Poll On ATI Linux Driver Possibilities

[Michael]

With the new fglrx driver coming out later this year, what do you think it will offer? Will it save ATI on the Linux landscape?

Share your ideas and vote in the poll.

[d2kx]

Longtime Phoronix readers know that Michael is in contact with AMD. They know that Michael isn't lying. And they see him hyping the "new ATI Linux Driver", as he calls it, more, and more, and more in the last weeks.

Remember that.

[Xipeos]

"Great!"
It'd better be. I just spent 150$ on an x1900XTX (against my better judgment (http://www.phoronix.com/forums/showthread.php?p=9053#post9053)).

Guess I trust Michael too much...

[lenrek]

Is too late, my side has completed the changes to Intel (CPU + Graphics). If there is anymore new system that we gonna purchase, it will most likely be a year or two from now.

So, for me, is too late. But then, if it does produce a better driver, maybe it will generate some excitement to some of my old boxes... Who knows?

[yoshi314]

i've already got two pci-e ati cards. one of which is pretty well supported by opensource driver (x300) . and one which is not (x1300). i really regret buying the latter - i don't use it anymore because of various fglrx issues that make using it a challenge. especially if u use a bleeding-edge like system like me (gentoo with lots of unstable scm-packages).

[sundown]

Michael, what did you vote ^_^?

I think it's going to be an improvment over the previous "old" driver but I don't have my hopes high just because I suspect that there will be lots of bug issues with the "new" one.

Jawbreaker would sing - "All I want is a life without parties" but all I want is smooth HD video playback on Linux and no problems with a 2nd display (smooth video playback on it as well). :(

[glussier]

I voted too little too late. For now I'm not buying any ATI video cards. If ATI's linux support, eventually proves to be on par with Nvidia, I might change my mind then. Up untill a few months ago I was purchasing only ATI cards, but for now, I'm one of their ex-customiers.

I know we need competition (nvidia vs ati or Intel vs AMD) but, I will support the underdog only if they can provide me with what I need.

[carpman]

Just AIGLX. Please let me get rid of Xgl.

[gfxdrone]

It makes no difference at all what ATI/AMD does now. They have a steady track record of failure, so just because they are saying they will change in those 'secret' ways that will be great if they ever land and work, the fact remain: drivers fail on new kernels, drivers break on new xorgs, driver roadmaps, that stupid 1 release per month no matter if things are worth releasing yet or not, all point to ongoing issues being the norm.

Is any ati driver even running on 2.6.23 for example? And will they backport support if they do get one working, or will users once again be stuck praying that the only working one works for them?

Only time will tell, anyone who buys an ati on faith that the changes will work and be solid is... well, an idealist at best.

AMD will need a minimum of 1 year (ONE) of no problems, supporting new xorgs, new kernels, before they are released, no breaks, no major failures, and, I almost forgot, a driver that actually does work.

I'm not holding my breath, if they have managed to reach these conditions by next summer, one year from now, or 1 year from when the new stuff comes out, then it might be worth taking a new look at them.

But this nonsense of not fixing things before they break and so on, that has to stop. This year's xorg failure was particularly stupid, since they even forced their biggest supporter, michael to have no support in fedora, which shows that not only is ati sloppy, but pretty darned clueless.

Here's hoping that some heads have in fact rolled in ati, and that amd has realized that if it doesn't get it's act together, everyone running linux is going to move to full native intel stuff. As it stands now, I'd be close to insane to recommend for any client considering linux machines that they run either amd or ati.

[Michael]

Michael, what did you vote ^_^?

You'll find out later this year when the new driver comes out.

[DarkFoss]

I'd like to see ATI''s Linux drivers performance on par with it's Windows drivers. Since that wasn't an option I didn't vote.
If the "new" drivers fail that..I'll be switching to Nvidia by years end.

[www1862]

Michael,i vote better.i trust you.when i want to buy a NV card,i decide that wait for the 'new' ati driver.but,i hope that it won't be too long before the 'new' relaese.sorry for my poor English.:D

[Regenwald]

i'm sure that michael is the only one who votet for the last option: "The best option for graphics drivers"...
:)
well, i can't believe that ati hasn't discovered the linux's possibilities and they simply can't ignore the market. when you read about autodesks enthusiasm in linux, more and more hollywood movies are made "by linux" (pixar etc.). you can't ignore the trend.

[user5124]

Poor drivers for year after year made me avoid even considering reviews for ATI.

After 4+ years of drivers so bad it made me laugh in sadness, I moved to nvidia.

The nvidia drivers on the other hand made me laugh in surprise. "It just works...wtf?...I even upgraded to the latest kernel and it's compiling easily/ok again!".

[Svartalf]

Longtime Phoronix readers know that Michael is in contact with AMD. They know that Michael isn't lying. And they see him hyping the "new ATI Linux Driver", as he calls it, more, and more, and more in the last weeks.

