AMD 8.37.6 -- The no R600 or AIGLX Support Version

[Michael]

The AMD/ATI fglrx 8.37.6 driver is now available... It contains X server 1.3 naming fix, AMDCCCLE v1.0 (adds translations and the 3D page), RHEL5 support, and some random fixes.

https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/linux_8.37.6.html

It was a year ago that AMD/ATI had delivered the fglrx 8.25.18 display driver, which corrected a bug causing issues with the Radeon X1000 512MB video cards, added new product support, and squashed a variety of other bugs. Well, today AMD delivered the 8.37.6 Linux display driver. Affecting this driver the most is support for X server version 1.3 and appended several other fixes along with a slight upgrade to the AMD Catalyst Control Center Linux Edition. While no AIGLX support was added or no new features introduced, the upgrade should be worth the time

http://www.phoronix.com/vr.php?view=10069


FEDORA USERS, READ THIS
Okay, here's the scoop on Fedora 7. WIth AMD fixing the X 1.3 problem, the drivers will build. However, when you install the RPMs for Fedora 7, aticonfig will crash and will remove your xorg.conf (though the original is renamed). If you manually update your xorg.conf and then restart X, fglrx will fail to load and you'll find a backtrace in your Xorg.0.log. So if you need the fglrx drivers to survive, don't upgrade to Fedora 7 just yet. Stay tuned to the forums for more information. This issue with 8.37 should not cause problems with Fedora Core 6 or previous versions.

[yoshi314]

it amuses (and embarasses) me to no end when i look at the linked page (on the ati site) and i see that i need Qt2 to use the panel and only Xorg <=7.1 is supported (and glibc 2.2-2.3 ) :]

does anybody from ati actually read this page ? :D

oh well. talk about timely r600 support. it was supposed to be in a couple of releases ago.

i think ati is doing some kind of psychological warfare on its linux users. i am a patient ati user but i've already given up hope for performance improvement and aiglx. i just want xorg and kernel compatiblity, since i cannot afford to buy a different video card, at least right now.

i can live with awful video tearing and system lockups. i guess i've given up on ati with this release.

[d2kx]

I like it how you always name the new drivers, keep this and think of more negative aspects when they bring us AIGLX, please :)

The release notes look damn small, but I can really see they are improving they linux driver. AIGLX Support and the new OpenGL driver which should give us more performance and we are on the same level with nVidia. Did they have the feature for AA/AF setting in the driver?

[yoshi314]

we should bet on the name for the next release.

i'm going for "nothing new edition" :]

[Michael]

I like it how you always name the new drivers, keep this and think of more negative aspects when they bring us AIGLX, please :)

The release notes look damn small, but I can really see they are improving they linux driver. AIGLX Support and the new OpenGL driver which should give us more performance and we are on the same level with nVidia. Did they have the feature for AA/AF setting in the driver?

All of the real details and changes can be found @ http://www.phoronix.com/vr.php?view=10069

[d2kx]

+ CCC offers AA/AF settings
+ CCC translation updates
+ Xorg Server 1.3 support
+ RHEL5 support
+ proper Vsync swapping
+ crashbug fix for logging of from X
+ Xv cropping bug fixed

I already spread the word about it in some german forums and on the Sidux forum, so I can have my new Sidux 2007-02 with Xorg Server 1.3 running tomorrow :)

Hopefully fglrx 8.38 comes in 2 weeks ^^ Not impossible.

[slacker]

mm, even 8.37 doesnt compile on kernel 2.6.22rc3.
I tried the patch here: http://www.phoronix.net/forums/showthread.php?t=2849
but no luck. any ideas?
thx

[yoshi314]

+ CCC translation updates the polish one is awful, though :]

[Raven3x7]

So can anyone confirm they fixed the xv playback issue that cropped up in 8.36.5?

[yoshi314]

So can anyone confirm they fixed the xv playback issue that cropped up in 8.36.5?which one? because the tearing got reduced but it's still there. it only appears HD resolution videos.

[the-me]

I can't build the module with Linux 2.6.21. It says me, that it's using a GPL-incompatible function called pvirt_ops (or so).

Is there any patch available?

Here's my build log:
http://nopaste.linux-dev.org/?1529

[Zhenech]

Compile your kernel with CONFIG_PARAVIRT=n

[Michael]

the polish one is awful, though :]

I agree, the Polish amdcccle translations aren't the best.

[Zhenech]

Ohkay, XVideo is still broken for me (Thinkpad Z61m, X1400).

