View Full Version : Catalyst 8.11 and Radeon 9600/9700 (R300)
tsuru
11-24-2008, 03:31 PM
Has anyone been able to run Catalyst 8.11 drivers on a Radeon 9600/9700 based machine yet? So far all my attempts with Ubuntu 8.11 and Gentoo have failed at module load time with errors like this:
[fglrx] Maximum main memory to use for locked dma buffers: 928 MBytes.
[fglrx:drm_alloc] *ERROR* [driver] Allocating 0 bytes
[fglrx:firegl_init_device_list] *ERROR* Out of memory when allocating device
heads
[fglrx:firegl_init_module] *ERROR* firegl_init_devices failed
I have read that issues with the AGP bus is causing it. Have there been any new developments on this front?
How much memory do you use? 32 or 64 bit? PAE enabled kernel?
tsuru
11-25-2008, 12:06 AM
How much memory do you use? 32 or 64 bit? PAE enabled kernel?
I'm on a 512 MB laptop 32-bit pentium-m. I'm not sure if PAE is enabled or not, it's the stock ubuntu intrepid kernel.
Stormking
11-25-2008, 01:42 AM
As mentioned in other threads, I've got the same problem. Hardware is a Mobility Radeon 9700 with 64MB. I'm running 32 Bit Ubuntu Intrepid with its standard kernel.
The problem appeared in the 8.10-Ubuntu-Special-Edition of fglrx and was not fixed in 8.11 (official installer).
Well some old Pentium M do not support pae, check
grep pae /proc/cpuinfo
If it does you can try the server kernel in the repository and see if it makes any difference.
tsuru
11-25-2008, 01:41 PM
Well some old Pentium M do not support pae, check
grep pae /proc/cmdline
I assume you meant /proc/cpuinfo... Either way PAE is not available. It appears I got a Dothan just before the 533 MHz bus w/ PAE enabled. :( Do the Catalyst drivers require PAE?
Yes, i meant that. Well it does not require it, it was just an option to try. My RV410 still runs with fglrx, not really optimal, but i can at least debug the fglrx packageing ;)
tsuru
11-25-2008, 03:06 PM
I see the RV410 also runs on PCI-E bus...sigh. Maybe it's time to play TAPS and just move on.
biasquez
11-26-2008, 02:31 AM
@Kano
i have same problem, ati mobility 9700 on ubuntu intrepid 8.10 (768 mb ram, 32 bit, pentium-M dothan)
grep pae /proc/cpuinfo -> nothing, cpu doesn't support PAE
Well you can compile another kernel for testing. Maybe 2.6.26 series or get one older 2.6.24 from hardy (don't forget the 2 header packages).
Stormking
11-26-2008, 06:00 AM
Well you can compile another kernel for testing. Maybe 2.6.26 series or get one older 2.6.24 from hardy (don't forget the 2 header packages).
So FGLRX needs PAE now? Even if I don't have 4 Gigs of RAM? Why?
I mean, my CPU supports it but currently, I am using Ubuntu's Generic Kernel which seems to be compiled without PAE.
No you don't need PAE, but that would be an easy test to try the other kernel. Now just fetch another one.
If you want you can try installing those 3 debs:
http://kanotix.com/files/kernel/kernel-update-pack-generic-28/kernel/
That's 2.6.28 from jaunty git - just without the -ub marker which is normally in the next jaunty kernel.
DocentDAMP
12-02-2008, 04:18 PM
Has anyone been able to run Catalyst 8.11 drivers on a Radeon 9600/9700 based machine yet? So far all my attempts with Ubuntu 8.11 and Gentoo have failed at module load time with errors like this:
[fglrx] Maximum main memory to use for locked dma buffers: 928 MBytes.
[fglrx:drm_alloc] *ERROR* [driver] Allocating 0 bytes
[fglrx:firegl_init_device_list] *ERROR* Out of memory when allocating device
heads
[fglrx:firegl_init_module] *ERROR* firegl_init_devices failed
I have read that issues with the AGP bus is causing it. Have there been any new developments on this front?
I have the same problem with Catalyst 8.11 on a Radeon Mobility 9700 on a 64-bit CPU (Athlon 64 3000+).
I hope this will be fixed in 8.12. Does ATI know of this problem?
linuxjacques
12-02-2008, 05:29 PM
same problem here with 8.11 on a thinkpad t42 Mobility 9600.
32-bit 2.6.26 kernel, 2GiB RAM, no PAE
scottial
12-04-2008, 11:12 AM
Same problem with 8.11 and Mobility Radeon 9700.
My device id (1002:4e50) is not included in driver header file fglrxko_pci_ids.h and in xorg driver fglrx_drv.so
I think AMD have missed same device id during driver compilation.
