AMD Releases Catalyst 9.6 For Linux

[phoronix]

Phoronix: AMD Releases Catalyst 9.6 For Linux

AMD has pushed out a new Catalyst driver update for Linux users this afternoon. The two new features in this release, as mentioned by their release notes (PDF), is support for new Linux distributions and MultiView support for Radeon hardware. The distributions now officially supported by the Catalyst 9.6 Linux driver is SuSE Linux Enterprise Desktop 11 and early Red Hat Enterprise Linux 4.8 support...

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

[hdas]

so soon? it is just 15th today.

[super.rad]

Does it support 2.6.30 kernel?

[FallenWizard]

The website still shows 9.5, and does the new driver have 2.6.29/30 support?

[Mazur]

No because who needs it? ATI is still in time where 2.6.10 is out.

[FallenWizard]

No because who needs it? ATI is still in time where 2.6.10 is out.


I know, I am using the Open Source drivers anyway, but I am just curious.

[JeanPaul145]

so soon? it is just 15th today.

It seems AMD is aware of that. Catalyst 9.6 isn't even available for download yet.

[bojo42]

oh no, the worst issue ever:
"System may stop responding when running Return to Castle Wolfenstein: Enemy Territory"

so even if i can have 3D on a 780g with the driver, i might not be able to use it for the only true reason ;)

[teuf31]

Enough is enough. Tomorrow, I buy a NVIDIA card for my mediacenter. I'm tired of waiting for an XvBA interface that will never come out. Never more ATI :mad:

[SyXbiT]

2.6.30 support..... WHY?
that would be supporting a new kernel WAY head of schedule.
why not wait until people get angry and go and buy nvidia.
then we'll really have a reason to improve

[kensai]

Another useless driver,

- No support for kernel 2.6.29 or 2.6.30.
- Legacy GPUs code is still there though disabled making the download too big and not letting legacy users enjoy this code while it is still there, what a selfish bunch.
-No support for Linux yet, just three outdated distributions.

Guys, I can go as far as to say that kernel 2.6.30 or .31 for that matter, support will be added by the 9.9 if we are lucky or 9.10 release.

Yesterday I was still an AMD supporter even though I hated ATI. Today, I hate AMD and ATI. Is so sad, I won't buy AMD and even less ATI anymore.

[bojo42]

@JeanPaul145: you can grab it here: https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/ati-driver-installer-9-6-x86.x86_64.run

[Mazur]

I just would like to know:

"Why?!"

You got one month to prepare good working driver with support, many people was even giving code to support new kernel's but no. This driver should be called 'Catalyst for stable/old Ubuntu', not Catalyst for Linux. You do not support stable kernels, you do not support alpha kernels. Then what the hell you support?

[mendieta]

It seems AMD is aware of that. Catalyst 9.6 isn't even available for download yet.

Just replace 9-5 by 9-6 in the URL for the download ...

[JeanPaul145]

Just replace 9-5 by 9-6 in the URL for the download ...

Yeah between that post and this one they changed it. I'm installing it right now, wondering in what wonderful ways the June '09 edition of fglrx will disappoint me.

EDIT: I see the whole xorg-server-patch-removal-causing-terrible-delays-on-unminimizing-windows thing hasn't been cleared up yet. Back to Metacity for me.

I must say, I'm getting mighty tired of always wondering what functionality will get broken with my AMD card, and then not getting timely fixes for that breakage once it happens.if this keeps up my next gfx card (admittedly more than a year or two away) will come from the stables of Nvidia.

@the AMD execs: I get the whole professional-businesses-make-us-money thing, but don't forget about the Linux desktop people please. Unreleased UVD2 code, mediocre performance on high-end gfx cards, updates barely even worth updating to (and getting less so every month)... This is the reward we get for buying an AMD gfx card. If you have a shortage of manpower, just hire more people. Don't give us the economic crisis excuse (developers don't cost *THAT* much, just fire a couple of execs with poor job performance reviews or something...), just listen to your customers. All of them.
P.S.: I'm fully aware of the FOSS drivers, but those don't really currently help those of us with an R600/R700 card now do they ;)

[mendieta]

That's nice. We've been all asking for Open Source graphics drivers and open specs for years. Now that AMD/ATI do it, we flame them. How cool are we?

Granted, NVIDIA's binary only support is better as such, so I would probably by an NVIDIA card in the short term if that;s the only thing going (and I've been _this_ close to doing it a lil ago out of frustration with ati + ubuntu 9.04).

But in the long run ATI and Intel are the way to go. And we should be respectful of them for doing the right thing.

[he_the_great]

That's nice. We've been all asking for Open Source graphics drivers and open specs for years. Now that AMD/ATI do it, we flame them. How cool are we?

Granted, NVIDIA's binary only support is better as such, so I would probably by an NVIDIA card in the short term if that;s the only thing going (and I've been _this_ close to doing it a lil ago out of frustration with ati + ubuntu 9.04).

But in the long run ATI and Intel are the way to go. And we should be respectful of them for doing the right thing.

But providing open specs and drivers didn't fix any problems so of course people are still mad. Things are supposed to be instantly better when you do that.

[kensai]

That's nice. We've been all asking for Open Source graphics drivers and open specs for years. Now that AMD/ATI do it, we flame them. How cool are we?

[...]

But in the long run ATI and Intel are the way to go. And we should be respectful of them for doing the right thing.

Their open source driver is a trash for rx600 cards and above, and their closed source drivers is a trash. So, am I going to be happy with buying something that doesn't work as it should because the company that makes it doesn;t even know how to make a driver for it.

At least NVIDIAs closed source driver is so good that you don;t even wish for an open source one.

[DanL]

If you're still using fglrx, you're doing it wrong.