[lowlands]

FEDORA USERS, READ THIS
Okay, here's the scoop on Fedora 7. WIth AMD fixing the X 1.3 problem, the drivers will build. However, when you install the RPMs for Fedora 7, aticonfig will crash and will remove your xorg.conf (though the original is renamed). If you manually update your xorg.conf and then restart X, fglrx will fail to load and you'll find a backtrace in your Xorg.0.log. So if you need the fglrx drivers to survive, don't upgrade to Fedora 7 just yet. Stay tuned to the forums for more information. This issue with 8.37 should not cause problems with Fedora Core 6 or previous versions.

Hi Michael,

Do you have any information (that you can disclose) if ATI soon will be releasing an updated driver that will fix this issue? Soon meaning something like in the next week or so.

For ATI: Fedora Core 6 has millions of users. It looks like Fedora 7 will be even bigger. I sincerely hope (and respectuflly encourage you) that you will issue a new release that will fix the problems on Fedora 7 at your earliest convenience.

Thanks,
Patrick

[Michael]

Hi Michael,

Do you have any information (that you can disclose) if ATI soon will be releasing an updated driver that will fix this issue? Soon meaning something like in the next week or so.

For ATI: Fedora Core 6 has millions of users. It looks like Fedora 7 will be even bigger. I sincerely hope (and respectuflly encourage you) that you will issue a new release that will fix the problems on Fedora 7 at your earliest convenience.

Thanks,
Patrick

AMD will not be deviating from their standard release cycle in order to provide a fix for a distribution that isn't even officially supported by AMD.

8.38 in June is the best case scenario.

[Svartalf]

oh well. talk about timely r600 support. it was supposed to be in a couple of releases ago.


And you'll wait a while longer, I suspect... R600 support is (currently...) only available with the Orca codebase- which means they've got a lot of tuning and optimization work to do before that will happen.

[Svartalf]

AMD will not be deviating from their standard release cycle in order to provide a fix for a distribution that isn't even officially supported by AMD.

8.38 in June is the best case scenario.

Which is a bit odd, considering that the distribution in question is the BASIS for the distributions they DO support- they'd be better served to just break down and support Fedora in the first place... :D

[gfxdrone]

This one is almost working, gets closer, some users actually did get it installed, though the deb created, fgrlx-driver (debian sid) is kicking out an error:

dpkg: regarding fglrx-driver_8.37.6-1_i386.deb containing fglrx-driver:
xserver-xorg-core conflicts with fglrx-driver
fglrx-driver (version 8.37.6-1) is to be installed.
dpkg: error processing fglrx-driver_8.37.6-1_i386.deb (--install):
conflicting packages - not installing fglrx-driver
Errors were encountered while processing:
fglrx-driver_8.37.6-1_i386.deb


Oddly enough however, after editing xorg.conf to use fglrx, one user got x to start and show that it's using 8.37.6 as its video driver, but another user failed to get x to start at all. So we dumped live testing and will wait to see what caused that issue.

All in all, more really bad work from ATI, and unless AMD comes out with a VERY strong statement about a serious change in directions by their new subsidiary ATI re Linux drivers, either Opening the driver source, or, at worst, opening the hardware specs so Linux developers can work on their own, ATI is permanently off our acceptable hardware list. So AMD's announcment better be really good. Not some marketing happy talk, something real. This is total b####t as far as I'm concerned, it's a joke, pathetic, and ATI sucks.

Whether ATI is incompetent, indifferent, or actually is deliberately pulling back on Linux video driver development as some have suggested makes no difference to end users, the result is the same, you can't count on them, their quality is garbage, and their support non-existent. Only a fool would buy ati today.

And I definitely am not interested in hearing ATI blame developers who are writing their installers for them for free for any errors, that's total nonsense, ATI needs to test their code that they put out, even when they do get people to write it for them for free, this approach is disgusting, it's almost incomprehensible to me how a company could do this bad a job without actively trying... with one exception, unless they are using bad outsourced programmers, which wouldn't surprise me to see at all, since one of the hallmarks I've seen of bad outsourced guys is terrible to non-existent testing and quality control.

[Svartalf]

This is total b####t as far as I'm concerned, it's a joke, pathetic, and ATI sucks.

I wish they'd listen to us here- they're NOT doing themselves ANY favors with this mess.

[gfxdrone]

On the remote chance anyone from ATI cares, here's the xserver-xorg-core version information:

apt-cache policy xserver-xorg-core
xserver-xorg-core:
Installed: 2:1.3.0.0.dfsg-5
Candidate: 2:1.3.0.0.dfsg-5
Version table:
*** 2:1.3.0.0.dfsg-5 0
500 http://ftp.us.debian.org unstable/main Packages
100 /var/lib/dpkg/status

If AMD wants to play with the big kid on the block, Intel, they'd better get this driver nonsense sorted out as soon as possible or all those lovely desktop rollouts aren't going to be running amd/ati, they'll be running Intel.