Remember that.

Oh, I know for certain that there's a new driver codebase in the works and that they're expecting to release it soon. The biggest thing I've got to ask is whether or not the thing is going to be something at least on a par with the windows driver's performance levels (And performance doesn't just mean speed, it also means correct rendering, memory consumption, etc...) or not.

[d2kx]

i'm sure that michael is the only one who votet for the last option: "The best option for graphics drivers"...

He hasn't.

[container]

I hope they will improve performance (as in speed, mostly). I suppose that is the biggest challenge. As for AIGLX, can it really be hard to implement? It seems funny that open source drivers can handle it, but it's still too difficult to do for the fglrx team.
But anyway, I'm looking forward to the new driver, and will postpone publishing my "ATI is deliberately keeping Linux drivers bad, possibly paid by Microsoft" hypothesis until it is out and tested.

[lenrek]

He hasn't.

Ah... I guess we know who that vote belongs to then... :p

Anyway, I guess we will know how the new driver turn out to be... But I think, judging from the poll so far, I guess not many of us is getting excited over it.

[Regenwald]

lol :)
well, i think that at least the opengl performance in windows, linux and mac are going to be nearly the same (perhaps 20% performance difference, not more). There was once an interview with an ati engineer and he said that the new driver will be based on the same codebase.
has someone still this link? i would like to read it again^^. (it was in "the truth about amd/ati" i think, but honestly, i don't know :) )

[Svartalf]

I hope they will improve performance (as in speed, mostly). I suppose that is the biggest challenge. As for AIGLX, can it really be hard to implement? It seems funny that open source drivers can handle it, but it's still too difficult to do for the fglrx team.


It's not that it's still too difficult- it's that they never really put the needed manpower on the currently shipping codebase versions of fglrx to fix things like that and the glaring, embarassing performance problem (Half of an already admittedly weak showing in the Windows world isn't very good at all...).

It appears that the situation has changed with the new owners- just not as quickly as they needed it to be happening. They've got something in hand that will allow them to give us AIGLX fairly easily, but concerns about them not understanding Linux well enough at the kernel level to get the performance up linger. I guess we'll see if they've figured out what they were doing wrong with their interrupt processing, etc. on the new drivers when they come out.


But anyway, I'm looking forward to the new driver, and will postpone publishing my "ATI is deliberately keeping Linux drivers bad, possibly paid by Microsoft" hypothesis until it is out and tested.

It wasn't that they were bought. ATI's upper management thought differently of the entire picture; they were going for where they thought the money was at- DirectX support and Set-Top boxes. Heh... We know where that's been going of late. OpenGL is very relevent everywhere OTHER than on Windows and X-Box as that's really the only segments that use them and the larger market doesn't USE Microsoft in the space; it's all VxWorks, RTEMS, pSOS, Linux, and Symbian. >:-) :D

They got a clue somewhere along the way (AMD buying them?) and got cracking on fixing the mess they made for themselves. The biggest question would be do they have it coming up or did they miss the mark with the new codebase...

[Svartalf]

lol :)
well, i think that at least the opengl performance in windows, linux and mac are going to be nearly the same (perhaps 20% performance difference, not more). There was once an interview with an ati engineer and he said that the new driver will be based on the same codebase.
has someone still this link? i would like to read it again^^. (it was in "the truth about amd/ati" i think, but honestly, i don't know :) )

Shouldn't be any performance difference if they've got the interfaces to the OSes done right. The Windows and Linux drivers are expected to be based off of the same main codebase, with interfaces to the OS accordingly done. The MacOS version of the OpenGL layer is a slightly different beastie, with Apple providing the userspace layers and the hardware vendors providing mainly the hardware level hooks for the OS and the userspace piece to drive, much more like the MesaGL/DRI code design.

[Regenwald]

what do you think: will it have support for opengl 3?

[Michael]

what do you think: will it have support for opengl 3?

Not when the initial new driver release is out.

[Malikith]

I think ATI's new driver will be pretty decent, when it comes out that is. Will it beat nVidia drivers? I think it may come close in the beginning and if ATI/AMD keep up the work on it, it will in time. But I think on release it will be good enough to please us all......

Or at least I hope.

[Michael]

I think ATI's new driver will be pretty decent, when it comes out that is. Will it beat nVidia drivers? I think it may come close in the beginning and if ATI/AMD keep up the work on it, it will in time. But I think on release it will be good enough to please us all......

Or at least I hope.

Your question will be answered in the near future.

[Nicolas]

Your question will be answered in the near future.

The problem is that the concept of near future is relative... We are always expecting that the next release is "the good one". Unfortunatelly it has been quite the opposite.