[bulletxt]

Everytime I see a new AMD fglrx release I start to laugh... and I can't stop. AMD Catalyst isn't a driver, it's a joke or just a bad dream.

[mendieta]

At least NVIDIAs closed source driver is so good that you don;t even wish for an open source one.

It is better, for sure, but I have wished it was open at least once (with a horrible bug not freezing my machine on logout by using KDM. I had to switch to GDM in Kubuntu because of this. This is documented in launchpad, and it could never fixed (because it is closed source).

Remember, the grass is always greener on the other side.

[DanL]

Remember, the grass is always greener on the other side.
"The grass is always greener where the dogs are sh!tt!ing."

[MaestroMaus]

It is better, for sure, but I have wished it was open at least once ... Remember, the grass is always greener on the other side.

I second that. My laptop keeps making the screen a mess when it is on battery...

[Louise]

oh no, the worst issue ever:
"System may stop responding when running Return to Castle Wolfenstein: Enemy Territory"

so even if i can have 3D on a 780g with the driver, i might not be able to use it for the only true reason ;)

I also have a 780G. Why don't you just use the 3200HD driver?

Don't flame me. I am asking, because I don't know the answer :)

[LavosPhoenix]

What the fuck! They said that that 9.6 would have support for 2.6.29, yet they FAIL to deliver on their promises ONCE AGAIN! I don't know what the fuck AMD's graphic driver development team has been doing all these months, but maybe they need a pay cut to encourage them to actually get work done. Seriously, this is ridiculous. Once again I will state that AMD needs to be sued for not providing a working product, since the hardware is useless without properly working drivers. And thus be forced to provide full refunds to all those with damages. After all, that's EXACTLY what they've done to those with X1000 and before. At the very LEAST they need to be forced to release the full source code since it does not contain anything that cannot be learned by disassembling the driver.

[ivanovic]

What the fuck! They said that that 9.6 would have support for 2.6.29, yet they FAIL to deliver on their promises ONCE AGAIN!
Uhm, when has AMD/ATI ever promised this? There was noone officially stating that the 9.6 release will have 2.6.29 support. There were many users speculating that it would be about time, but no single time any AMD/ATI official stated that this support would be included.

[Mazur]

I am sick of those drivers and I will never buy ATI again even if 3d and KMS support will be added to open source drivers. I preffer Intel and Nvidia.

[Heiko]

What the fuck! They said that that 9.6 would have support for 2.6.29, yet they FAIL to deliver on their promises ONCE AGAIN! I don't know what the fuck AMD's graphic driver development team has been doing all these months, but maybe they need a pay cut to encourage them to actually get work done. Seriously, this is ridiculous. Once again I will state that AMD needs to be sued for not providing a working product, since the hardware is useless without properly working drivers. And thus be forced to provide full refunds to all those with damages. After all, that's EXACTLY what they've done to those with X1000 and before. At the very LEAST they need to be forced to release the full source code since it does not contain anything that cannot be learned by disassembling the driver.

Who said that and when did they say that? I never heard AMD making any promisses about future driver releases. And did you really expect that they'd support a new kernel? You guys should know better by now... the first driver to support 2.6.29 or 2.6.30 will be 9.10, because thats when Ubuntu 9.10 gets released. So it will probably be a beta driver for Ubuntu. Don't expect these things earlier...

I'm getting a bit sick of all those people whining each month. AMD did lots of work to improve the situation over the last few years and in my opinion they definately succeeded. I agree the driver is not that good as nvidia's one, but it certainly does its work in most cases. Besides that, they opened up every possible document to write open source drivers. You guys really should applaud AMD for that, really... this is a huge step and most people posting here don't seem to appreciate that.

[Melcar]

Every month the little dogs start yapping. It's hilarious really.

[Louise]

Who said that and when did they say that? I never heard AMD making any promisses about future driver releases. And did you really expect that they'd support a new kernel? You guys should know better by now... the first driver to support 2.6.29 or 2.6.30 will be 9.10, because thats when Ubuntu 9.10 gets released. So it will probably be a beta driver for Ubuntu. Don't expect these things earlier...

I'm getting a bit sick of all those people whining each month. AMD did lots of work to improve the situation over the last few years and in my opinion they definately succeeded. I agree the driver is not that good as nvidia's one, but it certainly does its work in most cases. Besides that, they opened up every possible document to write open source drivers. You guys really should applaud AMD for that, really... this is a huge step and most people posting here don't seem to appreciate that.

Yes I am with you 100%.

In fact AMD is opening their drivers, which nVidia doesn't.

We are in a very existing time right now, where AMD is changing how graphics is done on Linux. Forever.

To those that complain: Don't you think you are being just a little bit too egoistic?

No one have ever done what AMD is doing to Linux right now. Just imagine how much money and risk AMD is putting into Linux, to give us what we really want. Open Source drivers, and in the long run NEVER have to download or think about GFX drivers again. It just works out-of-the-box.

[super.rad]

You guys should know better by now... the first driver to support 2.6.29 or 2.6.30 will be 9.10, because thats when Ubuntu 9.10 gets released. So it will probably be a beta driver for Ubuntu. Don't expect these things earlier...

Won't be much use then as Ubuntu 9.10 will use the 2.6.31 kernel

[Louise]

Every month the little dogs start yapping. It's hilarious really.

Yes :)

I think Michael is writing the articles for this very enjoyment :D

[energyman]

Who said that and when did they say that? I never heard AMD making any promisses about future driver releases. And did you really expect that they'd support a new kernel? You guys should know better by now... the first driver to support 2.6.29 or 2.6.30 will be 9.10, because thats when Ubuntu 9.10 gets released. So it will probably be a beta driver for Ubuntu. Don't expect these things earlier...