And if AMD needs some lessons on how to do a good modern hardware based business that also is open and communicates with its users, check out SUN, especially Jonathon Schwartz's blog (http://blogs.sun.com/jonathan/). AMD is also messing up in this regard with its total failure to communicate in any meaningful way to its users, even if they do get some word to phoronix, as if we should feel so grateful that they even deign go speak to their users in that way.

[biAji]

I upgrade my FC6 to F7 last night..

no aiglx support, bugs in F7...So, It seemed this release means nothing to me~~ :(

[yoshi314]

this is the second driver release i might call "8.35 does it better". if it weren't for xorg-server-1.3 i'd be back on 8.35, again.

now i'm really curious what excuses will ati come up with in the development cycle article, which will come out....someday.

it's hard to believe that any company would be actually able to release mediocre drivers for three years straight. they don't do this in the windows world, but linux is a different story.

there must be something very wrong groing on with the fglrx development.

[slacker]

Ohkay, XVideo is still broken for me (Thinkpad Z61m, X1400).

Same here: can't watch a fuc**n movie!!
I'm so pissed off :mad:
Dell Inspiron 6400 X1400 also.
How is it possible that newer drivers are worse than older??? :confused:
bleah, i'm disgusted..

[DirtyHairy]

Just for the notes, things have improved for me a bit on a Tinkpad T60 (X1400, Gentoo stable with xorg-server 1.2.0) with 8.37 . While xv simply was gone with 8.36, it has reappeared for me with 8.37. It is still broken though, I get colored stripes, the xv window isn't properly hidden by other windows, and when maximized, I get an odd "hanging" effect in playback with the video stopping every second and jumping back a frame (looks like double buffering issue). On the other hand, vsync with opengl now works for me, which it never did before making opengl finally an emergency replacement for xv (which it wasn't before since I find tearing in playback very annoying).

Still, I switched back to 8.35 which is the best-working version for me so far.

[sundown]

I, too, have the video playback issues. And we were so keen on this release after its delay :)

[slacker]

sundown which video card do you have? X1400? On a laptop?
Could it be a specific problem? Maybe ATI doesn't know about it, that's why they didn't fix it :confused:

[curzondax]

i get exactly the same fps numbers with this version..

[sundown]

Mobility Radeon X1600.

[bimbo]

This seems to work fine on my Thinkpad T60 (X1400). As far as I can tell, Video display is fine (though I don't have any HD content to test with but SD works fine) and amdcccle can switch screen config on the fly...

I stand corrected, tearing still happens, but it's only visible in fast scenes.

But at least it compiles cleanly for 2.6.21.3 (once paravirt is deactivated)

[d2kx]

curzondax, this is because the new OpenGL driver isn't integrated yet.

[lenrek]

I manage to install without any problem. So far everything works fine.


[fglrx] Maximum main memory to use for locked dma buffers: 555 MBytes.
[fglrx] module loaded - fglrx 8.37.6 [May 25 2007] on minor 0
ACPI: PCI Interrupt 0000:01:05.0[A] -> GSI 17 (level, low) -> IRQ 16
[fglrx] total GART = 130023424
[fglrx] free GART = 114032640
[fglrx] max single GART = 114032640
[fglrx] total LFB = 134217728
[fglrx] free LFB = 120320000
[fglrx] max single LFB = 120320000
[fglrx] total Inv = 0
[fglrx] free Inv = 0
[fglrx] max single Inv = 0
[fglrx] total TIM = 0



$> uname -a
Linux virgo 2.6.22-rc3 #2 Mon May 28 15:16:16 SGT 2007 i686 athlon-4 i386 GNU/Linux



$> lspci | grep VGA
01:05.0 VGA compatible controller: ATI Technologies Inc ATI Radeon XPRESS 200M 5955 (PCIE)

[gfxdrone]

Well, after reading the ATI / AMD press release disguised as an 'article' here on phoronix, several things become immediately obvious: forget about open drivers. Forget about any of the issues getting fixed. Forget about ATI actually ever supporting users. Forget about ATI ever testing and releasing working drivers, and taking responsibility for their drivers.

To summarize, forget about ATI completely in the Linux kernel, it's over, if the free drivers don't cut it for you, expect a non stop stream of breaks, kernel and xorg failures, and whatever else the future brings.