It would have been nice for AMD/ATI to have released a decent driver (with AIGLX support) in time for Ubuntu 7.10, but now it's too late.

But I'll keep my fingers crossed for the next one... I might be surprised.

I doubt the new driver will beat nVidia's... Not at first instance at least.

And I think it's difficult to compare performance between Windows and Linux, as there is much more things to consider than only the OpenGL driver. Of course the actual driver has very poor performance, but if we could only get near Windows performance, that would be nice.

[Xipeos]

wtf with the windows comparison already! ATI's OpenGL performance has always been bad, even in windows. The way I remember it, there's little performance difference in OpenGL between the two drivers.
Those of you asking for windows performance are asking for a bad driver! (they might listen).

I want a driver that can fulfill my graphics card's potential (and that's not the windows driver).

[Michael]

The problem is that the concept of near future is relative... We are always expecting that the next release is "the good one". Unfortunatelly it has been quite the opposite.


If I said anything other than "near future" or "later this year" or "this fall", AMD would probably send a hit squad after me :rolleyes:. Just know that it is in the near future and that fall in the US is very close, or if you consider fall the start when students go back to school, we are in fall already.

[www1862]

If I said anything other than "near future" or "later this year" or "this fall", AMD would probably send a hit squad after me :rolleyes:. Just know that it is in the near future and that fall in the US is very close, or if you consider fall the start when students go back to school, we are in fall already.
I hope so.God bless us!

[Michael]

FYI, if you subscribe to the Phoronix Forums and in your control panel options area have selected "Allow Administrator Email", you will be automatically notified when the announcement is made...

[Svartalf]

wtf with the windows comparison already! ATI's OpenGL performance has always been bad, even in windows. The way I remember it, there's little performance difference in OpenGL between the two drivers.
Those of you asking for windows performance are asking for a bad driver! (they might listen).

I want a driver that can fulfill my graphics card's potential (and that's not the windows driver).

Excuse me, but there's a big difference between the two- approximately half the overall performance under Linux versus Windows in many cases. We're not even AT the mediocre performance showing of the Windows OpenGL driver yet.

[Xipeos]

Excuse me, but there's a big difference between the two- approximately half the overall performance under Linux versus Windows in many cases.

Ok, I haven't used windows in quite a while, so I must believe you in this approximation. Don't you agree though that, game-wise, "barely/somewhat playable" (windows) is not that much better than "almost unplayable" (linux) ? (the current situation may not be *that* bad)


We're not even AT the mediocre performance showing of the Windows OpenGL driver yet.
Would you be happy with that kind of performance though? I wouldn't...

[d2kx]

OpenGL Windows Performance has improved a lot in the last years. It is as good as nVidias Performance "now".

[Svartalf]

Ok, I haven't used windows in quite a while, so I must believe you in this approximation. Don't you agree though that, game-wise, "barely/somewhat playable" (windows) is not that much better than "almost unplayable" (linux) ? (the current situation may not be *that* bad)


It is for at least part of the parts in the supported range for fglrx. In the end, even mediocre is better than what we currently have, which verges on abysmal for some of the chipsets.


Would you be happy with that kind of performance though? I wouldn't...


If you knew what I did as a pastime as much as anything else, you'd know that I do not consider the mediocre performance in Windows to be acceptable, let alone the worse performance under Linux. However, having said that, stellar's something that's going to be tough to get for them on a first cut of the new codebase derived drivers- I'm willing to ask for realistic goals, which would be parity with Windows at this point. Parity with what's about to come out for MacOS would be perfect, but I don't see that as a realistic goal for them- though they could surprise us all the same.

[Svartalf]

OpenGL Windows Performance has improved a lot in the last years. It is as good as nVidias Performance "now".

Depends on the application. There's some...odd...breakdowns in performance because of some design decisions they appear to have made in their drivers. (The silicon's not the one to blame in those instances...) Some games and some workstation apps can still drag the hottest machines to their knees in XP or Vista with ATI hardware and not with NVidia in the same context.

[Regenwald]

i've got some interesting "news" from this thread (http://www.forum-3dcenter.org/vbulletin/showthread.php?t=378087)
in a german 3d-forum. a "guest" (->anonymous) said that the new driver will be there in ~2 weeks. well, it's kind of rumour i think. (he talks about open source-driver, too. but he also knows the avivo project so it is not too fantastic since not everybody knows projects this kind). you can't take it that seriously, but i think every myth is based on a true event... :).
what dou you think? 2 weeks? sounds fantastically^^

[Michael]

i've got some interesting "news" from this thread (http://www.forum-3dcenter.org/vbulletin/showthread.php?t=378087)
in a german 3d-forum. a "guest" (->anonymous) said that the new driver will be there in ~2 weeks.

...

what dou you think? 2 weeks? sounds fantastically^^

Nope, it's not two weeks. It's ______...