Well when you add it (to the h) manually it would run too. Just with a logo in the corner, but with help of an older driver (or from another install the /etc/ati dir) you can avoid that.
Stormking
12-11-2008, 01:49 AM
8.12 is out and still not working for me. Thank you, ATI.
suacy
12-15-2008, 07:28 PM
8.12 is out and still not working for me. Thank you, ATI.
Do you get the same errors with 8.12?
Just so I get this straight, R300 support is broken in 8.11 and 8.12.
poofyyoda
12-15-2008, 07:48 PM
Just so I get this straight, R300 support is broken in 8.11 and 8.12.
Not for me. I've got an AGP 9600 card going on my secondary comp without problems, while on my main one I've upgraded my old R300 card (x550 PCIe) to an HD 4650. The only real problems I face are some opengl regressions causing more corruption in wine games and also a native game: tileracer. (with 8.12)
For me the 9600 and fglrx, works quite well infact. I use 8.11 at the moment on it since 8.12 regressed some of my stuff :( , but 8.12 works fine otherwise
dkasak
12-15-2008, 07:53 PM
This all smells like the way ATI removed r200 support ... http://www.phoronix.com/scan.php?page=article&item=550&num=1
Initially, things broke, and not a single peep was heard from ATI on what the problem was. People complained. Please hypothesised. Then ATI 'officially' removed r200 support.
Is this what's happening now?
It's instructive to consider recent infighting in the Xorg dev community about xf86-video-radeonhd vs xf86-video-ati. The trusty xf86-video-ati driver supports EVERY version from r100 to r600. But ATI ( AMD, whatever ) are very keen to push the development of xf86-video-radeonhd, which only supports later radeons ( r500, r600 ). This is no coincidence. While ATI can't actively break open-source support for earlier cards, they can at least try to entice developers across to a driver which doesn't support older cards. And with all the changes on the horizon ( GEM, UXA, DRI2 etc ), I expect xf86-video-radeonhd will be quickly updated, while users of older cards will have to stick with 'legacy' drivers.
Ren Höek
12-19-2008, 05:10 AM
Same Problem with
Catalyst 8.12 (fglrx 8.561)
Mobility Radeon 9700
Ubuntu 8.04
Would be really sad, if they stop support this way. If they can't fix the driver they should at least fix the support list...
Stormking
12-19-2008, 05:53 AM
Do you get the same errors with 8.12?
With 8.10-Ubuntu-Special-Edition (xserver-1.5-compatible-beta), official 8.10, 8.11, 8.12 and 9.1beta.
Just so I get this straight, R300 support is broken in 8.11 and 8.12.
I can only speak for myself and I have a Mobility Radeon 9700. But the same error message has been reported by people with Mobility Radeon 9600 and Radeon 9500.
[Knuckles]
12-27-2008, 01:00 PM
Same problem here, with a Mobility Radeon 9600. C'mon ATI, if you're going to deprecate, at least warn users first!
Everywhere says that the current drivers should support R300. SO WHY!?
Bah, have to go back to openSUSE 11.0 on one of my laptops because of this crap.
RealNC
12-27-2008, 02:19 PM
R300 works better with the open source radeon driver anyway. I would feel lucky if I were you :D
yotambien
12-28-2008, 08:49 AM
R300 works better with the open source radeon driver anyway. I would feel lucky if I were you
No, it does not. While it is true that the open source driver used to be more stable than AMD's binary, that changed since the ~May/June releases (before that fglrx was unbearable). Now the differences are marginal, i.e. they both crash just rarely. On the other hand, 3D performance with radeon sucks compared to fglrx, and it also lacks some features needed to correctly run some programs (*). On top of that, fglrx's Powerplay is simply absent in the open souce driver, which makes a huge difference for laptop users.
The only scenario I can imagine prefering the oss driver over fglrx--religion aside--is when using a composite desktop. I don't, so fglrx works well for me.
Having said this, the Windows version of the driver is miles ahead of fglrx in every sense.
(*) No, this is not FUD. I know of at least a program I happen to use that doesn't run as it should with the oss driver, and many more that keep falling to software rendering.
Just try Zero Ballistics. But the system seems to be a bit slow for that game anyway.
[Knuckles]
12-28-2008, 10:31 AM
yotambien: For R300 I'm stuck with catalyst 8.9, so all that goodness is not available for me.
I'm gonna do some tests with the radeon driver, at least for XBMC it seems to be comparable to fglrx.
As I wanted to use this pc as a media center only, if it has good performance and works with XBMC, it's golden.
But ATI could at least inform users that R300 is not working, it's fsckin lame to go to their site and follow that wizard-thingy where you select exactly your hardware versions just so they can send you to the same generic linux driver every time, and it doesn't even WORK.
yotambien
12-28-2008, 03:34 PM
;56530'] For R300 I'm stuck with catalyst 8.9, so all that goodness is not available for me.