I'm getting a bit sick of all those people whining each month. AMD did lots of work to improve the situation over the last few years and in my opinion they definately succeeded. I agree the driver is not that good as nvidia's one, but it certainly does its work in most cases. Besides that, they opened up every possible document to write open source drivers. You guys really should applaud AMD for that, really... this is a huge step and most people posting here don't seem to appreciate that.

well, I switched from Nvidia to AMD/ATI last year - because of their new open source driver politics. That is a lot applauding, in my opinion. I put my money on them.

I am in for new card in ca 3 month. Maybe 5. I am not sure if I buy the new cpu or new card first.

I will see where I put my money. AMD still has the advantage of the opensource initiative. But not supporting a kernel that is 3 month old is weak.Very weak. And before you answer, think about this:
with 2.6.30 out 2.6.28 won't get any fixes anymore. So you can either use the crappy 2.6.27 or 2.6.29, patch the driver and have dmesg spammed.

Btw, this:
[ 8540.955717] [fglrx:firegl_find_any_map] *ERROR* Invalid map handle!<3>[fglrx:drm_vm_close] *ERROR* map not found -> inconsistent kernel data!!! vma_start:0x7fedd842c000,handle:0xfdff0000

is that a real problem?

and 2.6.30 with all its improvements is completly out of question. Well, I did not expect .30 support. But support for .29 was more than due.

The opensource drivers are out of question for me. Thanks to no real power managment the fan is spinning at high speed all the time. Which means my usually quiet box becomes VERY LOUD.

Oh, and it uses a lot more power. Yeah, thanks.

[Kano]

The diff between 9-5 to 9-6 is so small that i did not even have to update pci-ids, because it does not support more cards. One of the most useless updates as it did not fix any issue for me.

[kensai]

I own nvidia and ati, so yes the grass is always greener on the other side, on the nvidia side, that is the best side to be.

As to the one that said no company has done as much as amd/ati for Linux, I just have to laugh. Intel has done way more things, way way way more. USB 3.0 is the most significant right now, they are constantly coding for the kernel and without them the kernel wouldn't have a whole lot of features even AMD users enjoy. AMD just gives some documentation that keeps their open source drivers a trash.

If it weren't for like <1% of the people applauding AMDs bad effort at doing something good, they would have realized they needed to change and do things right, so people can enjoy their hardware.

[ethana2]

Will my friend's twin 4850 crossfire rig work with the next Ubuntu?

[he_the_great]

:) not only is everyone complaining that AMD doesn't support the latest kernel, but not even Ubuntu is supporting it. 9.10 is when they will support it even though they release alphas before that. Of course Debian is still on 2.6.26 and don't know what Fedora 11 is on.

[E.Dulen]

and don't know what Fedora 11 is on.
2.6.29 :-/

[energyman]

I own nvidia and ati, so yes the grass is always greener on the other side, on the nvidia side, that is the best side to be.

As to the one that said no company has done as much as amd/ati for Linux, I just have to laugh. Intel has done way more things, way way way more. USB 3.0 is the most significant right now, they are constantly coding for the kernel and without them the kernel wouldn't have a whole lot of features even AMD users enjoy. AMD just gives some documentation that keeps their open source drivers a trash.

If it weren't for like <1% of the people applauding AMDs bad effort at doing something good, they would have realized they needed to change and do things right, so people can enjoy their hardware.

and intel shafted us more than once. So pleeeaase. AMD even worked together with linux kernel devs when creating AMD64. Intel didn't ask anyone for input when they crapped out Itanic.

[super.rad]

:) not only is everyone complaining that AMD doesn't support the latest kernel, but not even Ubuntu is supporting it. 9.10 is when they will support it even though they release alphas before that. Of course Debian is still on 2.6.26 and don't know what Fedora 11 is on.

Ubuntu karmic is currently using 2.6.30 although it will soon be moving to 2.6.31-rc1

[Louise]

As to the one that said no company has done as much as amd/ati for Linux, I just have to laugh. Intel has done way more things, way way way more. USB 3.0 is the most significant right now, they are constantly coding for the kernel and without them the kernel wouldn't have a whole lot of features even AMD users enjoy. AMD just gives some documentation that keeps their open source drivers a trash.

You are missing the point. Opening up the driver stack is showing your competitor your hand without seeing his.

Don't forget that ATi and nVidia are the two leading GPU vendors, so it is a very big deal.

[kensai]

You are missing the point. Opening up the driver stack is showing your competitor your hand without seeing his.

Don't forget that ATi and nVidia are the two leading GPU vendors, so it is a very big deal.

Not a big deal, nvidia owns this market, and until ati decides to not treat linux as a joke they won;t be respected.

[Louise]

Not a big deal, nvidia owns this market, and until ati decides to not treat linux as a joke they won;t be respected.

Again, you are missing the point.

The hardware is the same for Linux and Windows, which means that the secrets that is exposed in the open source drivers, can nVidia dig through.

And if they find hardware design flaws, they will set up benchmarks where these bugs are tricked.

Not to mention, if ATi have done something very clever, nVidia can just take it, and deal with the law suite later, if ATi ever finds out.

For a very long time that what Intel's strategy. To steal from AMD, and even if AMD won the law suite, the money Intel had made from the stolen technology was greater then the fee.

[krazy]

Their open source driver is a trash for rx600 cards and above,
My r700 card works great out-of-the-box for everything I need (EXA, Xv etc.) except 3d.
At least NVIDIAs closed source driver is so good that you don;t even wish for an open source one.;)

[cruiseoveride]

Buying an ATi card was the worst decision I ever made in my life. It really goes to show that paying a little more and getting an Nvidia is definitely worth it.

[Pfanne]

Buying an ATi card was the worst decision I ever made in my life. It really goes to show that paying a little more and getting an Nvidia is definitely worth it.

sorry if i need to correct you, but writing this post was the worst decision in your life...
go fail somewhere else please.