I can't believe that phoronix was suggesting that something of interest was about to be announced re AMD/ATI, what a joke, what we get is a weak rehash of nothing you couldn't figure out yourself by looking at ATI's junky driver releases.

And that final comparison to nvidia drivers, someone here is drinking the happy juice, nvidia stuff works, they release when the driver is working and ready, after a beta release cycle to catch bugs, usually it doesn't even require a patch, or they release when they need to fix stuff for xorg or new kernel, it's released very quickly, and it works.

I think phoronix is going a bit overboard to try to not offend their connections to ATI, this article was one of the most surreal fantasies I've read in a while. You really should have left that nvidia comparison off the end, you know that's not true, why did you write it? nvidia drivers are better in every way, they are released and usually work, their installer is better, they take responsibility for bugs to some degree at least, they continue to support legacy cards with active new driver releases, and so on. Just because ATI pushes out a bad product on a schedule doesn't make their process good, it makes it bad in fact, the mark of bad software is exactly this: not release when ready, after beta release to catch bugs, but release when the schedule says you should. No wonder ATI drivers suck, now it's totally clear, their process sucks, and obviously their management must be bad too to make such a typically pointy headed boss decision.

Conclusion: if this is the AMD announcement, and if this is the best they can do, then AMD might as well write off all large scale corporate / organizational rollouts today, only an idiot IT guy would buy AMD / ATI if they're going to roll out Linux desktops.

And regular users, be warned, this article clearly states: no changes, more of the same, bad testing will continue, bad methods will not be fixed, and nothing will be opened. So forget ATI if you have any sense at all.

The VERY BEST we can now hope for, and I mean the very best, I don't expect it at all though, is for AMD/ATI to at least open their hardware specs so that free driver authors can work with fewer restrictions.

I have been a loyal AMD fan for years, but this is a serious mistake on their part, and is going to end up losing them the Linux market goodwill they built up over the past years, unless, again, they come out with a bombshell and open their processes, their drivers, and their hardware specs.

I feel so sorry for ATI card users who want to run Linux, what a pity, and how depressing. And phoronix should be ashamed of themselves for printing such utter nonsense, I hope they at least paid you for the advertising space...

AMD is now making the exact mistake Sun Microsystems made a few years ago, failing to adapt to new models, pursuing strategies that worked years ago, but which are not nearly as effective today. And they can expect to experience the same thing Sun did, until they learn their lesson, drops in share prices, marketshare, and failure to expand their market. However, Sun shows the way, you can change as a company, but it takes determination and the willingness to take risks. ATI is clearly hopeless and will not change until AMD forces change, and AMD will only force this change when they lose enough market share and their stock price drops enough to make taking the chance to lead worth it.

[d2kx]

Debian/sid: had to install xserver-xorg-core 1.3.0.0-6 from the Incoming folder first for depency reasons, only started with aticonfig doing the Xorg configuration (my own only got me a black screen). Then it hangs on VT console change (was fixed for me in 8.36.5).

Wanted to try the new Fedora 7, but it doesn't work.

Excellent.

[lenrek]

Well, after reading the ATI / AMD press release disguised as an 'article' here on phoronix, several things become immediately obvious: forget about open drivers. Forget about any of the issues getting fixed. Forget about ATI actually ever supporting users. Forget about ATI ever testing and releasing working drivers, and taking responsibility for their drivers...

Anger won't solve the problem. But hey, I am stuck with this laptop, and I know I have no choice (at the moment). If this is a desktop, I would have throw away this ATI card long time ago. Perhaps from ATI perspective, maybe they think we should be grateful that we are actually getting some support them.

I am using this laptop for more than 1 and half year now. Everytime there is a release, I was hoping for something new, something that can give me those fancy stuff I have been seeing (Composite, AIGLX, ...), and searching the net for more info, which finally led me to this forum. After reading stuff around here, I know, I should kiss my hope good bye and just be realitic. If ATI can give me stable driver, I am happy.

However, am I content with this situation?

Well, I can only say, if, by the time, I can getting my new laptop and the situation still remains the same, then I will be kissing ATI good bye as well... :D

[gfxdrone]

The one bright spot is that the experimental ati xorg drivers are working very well on older hardware. I really think that ATI should just admit their incompetence and stop even trying, just open the specs up and let Linux hackers do it, I'm running experimental xorg ati drivers and they are great on old hardware, the stuff works. That's the hardware ATI doesn't even have a driver for anymore. I don't know what's wrong in the ATI corporation, and I don't know why Phoronix is pretending that what is wrong is just fine, but it's not to Phoronixes credit to publish this kind of blatant pandering to ATI, it's pathetic to see a website do this, it reminds me of MS fanboy sites reviewing Vista.