[Xipeos]

Nope, it's not two weeks. It's ______...

I think I'll start a "Michael haters group". :)

[Regenwald]

well, you're not the only one, i would join^^
on 10th september, they are going to present their new barcelona cpu. perhaps there are similarities/overlapping?! so it seems to be this month because there is still no 8.41?!

[Michael]

You will find out in due time... :D

[Regenwald]

damn you make me crazy :O ;)
well it is very exciting! however, i'm still happy with my nvidia 7600gs with xinerama ;)

[Xipeos]

Michael, I remember there was a promise for a review showing cedega gold*, but it never came through. I'm assuming that cedega didn't meet the expectations, so the review was dropped.

I'd hate it if something similar happened with fglrx, after all these promises and your obvious certitude for improvement.

Can we at least be sure that your information is correct (ie. you're getting more than just some promises from ATI) ? It'd make you look like a "joe jerkoff", and people may really start hating you if ATI failed to deliver as promised.

Hopefully, you can answer the question without fear from the hit squad.

* as in "US gold", "dacian gold"

[Michael]

Michael, I remember there was a promise for a review showing cedega gold*, but it never came through. I'm assuming that cedega didn't meet the expectations, so the review was dropped.

Actually, in that case Transgaming would not send us the final build of Cedega that is available to the customers, just the preview/media build that still contained bugs.

Can we at least be sure that your information is correct (ie. you're getting more than just some promises from ATI) ?

The information shared thus far remains correct about the claims of what this driver will offer (performance improvements, R600, etc) plus more.

[Xipeos]

I'm relieved :)

AND

<insert big grin here>

[ivanovic]

The information shared thus far remains correct about the claims of what this driver will offer (performance improvements, R600, etc) plus more.

Ugh, if the guest in that other forum forum is correct (I hope he is) this "plus more" could mean open source drivers. If this was true I would directly consider buying a new graphics card from AMD/ATI. If it is not I will stay with my good old 9800Pro that is by now at least working using the open source drivers though speed is not this good. (UT2k4 is playable at 800x600 with medium details, Doom3 is just a slideshow as well as many other 3D games)

I hope this guy was not some strange kid speculating as maybe all of us were at the times when AMD did buy ATI...

[Malikith]

Hey Xipeos,

I took your signature for fun and I made something out of it, check this out, Michael should get a laugh out of this ;).

Originally Posted by Michael
Nope, it's not two weeks. It's ______...

If theres any meaning behind that towards the ATI drivers, count the characters haha..

_ _ _ _ _ _ . . . (9 characters)

S E P T E M B E R (9 Characters)

Hahahaha. Of course it probably doesn't mean anything ;). But I'm throwing out a good guess out there. Maybe that'll spark you guys up a bit. Definitely is fun. Plus my birthday is in this month so maybe that'll contribute, ATI/AMD can give me a birthday gift.

[Svartalf]

Nope, it's not two weeks. It's ______...

<*sigh*> You're such a brat, Michael... You're being needlessly cruel stringing us along like this. I'd almost swear you've been taking lessons from Howard Tayler in that regard. >:-) :D

[Svartalf]

Ugh, if the guest in the 3dcenter forum is correct (I hope hr is) this "plus more" could mean open source drivers. If this was true I would directly consider buying a new graphics card from AMD/ATI. If it is not I will stay with my good old 9800Pro that is by now at least working using the open source drivers though speed is not this good. (UT2k4 is playable at 800x600 with medium details, Doom3 is just a slideshow as well as many other 3D games)

Don't get your hopes up... They spent a long time in the making for those drivers- a LONG time. I would be tickled beyond words for opened sources, but that's NOT likely.

[Regenwald]

other question: are there "special" drivers from ati/nvidia for the professional series (firegl, nvidia quattro? as far as i know, nvidia do have seperate drivers?! does anyone know this?

[Michael]

other question: are there "special" drivers from ati/nvidia for the professional series (firegl, nvidia quattro? as far as i know, nvidia do have seperate drivers?! does anyone know this?

NVIDIA does have releases that are specifically designed for the Quadro or their VCS professional systems, but the normal consumer drivers should work with those GPUs as well. The workstation releases just have different quality assurance testing, etc.

ATI also does workstation testing and issue a new FireGL series update like every three months or so. It's the same driver, just additional testing goes into the release for the workstation programs.

[Michael]

Nope, it's not two weeks. It's ______...


It's tonight!

[Svartalf]

Seems that my naming might just be apt for it after all. >:-) :D

[Thetargos]

Had I seen this poll earlier than yesterday when I read about the new drivers, I most likely would have voted something like "Too little, too late"... However now since the 8.41 preview and numbers, I can only say that the next drivers will be GREAT, it was about dayum time, too!!! :D