[cruiseoveride]

You are missing the point. Opening up the driver stack is showing your competitor your hand without seeing his.


ATi drivers have absolutely no worth. THE WORLD KNOWS ATI SUCKS. You cannot embarrass yourself any further. Ok so a few devs at Nvidia may die of laughter when they see the ATi code, but you can't do anything about that.

sorry if i need to correct you, but writing this post was the worst decision in your life...
go fail somewhere else please.
fanboy FAIL.

[linuxhansl]

Enough is enough. Tomorrow, I buy a NVIDIA card for my mediacenter. I'm tired of waiting for an XvBA interface that will never come out. Never more ATI :mad:

You go ahead and do that and be forever locked into a binary driver with a company that will never change policy.

I will instead start using the open source ATI driver when it's ready, help with debugging, and maybe even contribute some code. Until then for 3D I keep a 2.6.27.x kernel around (which is still updated with security patches, btw).

[Louise]

ATi drivers have absolutely no worth.

How old are you?


THE WORLD KNOWS ATI SUCKS. You cannot embarrass yourself any further. Ok so a few devs at Nvidia may die of laughter when they see the ATi code, but you can't do anything about that.


nVidia fanboy :)

I actually think that it is ATi's latest card that is the fastest :)

[mendieta]

My r700 card works great out-of-the-box for everything I need (EXA, Xv etc.) except 3d.
;)

I second that, my HD 3200 IGP worked wonders for everything but 3D. I can't wait for 3D Support to kick in and switch over for good.

[Spanner]

Buying an ATi card was the worst decision I ever made in my life. It really goes to show that paying a little more and getting an Nvidia is definitely worth it.

That's pretty extreme - but I'd say that buying a 4870 was the worst mistake of my PC-building 'career'. It wasn't based on cost, it was to support AMD's "open" approach.

To be sure I expected *some* pain with fglrx, but I had no idea just how much, or how consistent it would be. Now I just can't wait to throw cash at a GeForce card, to hell with open documentation!

[Jimmy]

Linux 2.6.30 seems to work just fine with Nvidia's drivers. Even on my really old SLI GeForce 6600s.

Don't know what to say other than I feel sorry for the ATI folk.

[BhaKi]

People, please don't underestimate the value of open specifications. It's true that nVidia's closed-source Linux driver is better than AMD's Linux drivers at the moment. But don't just assume that Linux-support is everything. "You must use Windows or Linux in order to use our card" is almost as bad as the "You must use Windows in order to use our card"-attitude that we've been fighting against for years. For instance, what if you decide tomorrow to switch to some obscure OS like OpenBSD? With nVidia's cards you'll only be able to use resolutions like 640x480 and that too without any acceleration or compositing. With AMD's open specs, independent developers can write drivers. That could be myself or the OpenBSD developers. THAT is true freedom.

[energyman]

don't forget that nvidia betrayed partners and users over the last couple of years. More and more shit is unconvered about their mass dying chips.

[Pulfer]

My r700 card works great out-of-the-box for everything I need (EXA, Xv etc.) except 3d.
;)

How did you get Xv to work? What driver do you use? What's your xorg.conf settings?

[grege]

People, please don't underestimate the value of open specifications. It's true that nVidia's closed-source Linux driver is better than AMD's Linux drivers at the moment. But don't just assume that Linux-support is everything. "You must use Windows or Linux in order to use our card" is almost as bad as the "You must use Windows in order to use our card"-attitude that we've been fighting against for years. For instance, what if you decide tomorrow to switch to some obscure OS like OpenBSD? With nVidia's cards you'll only be able to use resolutions like 640x480 and that too without any acceleration or compositing. With AMD's open specs, independent developers can write drivers. That could be myself or the OpenBSD developers. THAT is true freedom.

Except Nvidia make BSD drivers as well, at least for FreeBSD the last time I tried it.

However I take your point. I have a Media Box based on a Gigabyte MA78GM-S2H and use the open source driver with Ubuntu Jaunty. This box has no need of 3d or compiz effects. The 2d desktop is fast and stable, and the xv video is smooth. The box runs a Samsung LCD TV through the VGA connector. It plays and records Digital TV and plays xvid videos. The TV goes to sleep and wakes when it should. For this purpose the open source driver is great and updates occur through the normal update process.

The next step of adding proper 3d support for all cards will free us from fglrx for ever. When that happy day arrives we will all be better off because everything will "just work" - any distro, any kernel.

Personally I would rather AMD stopped development of fglrx for the end user altogether (keep it for the corporate FireGL) and put resources into finishing the open source stack.

My main computer is using Nvidia and the ATI HD3450 card I bought a year ago in high hopes of open source goodness gathers dust. By the time I can use it, it will be so out of date it will be replaced. Lucky it did not cost much.

I applaud the effort being made to create the open source driver. When finished it will be a huge boost for GNU/Linux. Sometimes it is hard to be patient.

[krazy]

How did you get Xv to work? What driver do you use? What's your xorg.conf settings?
Kubuntu Jaunty (using 2.6.30 now, but it also works on the default 2.6.28 kernel)
my xorg.conf is empty. The kernel uses radeon by default.

[grege]

How did you get Xv to work? What driver do you use? What's your xorg.conf settings?

This works for both my HD3000 series cards.

You need a 6.12 driver or newer

Section "Device"
Driver "radeon" # can also put "ati"
Option "DRI" "true"
Option "AccelMethod" "EXA"
EndSection

Ubuntu Jaunty has a 6.12.1 in the repos or you can add xorg edgers from PPA to get newer drivers.

This will give 2d only - there is no 3d support for newer cards yet.