It's clear that volunteer hackers are in fact doing better work with fewer resources than ATI, so why doesn't ati just admit that they can't do this work, for whatever reason, and stop pretending? Then open the specs and let some real programmers start working on it.

My anger is caused by two things: one, phoronixes pandering to ATI, down to printing blatantly false statements about nVidia versus ATI Linux drivers, and two: the continuing mess in their driver process, and the pathetic support ATI is providing its Linux users. I guess I really am going to give up now on this situation ever improving.

[yoshi314]

yeah i get that impression too. every shortcoming in the driver is countered by "however" statement followed by some phony reasons why ati users shouldn't worry.

i got used to it already. it's like watching commercials on tv - you just don't notice them anymore.

[uhoh]

8.37 doesn't work at all for me with mobility x600, on debian/unstable. startx switches from textmode, screen goes blank for a short while (as it usually does), then all I get is some faint green vertical lines and everything is locking up. Only option is alt+sysrq+s,u,b to do a half-clean reboot.

Downgrading xserver-xorg-core to the pre-1.3 version, and installing 8.36, everything works as before. (still I get screen corruption on the lower right corner at times, occasional mouse pointer corruption, only opengl for the first 2048 horizontal pixels, no aiglx, 50% chance of lockup when switching VT.).

ATI drivers on linux are hell, and the press releases about getting better are plain lies. Avoid ATI cards at all costs if you plan to run Linux.

:mad:

[gfxdrone]

This phoronix press release was a SERIOUS mistake on AMD's part, and on Phoronixes. AMD had to announce something of substance, something real, and instead they just tricked phoronix into printing a stupid corporate PR piece. That is a VERY bad sign in my opinion, it suggests that the problems at ATI may go higher up than I had thought, possibly into AMD itself.

If I were phoronix, I'd think very carefully about being used in this way again, at least if Phoronix wants to maintain any credibility. And I'd definitely not print the article without serious critical commentary, like pointing out what nonsense it is in almost every respect.

[Huenengrab]

I'm not even going to try the new drivers. Its basically all the same with fglrx. When ATI/AMD releases their hopefully better OpenGL-Driver, I'll give it a shot, as long as it happens this year. As soon as UT3/Quake Wars gets released, I'm planning on building a new machine, that'll be as linux-compatible as possible. If ATI/AMD didn't improve anything until then, ATI-graphics are dead for me.

[curzondax]

curzondax, this is because the new OpenGL driver isn't integrated yet.

wait, what? where is that info? any approximate dates on it? :)

[uhoh]

I'm not even going to try the new drivers. Its basically all the same with fglrx. When ATI/AMD releases their hopefully better OpenGL-Driver, I'll give it a shot, as long as it happens this year. As soon as UT3/Quake Wars gets released, I'm planning on building a new machine, that'll be as linux-compatible as possible. If ATI/AMD didn't improve anything until then, ATI-graphics are dead for me.

Why the "if"? ATI/AMD has proven themselves to be completely untrustworthy for anything related to Linux. Save yourself the trouble and buy nvidia (or maybe intel if what they come up with is fast enough), even if ATI magically fixes their driver problems.

[partha]

The AMD/ATI fglrx 8.37.6 driver is now available... It contains X server 1.3 naming fix, AMDCCCLE v1.0 (adds translations and the 3D page), RHEL5 support, and some random fixes.

https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/linux_8.37.6.html



http://www.phoronix.com/vr.php?view=10069


FEDORA USERS, READ THIS
Okay, here's the scoop on Fedora 7. WIth AMD fixing the X 1.3 problem, the drivers will build. However, when you install the RPMs for Fedora 7, aticonfig will crash and will remove your xorg.conf (though the original is renamed). If you manually update your xorg.conf and then restart X, fglrx will fail to load and you'll find a backtrace in your Xorg.0.log. So if you need the fglrx drivers to survive, don't upgrade to Fedora 7 just yet. Stay tuned to the forums for more information. This issue with 8.37 should not cause problems with Fedora Core 6 or previous versions.
This is indeed bad news for the F7 adopters with ATI cards who who want 3D. Looks like they have to settle for the open source driver.

Michael, I posted your statement above to the FedoraForum.