I didn't want to sound over excited about fglrx. The 'goodness' I was talking about is better 3D support, performance and powerplay, as compared to the oss driver. This doesn't mean that it is good in absolute terms, you only have to compare it with the Windows version, or try to hook a second monitor, or look at all the problems people seems to have when going composite or playing wine games.
In any case, I'm using what's available in Debian unstable, which happens to be whatever version AMD released in July or perhaps even June. Maybe I'm luckier than you for it works well here.
I'm gonna do some tests with the radeon driver, at least for XBMC it seems to be comparable to fglrx.
I'd be interested in knowing how it goes (I've got the same hardware). I don't think fglrx is any better than radeon for video playback.
But ATI could at least inform users that R300 is not working, it's fsckin lame to go to their site and follow that wizard-thingy where you select exactly your hardware versions just so they can send you to the same generic linux driver every time, and it doesn't even WORK.
That's for sure. Especially when you think that the mobility radeon 9600 and 9700 cards are from 2003 and 2004, respectively, and only this year they started to provide something acceptable. I don't know what I would've done during these last 3 years if the oss driver didn't exist.
[Knuckles]
12-28-2008, 05:15 PM
The oss driver works ok with multiple monitors -- even better than fglrx in my opinion.
I've been testing with XBMC, and it works fine, so i'll be keeping the oss one instead of using an old kernel and X.org just to please his majesty the fglrx.
I'm getting a bit of tearing in the video, I've got to investigate it more (unfortunately I didn't test it that far with fglrx, so I can't compare if there's a tearing problem with it too...), since XBMC uses opengl to output.
My media center project is now on hold for a while until I can get color with s-video tv-out on my crappy sony tv (I think it's one of those where i'll have to rig a scart converter, but I gotta look more into it before cracking it open).
yotambien
01-14-2009, 05:39 PM
So, what's going on with R300 cards fglrx? Is this generation of cards not supported anymore or 'just' a persistent bug?
[Knuckles]
01-15-2009, 04:25 AM
So, what's going on with R300 cards fglrx? Is this generation of cards not supported anymore or 'just' a persistent bug?
I'd say they are efectively ignoring it, in hopes it'll go away.
When will R300 support be fixed on your dr-*NA NA NA CAN'T HEAR YOU NA NA*...
King InuYasha
01-29-2009, 08:02 PM
Still broken in Catalyst 9.1
"(EE) No screens detected"
That message has been going on ever since I started using Ubuntu intrepid.
And no, I can't downgrade. The CD drive broke and the laptop doesn't support booting from USB.
[Knuckles]
01-30-2009, 05:43 AM
Still broken in Catalyst 9.1
"(EE) No screens detected"
That message has been going on ever since I started using Ubuntu intrepid.
And no, I can't downgrade. The CD drive broke and the laptop doesn't support booting from USB.
You might wanna try taking your hdd out and installing in a desktop. If the hdd is sata, no problems connecting, if it is pata, 2.5" hdd boxes for connecting via usb are very very cheap.
Just a suggestion, of course, I have a lot of old laptops, and that's how I normally solve the problem :cool: .
King InuYasha
01-31-2009, 01:29 PM
;60429']You might wanna try taking your hdd out and installing in a desktop. If the hdd is sata, no problems connecting, if it is pata, 2.5" hdd boxes for connecting via usb are very very cheap.
Just a suggestion, of course, I have a lot of old laptops, and that's how I normally solve the problem :cool: .
It is PATA, but I think I have a suspicion that its actually Mesa 7.2 causing the problems with fglrx. My Fedora 10 machine can use fglrx just fine, and that has Mesa 7.3rc1. Ubuntu Hardy had Mesa 7.1 I think, ehh. Now, I'm attempting to use prevu tool to build Mesa 7.3 for Intrepid.
thinkpad42
02-02-2009, 01:16 AM
i was able to get catalyst 9.1 working on my thinkpad with a mobility 9600 (asic 4e50) running on intrepid.
I had to add to the xorg.conf under the device section
ChipID 0x4E56
This makes the driver think you're running on a mobility 9550. i have also changed the chipid to match a mobility 9800, mobility x300, and a desktop 9600. The desktop 9600 almost worked, but after logging in the screen would change to black and i'd have to hard cycle it. the mobility 9800 and mobility x300 worked but i think at the same level as the mobility 9550.
It was actually took a lot of my sunday to get it working. Also a lot of reboots were needed. haven't used fglrx since hardy. if anybody needs additional help getting it going let me know.
playing a video seems to cause an x restart.
vBulletin® v3.8.4, Copyright ©2000-2010, Jelsoft Enterprises Ltd.