[krazy]

Personally I would rather AMD stopped development of fglrx for the end user altogether (keep it for the corporate FireGL) and put resources into finishing the open source stack.
Hah! that's what they did and many short-sighted people bitched (http://www.phoronix.com/forums/showthread.php?t=15796) about it (http://www.phoronix.com/forums/showthread.php?t=15798). :)

I applaud the effort being made to create the open source driver. When finished it will be a huge boost for GNU/Linux. Sometimes it is hard to be patient.
agreed.

[mendieta]

I upgraded to 9.6, working just dandy.

It did take me hours and hours (over several weeks) to get 9.5 working in Ubuntu 9.04. I found the answer here:

http://forum.compiz-fusion.org/showthread.php?t=6794

Specifically:
http://forum.compiz-fusion.org/showthread.php?t=6794&page=79

sudo rm /etc/ati/amdpcsdb
sudo aticonfig --initlal -f --overlay-type=Xv
sudo aticonfig --set-pcs-str="DDX,EnableRandr12,FALSE"
ssudo aticonfig --dtop=horizontal


Sadly, I reported this in the ATI unofficial bug-tracker with all sorts of details and no one ever answered.

[mendieta]

This will give 2d only - there is no 3d support for newer cards yet.

True, we should all keep this under the pillow:
http://www.x.org/wiki/RadeonFeature

:D

[Wintervenom]

See posts 319 (http://bbs.archlinux.org/viewtopic.php?pid=569846#p569846) and 321 (http://bbs.archlinux.org/viewtopic.php?pid=569893#p569893) at The AMD/ATI Catalyst Bar & Grill (http://bbs.archlinux.org/viewtopic.php?id=57084).

[Laughing1]

Does this fix still work for with kernel-2.6.30 final?

http://www.rage3d.com/board/showpost.php?p=1335866196&postcount=5

This is the same patch, correct? http://starowa.one.pl/~uzi/xorg-driver-video-fglrx-2.6.29.patch because the one in the above link doesn't work.

[Wintervenom]

Over here, we're using: http://aur.archlinux.org/packages/catalyst/catalyst/2.6.29.diff

---

Also:
It seems this patch will make fglrx work with kernel 2.6.30, but it first needs a one-line patch to the kernel.
http://forum.ubuntu.org.cn/viewtopic.php?f=42&p=1329973
...and a translator.

[Darkfire Fox]

The trolls are out in force, I see. Does this happen EVERY month? :confused:

[Nille_kungen]

I like all my ati cards and they work good.
I will keep buying an ATI cards.
If you run an development version of an dist then things will not work as they should, so if you chose to run unstable then it will be unstable on non working.

And btw the only card i have thats not working is an nvidia quadro nvs 135, that doesn't work with kde4 and that bugs me alot.

[eL_vErDe]

fglrx...

Did I told you my own history ? I was used to have ATi cards, deal with crappy driver, running an archived kernel...
One day I was really fed up of apll that shit and I bought two nVidia cards for my computers.
The only computers that still haves ATi are my two laptops running radeon/radeonhd.

My life has been really improved :D

If you don't care about, 3D use the free drivers. If you do, sell this shit and buy an nVidia card.

Don't rely on ATi, they are too stupid to release a working driver (I still can't believe we're still stuck on the same issues than 2 or 3 three years AGO !)

[Wintervenom]

The trolls are out in force, I see. Does this happen EVERY month? :confused:

Every... single... month.

[RealNC]

Another month, another useless AMD update. No support for a 3 months old kernel.... Why am I not surprised?

[susikala]

Another Catalyst update, another hoard of users bitching about AMD.

It gets tiring, really. Yes, everyone agrees that Nvidia's binary blob is better, but people should consider the following (and I believe this has been stressed here on the forums a couple of times already):


Catalyst is _mostly_ developed for the enterprise class of AMD users, which is, I believe, still AMD's main (and more or less only) source of income on Linux. Whether one or even 10,000 users say they'd "pay with their wallets" doesn't matter the least bit, since the Linux market is still very small and I don't think AMD or Nvidia cares a rat's ass, really, whether Linux users buy their cards or not.
Nonetheless, although they have no monetary incentive (so far as I can tell) to open up their specs, AMD did. Now whatever you do, AMD Linux users have been withdrawn the right to complain... or at least complain justfully. Which makes you, if you do complain, assholes. If you have so much time to do that, go ahead and improve the open source drivers.
Nvidia doesn't even support Linux, as far as I'm concerned, since binary blobs can't be even considered real support in a FOSS environment. Can you debug/improve/learn them? No. They're anti-Linux and shouldn't be really used. They wouldn't work on a Unix with some more restrictive approach to binary blobs, like OpenBSD.

[Heiko]

Last year I bought an HD4870 for my new machine. For another machine that functions mainly as a server and plays movies on my tv as well I bought a HD3450. Both machines are running fine using Ubuntu 8.04, and at the moment I'm switching to Ubuntu 9.04 and so far it works fine. I don't have any reason to upgrade to a newer kernel, so I'm quite happy the way AMD drivers turn out. I even program my own OpenGL programs (OpenGL 3.0 that is) and they run fine on my AMD hardware.

I'm still hoping for more improvements of course, but I think my expectations are somewhat more realistic than those of other people around here. Of course I see that if you're a Fedora user things are a bit nasty. I think if AMD starts supporting Fedora officially, the monthly complaints will drop quite a bit.

But be honest... how many Linux users, are upgrading to a newer kernel than the one that is in their distro? I think this is a minority, so supporting Fedora (which does use a newer kernel) officially would help decrease the complaints a lot.

[RealNC]

Another Catalyst update, another hoard of users bitching about AMD.