Partha
www.partha.com

[fanATic]

You're completely right gfxdrone.ATI or AMD had enough time,don't care for merging.It's their tradition,lousy linux support.They already had their punch with detonators and this wake them up from dreams for making catalyst drivers.First of all, I don't even like name of drivers "frgl" (it reminds me of fireGL and Quadro line).Second, on their driver page, on linux driver's, Radeon X1950 (pro)is not listed, unlike on Windows tm's driver page.Again no AIGLX and,yes Michael no Xorg 7.2 support
50.9MB 8.37.6 May 31, 2007 Automated installer and Display Drivers for XFree86 4.3 and X.Org 6.7, 6.8, 6.9, 7.0, 7.1.I already try it.I will install FC7.Now I will sell my X1950pro and buy some nVIDIA instead (it seems now,no use of my 975XBX2 crossfire support :D),maybe some 7100,7200 or something turbocache(no hypermemory warp speed) crap.At least I will have AIGLX and 3D support.I can run beryl on my noFX5200(Dawn)-64 MB card.So what is use of my powerful RV570? Only on XP,Vista? No thanks.Maybe improvement around folding@home, nice job ATi or maybe AMD, M$ is proud of you.
And by the way nVIDIA drivers are some 13 MB's comparing to whoping 50 ATI's.So what's the problem CLI kind of driving.No problem for me.

[bimbo]

I'm seeing much higher temps at times with 8.37 than with Feisty's stock 8.34 on my T60. Is anyone else seeing symptoms like that?

Temperature occasionally jumps up 15 degrees for no apparent reason which makes the fans go crazy, then slowly goes down.

See also http://www.phoronix.net/forums/showthread.php?t=2890 which I eventually solved by going back to stock Ubuntu drivers...

The following is the output of the thinkpad fan control script. The fourth integer in the bracketed list gives GPU temperature, which jumps up at times for no reason (see the 3rd line, fan didnt run at all before, then it jumps up 11 degrees which make it run full speed):

L=0->0 EC=00 RPM= 0 T=(48 34 35 61 30 _ 28 _ , 35 46 45 , _ ) Z=0+00000_0_,000,_
L=0->0 EC=00 RPM= 0 T=(48 34 34 61 30 _ 28 _ , 35 46 45 , _ ) Z=0+00000_0_,000,_
L=0->7 EC=00 RPM= 0 T=(48 34 35 70 30 _ 28 _ , 36 46 45 , _ ) Z=0+00060_0_,000,_
> Changing fan level: 0->7 (temps: 48 34 35 70 30 _ 28 _ , 36 46 45 , _ )
L=7->7 EC=07 RPM=2561 T=(48 34 35 69 30 _ 28 _ , 36 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=2561 T=(48 34 35 71 30 _ 28 _ , 36 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=2561 T=(48 34 35 74 30 _ 28 _ , 36 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=4287 T=(48 34 35 72 30 _ 28 _ , 35 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=3916 T=(48 34 35 72 30 _ 28 _ , 35 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=3916 T=(48 34 35 71 30 _ 28 _ , 35 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=3967 T=(47 34 35 71 30 _ 28 _ , 35 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=3967 T=(47 34 35 70 30 _ 28 _ , 35 46 45 , _ ) Z=6+00060_0_,000,_
L=7->7 EC=07 RPM=3967 T=(47 34 35 71 30 _ 28 _ , 35 46 45 , _ ) Z=6+00060_0_,000,_
L=7->4 EC=07 RPM=3980 T=(47 33 35 64 30 _ 28 _ , 35 46 45 , _ ) Z=4+00020_0_,000,_

[bimbo]

So since I can't seem to fix that temperature jump issue, has anyone got a patch for 8.35 to get the kernel module build on 2.6.21.3 (no it's not the paravirt ops issue)?

[Michael]

So since I can't seem to fix that temperature jump issue, has anyone got a patch for 8.35 to get the kernel module build on 2.6.21.3 (no it's not the paravirt ops issue)?

You should be able to replace the 8.35 firegl_public code with the 8.37 firegl_public code and it should build against 2.6.21 or 2.6.22.

[the-me]

Just for the notes, things have improved for me a bit on a Tinkpad T60 (X1400, Gentoo stable with xorg-server 1.2.0) with 8.37 . While xv simply was gone with 8.36, it has reappeared for me with 8.37. It is still broken though, I get colored stripes, the xv window isn't properly hidden by other windows, and when maximized, I get an odd "hanging" effect in playback with the video stopping every second and jumping back a frame (looks like double buffering issue). On the other hand, vsync with opengl now works for me, which it never did before making opengl finally an emergency replacement for xv (which it wasn't before since I find tearing in playback very annoying).

