

[aceman]

Will all you guys scoring points on each other calm down for a minute, please - I'm not quite finished here yet.

I tried this: I updated as follows (from Slackware-current packages, dated March)
libdrm: 2.4.20 --> 2.4.23
pixman: 0.16.5 --> 0.20.2
xorg-server; 1.7.7 --> 1.9.5
mesa: 7.8.1 --> 7.9.2

and X wouldn't start

(EE) module ABI major version (6) doesn't match the server's version (8)
(EE) Failed to load module "vesa" (module requirement mismatch, 0)
(EE) Failed to load module "fbdev" (module does not exist, 0)


What did I leave out, or do wrong? Do I have to compile manually? I'm not keen on the
Option IgnoreABI
approach to this, and it probably leaves things tolerating each other more than working together.

You must also install the driver packages (xorg-video-vesa, xorg-video-ati, etc) that come with the updated server in Slackware-current. I am also using it with ATI card and it works fine. Or compile the driver yourself after you installed xorg server 1.9.

[Kano]

You may have to play with agp aperture size in bios.

[Kano]

One other trick i needed for a radeon 9700/9800 agp card was that it only got stable after +0.1v of extra agp voltage.

[AnonymousCoward]

Why are you dumping this Windows crap in an open source forum? Have you no sense of decency?
You're really reinforcing the stereotype of the bad mannered German.

[business_kid]

You must also install the driver packages (xorg-video-vesa, xorg-video-ati, etc) that come with the updated server in Slackware-current. I am also using it with ATI card and it works fine. Or compile the driver yourself after you installed xorg server 1.9.

Thanks for thinking of me, aceman. I'll grab those from slackware-current, which seems to have overhauled video stuff recently. I have traced the ABI version error somewhat, and what's happening I gather is this:
1. xorg-server guys are bumping up required ABI versions at a rate of knots.
2. xf86-video-ati just checks what abi it's supposed to be
3. On my box it's getting that wrong, probably due to some outdated include file.

I have discovered /usr/include/xorg/xf86Module.h which requires video ABI 10 but it's getting 6 from somewhere. I'll do a methodical search today. The one fear I have is that mesa is picking it up. That's a mighty chunk to be trawling through.

[business_kid]

hey kano you are really helpfull i only pay with that cpu to agp stuff because of your writing so thank you very much
in general your knowledge basement is just epic
i allready put +0.1v on agp voltage,,,,
and there is no stability problem without extra voltage and with....
the hardware of this system runs just great if the driver works the opteron 144-1,8ghz runs hat 2,7ghz and the hd4650 is very fast...
there is just a ugly software bug or maybe a software fixable hardware bug only fixed on windowsXP..
"agp aperture size in bios."
i really brudeforce all kind of settings in the bios there is no triggering for that stuff or what do you mean?

If it's any comfort to you, before post #1(re windows and hd4650) I did quite some research, and found the hd4650 agp was never successfully got going by anyone except in XP, and he didn't really know how he did it. When I shoved in the nv mx-440 xp found it straight away on setup, and loaded a driver. AGP is yesterday, it seems. The difference was: mx-440 was out when nvidia were still working out of someone's shed (pre 2001), whereas the 4650 dates from after support was dropped.

[Kano]

Best get rid of all agp boxes, even 30? cpus are faster than those you have got.

[agd5f]

AGP has always been problematic regardless of the OS or GPU. You can disable AGP in the open source driver with radeon.agpmode=-1

[agd5f]

i think this option is the default for all agp cards....

It's not. The default is to attempt to use AGP.

on linux the system boot up and is software rendering only but it works stable and without problems.

do you think "radeon.agpmode=-1" makes it better than that ?

With agpmode=-1, you will still get acceleration, it just uses the on-chip gart rather than the gart provided by the AGP bridge which is slower than AGP on AGP systems. If you are not getting acceleration, most likely the module has failed to load properly for some reason. Check dmesg output.