It gets tiring, really. Yes, everyone agrees that Nvidia's binary blob is better, but people should consider the following (and I believe this has been stressed here on the forums a couple of times already):

Catalyst is _mostly_ developed for the enterprise class of AMD users, which is, I believe, still AMD's main (and more or less only) source of income on Linux. Whether one or even 10,000 users say they'd "pay with their wallets" doesn't matter the least bit, since the Linux market is still very small and I don't think AMD or Nvidia cares a rat's ass, really, whether Linux users buy their cards or not.

Bullsh!t. Lots of Linux users have ATI cards. And we PAID for them. We didn't get them for free. And even if we did, someone else paid for them. AMD doesn't give out hardware for free.


Nonetheless, although they have no monetary incentive (so far as I can tell) to open up their specs, AMD did. Now whatever you do, AMD Linux users have been withdrawn the right to complain... or at least complain justfully. Which makes you, if you do complain, assholes. If you have so much time to do that, go ahead and improve the open source drivers.

Bullsh!t. We have every right to complain since the drivers don't work correctly. We are customers. If AMD starts putting "Not suitable for Linux" stickers on their cards, THEN we won't have a right to complain.

Nvidia doesn't even support Linux, as far as I'm concerned, since binary blobs can't be even considered real support in a FOSS environment.

Bullsh!t. Not everyone is a Freetard. Not all of us care about "FOSS enviroments". Not even the kernel developers care about "FOSS environments". As Linus Torvalds put it: "Go suck on RMS." We care about working products and NVidia delivers those. They SUPPORT Linux because they want to stay market leaders. Their main target on Linux are workstations too, but they still deliver very good drivers suitable for multimedia and gaming desktops.

[Lukyn]

Hey guys, just relax. Whats going on? U are free to choose. Just sell your ATI and buy the nVidia. I know it now. I was so stupid that I buy ATI again with hope that ATI drivers are going better and better. Well, I was wrong. I didn't know that elite of Creative programmers now works for AMD/ATI.... I have feeling that in last changelog is missing, that CCC panel have now nicer colors...

And ppl with a little bit older ATI card, whitch is unsupported at all, can...welll...then can't do anything. They are just a garbage probably or I don't know. Well nVidia still maintain old Riva cards in drivers.... It is sad ATI. However, I can choose, so I see that future can be better, but not with ATI :)

[teuf31]

You go ahead and do that and be forever locked into a binary driver with a company that will never change policy.

I will instead start using the open source ATI driver when it's ready, help with debugging, and maybe even contribute some code. Until then for 3D I keep a 2.6.27.x kernel around (which is still updated with security patches, btw).

I use the open source ATI drivers since the begining... very good and stable drivers but it still missing the full 3D features and mainly from my side the UVD2 support.

Last year I'd buy a 3200 IGP card because I thought AMD was the best way regarding the openning of the documentation and the best full hardware decoding for HD videos. But one year later, I cannot use my hardware as expected.

I really need UVD2! open or closed source, at this step, I really don't care because my excellent Gigabyte MA78GM is near to finish into the trash.:(

... Yes I know, allways the same answer : "Stop crying and write yourself your own driver!". But what about the UVD2 documentation ?

About closed source driver, I can understand that the driver cannot deliver all the features in the same time as the windows driver
.... but one long, very long year later ??????????

[BlackStar]

I like the smell of trolls in the morning. Always the same people bitching about the same things every chance they get. Does this forum have a 'block user' function?

Back on topic: does anyone know if 9.6 supports EXT_geometry_shader? There were rumours that this was added in the beta, but I never got a chance to test.

[susikala]

x3 Bullsh!t
Why don't you go use Windows with that approach, then? You get everything you paid for and I hear it works well.

[Heiko]

I like the smell of trolls in the morning. Always the same people bitching about the same things every chance they get. Does this forum have a 'block user' function?

Back on topic: does anyone know if 9.6 supports EXT_geometry_shader? There were rumours that this was added in the beta, but I never got a chance to test.

From what I heard they only increased the maximum number of geometry shader uniforms from 0 to 512 in driver 9.6, which could be seen as a sign that geometry shader support is in the works... But no actual support at the moment.

[krazy]

I know I'm feeding the trolls, but honestly..
Bullsh!t. Not everyone is a Freetard. Not all of us care about "FOSS enviroments". Not even the kernel developers care about "FOSS environments". As Linus Torvalds put it: "Go suck on RMS."
Basically, I want people to know that when they use binary-only modules,
it's THEIR problem. I want people to know that in their bones, and I
want it shouted out from the rooftops. I want people to wake up in a
cold sweat every once in a while if they use binary-only modules.

Why? Because I'm a prick, and I want people to suffer? No.

Because I _know_ that I will eventually make changes that break modules.
And I want people to expect them, and I never EVER want to see an email
in my mailbox that says "Damn you, Linus, I used this binary module for
over two years, and it worked perfectly across 150 kernel releases, and
Linux-5.6.71 broke it, and you had better fix your kernel".

See?

I refuse to be at the mercy of any binary-only module. And that's why I
refuse to care about them - not because of any really technical reasons,
not because I'm a callous bastard, but because I refuse to tie my hands
behind my back and hear somebody say "Bend Over, Boy, Because You Have
It Coming To You".

I allow binary-only modules, but I want people to know that they are
_only_ ever expected to work on the one version of the kernel that they
were compiled for. Anything else is just a very nice unexpected bonus if
it happens to work.

And THAT, my friend, is why when somebody complains about AFS, I tell
them to go screw themselves, and not come complaining to me but complain
to the AFS buys and girls. And why I'm not very interested in changing
that.

Linus

[energyman]

I like the smell of trolls in the morning. Always the same people bitching about the same things every chance they get. Does this forum have a 'block user' function?