Still, I switched back to 8.35 which is the best-working version for me so far.
First I thought XVideo is running now fine.
Now I saw, I've got the same issue. Mplayer has this one, xine is crashing with XV, so they didn't fix it!!!

[bimbo]

Currently, I'm mulling over importing a T61 with GMA X3100 from the US (I live in Switzerland, Lenovo doesnt sell those here) and ebay my 1 month old T60...

[hobbes]

I have a new installation of Ubuntu feisty and this message on var/log/Xorg.0.log

(II) fglrx(0): driver needs X.org 7.1.x.y with x.y >= 0.0
(II) fglrx(0): detected X.org 7.1.0.0
(EE) fglrx(0): atiddxDriScreenInit failed, GPS not been initialized.
(WW) fglrx(0): ***********************************************
(WW) fglrx(0): * DRI initialization failed! *
(WW) fglrx(0): * (maybe driver kernel module missing or bad) *
(WW) fglrx(0): * 2D acceleraton available (MMIO) *
(WW) fglrx(0): * no 3D acceleration available *
(WW) fglrx(0): ********************************************* *

i googled for this and found nothing useful.

fglrx does not seems to load, even if manually forced.

i'm really stuck right there.

I followed, as always did, the instruction on this wiki.

http://wiki.cchtml.com/index.php/Ubuntu_Feisty_Installation_Guide#Method_2:_Install _the_8.37.6_Driver_Manually

With my previous feisty installation and 8.36.5 driver at least was working.

Anyone?

[gfxdrone]

If you update to latest xserver-xorg-core in debian sid you can now install the 8.37.6 driver without using hacks.

Hopefully users will be able to enjoy at least one functioning driver now, finally, after all this time of nothing.

ATI continues to be the driver of choice for all serious masochists and fans of low quality software and development processes, and seems firmly committed to maintaining that position.

[Svartalf]

curzondax, this is because the new OpenGL driver isn't integrated yet.

Heh... How do you come by that info- not that it's at all wrong, mind...

[Nicolas]

Well. That was another useless driver release for me (and for most people).

But hey, AMD/ATI development cycle shows that a driver version takes 11 weeks, due to testing and bla bla bla.

I guess they have about 4 developers there:
The first develops the "to-be-realeased" driver.
The second developes the "next" version.
The third pretends to test the drivers.
And finally the fourth is on it's own on the "incredible new OpenGL driver".

Jokes apart, I'm deeply disappointed, but I kind of expected this.

I guess we just got confirmation that we will continue to receive monthly useless driver updates.

By the time they implement AIGLX support, I really hope I have already updated my notebook, and I definitelly won't buy ATI anymore. I don't have any hope anymore, and I use Linux about 10 hours per day.

I'm a developer and I really fail to understand why does it take SO LONG to solve the common problems that people have been pointing out for MONTHS already (if not years?). AIGLX can't be that hard to take months for them to start doing something. nVidia solved this problem quickly. My friend's running integrated Intel cards use AIGLX with no problems AND DECENT performance. I bet even with the ultra-slow performance of the AMD/ATI Linux OpenGL implementation, my X700 Mobility must have better performance then an Intel Integrated Video Card.

I don't care about CCC and anisotropic/antialiasing options. Performance is bad as it is. Anti-aliasing/Aniso would just make things worse.

Someone posted that AIGLX was useless. I don't think so. User Interface is very important.

I'll check back when 8.53.6 is released with initial buggy betta AIGLX support...

[yoshi314]

ati really needs to get it together. i find it hard to believe in what is written in development cycle article.

maybe the monthly release cycle is too frequent for linux team to actually implement *something* within the drivers.

[DarkFoss]

Personally I find these drivers to be better than any since the 8.28.8's.. glxgears went from 125 to over 2k
Flatout under Cedega is smoother fps dipped to 44 but hovered around 70+ for the most part, although there is still plenty of pop-ups and missing textures. PPracer was a steady 125 fps aa enabled.. I haven't tried vid playback yet..

For me these are keepers

[Xipeos]

You're completely right gfxdrone.ATI or AMD had enough time,don't care for merging.It's their tradition,lousy linux support.They already had their punch with detonators and this wake them up from dreams for making catalyst drivers.First of all, I don't even like name of drivers "frgl" (it reminds me of fireGL and Quadro line).Second, on their driver page, on linux driver's, Radeon X1950 (pro)is not listed, unlike on Windows tm's driver page.Again no AIGLX and,yes Michael no Xorg 7.2 support
.I already try it.I will install FC7.Now I will sell my X1950pro and buy some nVIDIA instead (it seems now,no use of my 975XBX2 crossfire support :D),maybe some 7100,7200 or something turbocache(no hypermemory warp speed) crap.At least I will have AIGLX and 3D support.I can run beryl on my noFX5200(Dawn)-64 MB card.So what is use of my powerful RV570? Only on XP,Vista? No thanks.Maybe improvement around folding@home, nice job ATi or maybe AMD, M$ is proud of you.
And by the way nVIDIA drivers are some 13 MB's comparing to whoping 50 ATI's.So what's the problem CLI kind of driving.No problem for me.