Back on topic: does anyone know if 9.6 supports EXT_geometry_shader? There were rumours that this was added in the beta, but I never got a chance to test.

wrong. I defended AMD in the last couple of month. But no 2.6.29 support really pisses me off. Why shouldn't I be allowed to say that? That kernel is out for thre(!) freaking month! 2.6.28 is unsupported, 2.6.27 a pile of crap.

And I have a good reason, why I want the latest kernels - I have problems with mtrr - but before I can even think about reporting on lkml I must be able to reproduce all the problems with the latest stable kernel - oh look, I can't.

[susikala]

wrong. I defended AMD in the last couple of month. But no 2.6.29 support really pisses me off. Why shouldn't I be allowed to say that? That kernel is out for thre(!) freaking month! 2.6.28 is unsupported, 2.6.27 a pile of crap.

And I have a good reason, why I want the latest kernels - I have problems with mtrr - but before I can even think about reporting on lkml I must be able to reproduce all the problems with the latest stable kernel - oh look, I can't.

There's nothing to defend here or go against. AMD has made the right move to finally support Linux, and some people are just acting like jackasses. This just shows people love complaining, and would do it any way -- they complained before AMD opened up the specs, and they complain now why 3D isn't yet working perfectly in the FOSS drivers, and they complained when AMD deprecated < r500 support, and so on and so forth. The same people who complain are the people who use Linux with a Windows mindset (i.e. "just work", no consideration for other aspects). They should just... hmph, use Windows and be happy with that?

[RealNC]

I know I'm feeding the trolls, but honestly..

Now you're replying with something about binary blobs to something about Free Software. My problem is not Free Software or binary blobs. My problem is drivers that don't work.

NVidia's driver is binary too, but it works.

Btw, isn't it funny that so many people are angry and are replying here with flamage about AMD, and you dismiss them all as trolls? Well so be it. We're all trolls. We have nothing better to do than trolling here. Which is natural since you can't do much with fglrx. NVidia owners have their hands full playing games and watching movies, so there's not much going on in the NVidia section of this forum. Explain a lot, I think.

[Tinuva]

Alright I might not have the latest and greatest ATI card, but I do have a R500 in my laptop, and the open source drivers work wonders on it. I for one are happy with what ATI did by open sourcing their drivers.

What we really need is for the latest cards' 3D to also enter the opensource driver and everyone should be fairly happy.

[val-gaav]

Bullsh!t. We have every right to complain since the drivers don't work correctly. We are customers. If AMD starts putting "Not suitable for Linux" stickers on their cards, THEN we won't have a right to complain.

Do those cards have any stickers besides the ones saying about Windows support ? If no then I fail to see your point. If they were to put a "we do not support Linux sticker" then they should do so for all the rest of OSes like Solaris, Haiku, FreeBSD GNU/Hurd etc. There is so many of those that I think there would simply be no place on the box for them ;)


Not even the kernel developers care about "FOSS environments". As Linus Torvalds put it: "Go suck on RMS." We care about working products and NVidia delivers those.
Well ... get your facts first :
https://www.linuxfoundation.org/en/Kernel_Driver_Statement
https://www.linuxfoundation.org/en/Lf_statement



Personally I would rather AMD stopped development of fglrx for the end user altogether (keep it for the corporate FireGL) and put resources into finishing the open source stack.



Well since AMD/ATI dropped support for my rs690 in the last catalyst release it really helped me ... I no longer face the "Damn maybe I should try fglrx for better performance" issue. Though I was happy with open source drivers for quite a while, yet from time to time (mostly when trying some games) I had this urge to try fglrx.

I think a lot of angry people here would be the same. Once there is 3d support for r600 and up in open source drivers fglrx should just be targeted at all those FireGL cards and the support for normal customer r600 r700 cards should be taken out of those.

[RealNC]

Read. I said "Now you're replying with something about binary blobs to something about Free Software."

Geez, READ.

[val-gaav]

I'm sorry but I just corrected your false statement that:

Not even the kernel developers care about "FOSS environments". As Linus Torvalds put it: "Go suck on RMS." We care about working products and NVidia delivers those.

I don't see me misreading it.As for your reply in the second post (which by the way I didn't read before submiting, as it takes time to write a long post ;) ) it states what you want and it's a fair opinion, yet you are wrong to back it up with false statement about kernel developers.

[RealNC]

I'm sorry but I just corrected your false statement that:

I don't see me misreading it.As for your reply in the second post (which by the way I didn't read before submiting, as it takes time to write a long post ;) ) it states what you want and it's a fair opinion, yet you are wrong to back it up with false statement about kernel developers.

Sure. That's why Linus said:

"Well, you do have to realize that Linux has never been an FSF project, and in fact has never even been a "Free Software" project."

Tell me, what has this to do with binary blobs? Again: READ. Someone wrote that Linux is about having a "FOSS environment". It's not! That's what I replied to.

Why are we even having this discussion?

[mendieta]

I don't see me misreading it.As for your reply in the second post (which by the way I didn't read before submiting, as it takes time to write a long post ;) ) it states what you want and it's a fair opinion, yet you are wrong to back it up with false statement about kernel developers.

Thanks for clarifying that. You saved me the time. Linux is pragmatic while RMS is more philosophical, but both dislike the use of binary-only code.

I the meantime, I posted a pointer to a thread that has been incredibly useful to a ton of people to get fglrx running, I suggest folks having problems take a look at it. The fglrx driver is pretty shaky: one setting off and you get a hard crash. OTOH, once you get it running it does give you good 3D accel (for 2D ONLY you are much better off with the open source drivers, even in newer chipsets)

[blindfrog]

Half-Life 2 Episode 2 works once again great (DX9) with wine-git! :) Hopefully they'll get proper video acceleration working soon...