The fglrx driver DOES support X.org 7.2 and the installer should work without problems. For pre 8.37.6 drivers, you just had to "export X_VERSION=x710" before running the installer (doesn't anyone read those output messages?!), but the new one detects "X.org 7.1 or later"

This, of course, only if you're not too lazy or incapable of departing from your distro setup tools.

[lenrek]

...
This, of course, only if you're not too lazy or incapable of departing from your distro setup tools.

He is talking about FC7, and it is a known problem, at least for now.

http://www.phoronix.net/forums/showthread.php?t=3085

[Xipeos]

He is talking about FC7, and it is a known problem, at least for now.

http://www.phoronix.net/forums/showthread.php?t=3085

They are not X.Org related and I will explain more in the very near future

Then perhaps it's a kernel problem? If that's the case then you can probably use the vanilla kernel (unless it breaks some drivers/features specific to fedora).

[Michael]

Then perhaps it's a kernel problem? If that's the case then you can probably use the vanilla kernel (unless it breaks some drivers/features specific to fedora).

Not a kernel problem.

[Alistair]

Michael: Direct question for the ATI folks if you could

Given a working and functional setup that gets my ATI X1650Pro AGP 512Mb card up and running with opengl and fglrx drivers in xorg,
WHY is it that although glxgears and fgl_glxgears indicate the x1660pro card is producing 1.5 to 1.75 times the framerate of my 9600 agp card in default size, when I am playing World Of Warcraft in full screen on 1440x900 screen the x1650Pro agp card runs at about 1/2 to 1/4 the framerate of my ATI 9600 card?

I will note that the framerates in wow **when in small enclosed areas** or when ** looking straight up at the sky ** on the x1650Pro are 1.3 to 1.5 times faster. I am using the configuration (bios agp gart=512, video vram=256 in boot command line and xorg.conf maxgartsize=256) that works to get opengl working on the 1650pro to run the the 9600 --

I've got AGP 8x mode enabled on the x1650pro but somehow I get the idea that areas with lots of textures or texture work to be done are the cause of the dropoff in framerate. I am currently running the 8.37.6 drivers - and in fact have reasonable confidence in them as they have subtly improved the quality of my 2D rendered objects (my desktop fonts are clearly smoother and sharper than on 8.35, although fonts in things like Konsole and kwrite dont appear to have benefitted as much as fonts on webpages and on kicker components)

not that I'm whining (much) but I'm suspecting that this is a code path issue and likely fairly easy to resolve.

[Michael]

Michael: Direct question for the ATI folks if you could

Given a working and functional setup that gets my ATI X1650Pro AGP 512Mb card up and running with opengl and fglrx drivers in xorg,
WHY is it that although glxgears and fgl_glxgears indicate the x1660pro card is producing 1.5 to 1.75 times the framerate of my 9600 agp card in default size, when I am playing World Of Warcraft in full screen on 1440x900 screen the x1650Pro agp card runs at about 1/2 to 1/4 the framerate of my ATI 9600 card?

I will note that the framerates in wow **when in small enclosed areas** or when ** looking straight up at the sky ** on the x1650Pro are 1.3 to 1.5 times faster. I am using the configuration (bios agp gart=512, video vram=256 in boot command line and xorg.conf maxgartsize=256) that works to get opengl working on the 1650pro to run the the 9600 --

I've got AGP 8x mode enabled on the x1650pro but somehow I get the idea that areas with lots of textures or texture work to be done are the cause of the dropoff in framerate. I am currently running the 8.37.6 drivers - and in fact have reasonable confidence in them as they have subtly improved the quality of my 2D rendered objects (my desktop fonts are clearly smoother and sharper than on 8.35, although fonts in things like Konsole and kwrite dont appear to have benefitted as much as fonts on webpages and on kicker components)

not that I'm whining (much) but I'm suspecting that this is a code path issue and likely fairly easy to resolve.

Have you tried any Linux native games other than using WINE/Cedega to compare the performance? Cedega and the fglrx drivers haven't been best mates. In a game like Quake 4 or Enemy Territory you should notice improved performance.