[val-gaav]

@RealNC

FOSS stands for both free software and open source. As quoted from wikipedia:

'F/OSS' is an inclusive term generally synonymous with both free software and open source software which describe similar development models, but with differing cultures and philosophies. 'Free software' focuses on the philosophical freedoms it gives to users and 'open source' focuses on the perceived strengths of its peer-to-peer development model. Many people relate to both aspects and so 'F/OSS' is a term that can be used without particular bias towards either camp.

So mind you while Linux is not a FSF project and Linus prefers to call it open source rather then free software, when you use FOSS in your statements it is just as wiki states a term to be used with both camps.
Linux is pragmatic while RMS is more philosophical, but both dislike the use of binary-only code.

This sums it up well.

[Tares]

My HD4850 doesn't see any changes. Well, that's a good thing too ;-) nothing is broken :D

Although playing movies with compiz enabled is smooth with no issues, but when I try to rotate the cube its choppy and laggy :/ dunno why, because benchmarks shows 200fps.

[mirv]

Why do I have the feeling that the trolls are only out in force because almost all of their arguments have no merits anymore?
I'm not sure why, but this month the trolls are really getting to me. AMD's drivers work fine for me. Ok, compiz may have some trouble (guess what I don't use) but I've not had a single problem with AMD's drivers in a very long time. AMD's drivers also don't completely replace my glx stack (which I personally think is a good thing).
AMD's drivers are great. They're fantastic. They work. I want completely open source drivers, but in the mean time I can play all games I want, I can watch anything I want, dual monitors work perfect, and it all just works.

[mendieta]

Although playing movies with compiz enabled is smooth with no issues, but when I try to rotate the cube its choppy and laggy :/ dunno why, because benchmarks shows 200fps.

Have you tried the 10 bit enable?


aticonfig --set-pcs-val=DDX,OGLFMTA2R10G10B10Enable,1


From here:
http://forum.compiz-fusion.org/showthread.php?t=6794

HTH!

[energyman]

There's nothing to defend here or go against. AMD has made the right move to finally support Linux, and some people are just acting like jackasses. This just shows people love complaining, and would do it any way -- they complained before AMD opened up the specs, and they complain now why 3D isn't yet working perfectly in the FOSS drivers, and they complained when AMD deprecated < r500 support, and so on and so forth. The same people who complain are the people who use Linux with a Windows mindset (i.e. "just work", no consideration for other aspects). They should just... hmph, use Windows and be happy with that?


that 'right move' amd made is the single reason, why I have an amd graphic card on an amd board with amd cpu.
And don't tell me anything about 'windows mindset'. I invite you to look at my postings in the past.

I am also spending a lot of time on gentoo-user to help people. Windows mindset?

Because I am disappointed that AMD blew the chance away to be compatible with the not-even-latest kernel?

Go and kid yourself. You failed to amuse me.

[Pulfer]

Kubuntu Jaunty (using 2.6.30 now, but it also works on the default 2.6.28 kernel)
my xorg.conf is empty. The kernel uses radeon by default.

I switched from radeonhd to radeon driver and now playing videos with Totem loads CPU far not that much (~35% -> ~10% on my "test ones"). Looks like Xv is working fine. Kernel is 2.6.30 too.

Actually, the only thing I really miss in this driver is fan speed control (too noisy by default). I don't even need 3D much.

[NeoBrain]

Any idea when AMD will update the openSUSE RPMs? I'm still stuck with 9.2 right now as I don't want to install the driver manually because I never got it working that way...

Apart from that, any experiences whether a) alpha-blur works with Compiz or KWin now b) starting a second X server (xinit -- :2) with 3D accel works or if it does start at all?

[mendieta]

Any idea when AMD will update the openSUSE RPMs? I'm still stuck with 9.2 right now as I don't want to install the driver manually because I never got it working that way...

Apart from that, any experiences whether a) alpha-blur works with Compiz or KWin now b) starting a second X server (xinit -- :2) with 3D accel works or if it does start at all?

I think you can create SUSE RPMS from their installer, if they don't work you uninstall them and revert back. You can certainly do that for Ubuntu. Try passing --help to the installer.

[he_the_great]

Do those cards have any stickers besides the ones saying about Windows support ? If no then I fail to see your point. If they were to put a "we do not support Linux sticker" then they should do so for all the rest of OSes like Solaris, Haiku, FreeBSD GNU/Hurd etc. There is so many of those that I think there would simply be no place on the box for them ;)


Well ... get your facts first :
https://www.linuxfoundation.org/en/Kernel_Driver_Statement
https://www.linuxfoundation.org/en/Lf_statement


Linus isn't on that list.

[BlackStar]

wrong. I defended AMD in the last couple of month. But no 2.6.29 support really pisses me off. Why shouldn't I be allowed to say that? That kernel is out for thre(!) freaking month! 2.6.28 is unsupported, 2.6.27 a pile of crap.

And I have a good reason, why I want the latest kernels - I have problems with mtrr - but before I can even think about reporting on lkml I must be able to reproduce all the problems with the latest stable kernel - oh look, I can't.
Don't take it personally, RealNC was the one I had in mind. :)

For what it's worth, Catalyst can be patched to work with 2.6.29 (using it on Arch amd64 without problems - no idea about 2.6.30).

Edit: the kernel devs generally frown on reports with tainted kernels. Maybe you could move to the OSS drivers, reproduce the bug and report it?

Edit 2: Heiko, thanks for the hint on EXT_geometry_shader. Hope to see that implemented in the not-too-far future.

[val-gaav]

Linus isn't on that list.

Sure he isn't. Linus didn't sign that for whatever the reason he had for not doing so. Maybe he was pragmatic or wanted to remain neutral in a way.

though see this earlier post for Linus opinion about blobs:
http://www.phoronix.com/forums/showpost.php?p=78838&postcount=78

[Mithrandir]

Stop complaining and go contribute to open source ATI drivers. You have all required information to write good drivers.