Slashdot Mirror
The State of ATI Drivers on GNU/Linux
An anonymous reader writes "After 50 days of the Phoronix editor-in-chief exclusively using ATI Radeon hardware in his system, he has issued his final blog post dubbed The State of ATI Linux. Topics covered include the very low frame-rate performance, image quality, overclocking X.Org 7.1 support, Big Desktop/Dual Head, Linux CrossFire, and other relevant items to gamers and Linux enthusiasts. From the article 'While discussing this trial with a colleague that was not familiar with the quality of ATI's Linux drivers he immediately classified ATI Technologies as attempting to fine-tune a hull on a ship while there is a giant hole in the side. However, is this truly the case?'."
Yes.
The state of ATI drivers on Windows is pure crap. It's even worse on Linux.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Forget ATI's proprietary drivers. The fact remains that they are useless blobs of binary code. For old hardware, I get better performance and reliability using the radeon driver which comes with Xorg.
This free driver, however, needs improvement. This is where energy needs to be focused.
Other reason: I use OpenBSD. I can't use fglrx. Don't care about it. Won't do me any good. When people tell me about ATI drivers on Linux (read: fglrx nonsense), I tune them out. It is worthless information. It really is such a shame that Linux people are so willing to sell their soul to a proprietary driver.
A good driver (read: free) would mean freedom for all platforms. This would also give developers a chance to update the driver as things change, which would fix reliability problems.
I'll go you one better. Binary drivers aren't compatible with Free Software, period.
It's amazing to me the author of the article can put out as much verbiage as he did about this issue without ever once mentioning the real problem here - ATIs refusal to document the card interface so that the hardware can be properly supported.
Until they do, their customers that use Linux, *BSD, etc. remain broadly unsupported. Only a small subset of free systems even have the option of using the mushware they want to substitute for documentation, and at a cost many will not pay. They're making themselves irrelevant in what is probably the fastest growing segment of the computer market. Why would a free software user shell out big bucks for the latest ATI *or* Nvidious card only to face the choice of running it without accelleration for the same performance as a much cheaper card, or with buggy opaque mushware that that doesn't perform that much better and taints your system, assuming it will even run on it, which it often won't?
=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Friends don't let friends enable ecmascript.
Having some experience with writing free software drivers for ATI cards I must (sadly) concur.
The problem is that it takes up to a year to make a half-way usable 3d driver - with the specs. Which means the developers should get specifications at least 6 months before the release of the new card for the drivers to be of relevance.
The reality is that we are lucky to get specifications 1 year after the release has happenned.
With regard to open hardware this is a great idea for many reasons beside the availability of specs.
For example, I would really, really, not mind paying extra $5 so that the graphics card does not lockup the moment it receives a slightly malformed command. Or so that it has a timeout and does not hang the PCI bus forever on a wrong address. The general-purpose CPUs have got this for ages - they just throw an interrupt and go on.
It'd be less of a problem if they
1. Spent more time to document the damn hardware
2. Opened the interface to the public.
The problem is the hardware is always in a state of flux and just incremental improvements. Your GeForce 7800 is probably based on the same HDL source as the 6600 with appropriate changes. This means that legacy symbol names from one project creep up into the new space. You get odd names, combined with lack of comments and documentation [compliance] leads to hardware with "oops" that the drivers have to work around.
Things could vastly improve for the customer if they stopped pretending that they know best. I know for a fact that companies like ATI and Nvidia spend a good deal of time [re: cost of the video card] in DRM technologies. Basically they don't give two shits about you as a customer so long as you
a) feel inadequate with your 75W GPU and buy the next best thing next quarter
b) fully comply with their "dominance" of your machine, force you to run windows, force you to use their bloaty drivers, use their drm, etc
I tolerate Nvidia solely because their kernel modules work decently [well not anymore as they're not keeping pace with xorg development]. Opening the 2D and 3D accelerators to the public can only serve to make the hardware more popular.
Their value is in the hardware and the ability to develiver it. Not the interface that puts a triangle on the screen.
Tom
Someday, I'll have a real sig.
Well, unfortunately, ATI doesn't see it that way. Their profit margins are razor thin. They're not about to sink more than token effort into supporting a minority user base. They would make negative profit. To them, if you want to use their products, you can use them the way they're intended. Otherwise, buy something else. The fact that something else doesn't exist is not their probem. This is a business we're talking about, and in business, profit is the top concern. If you want to do humanitarian work, you start a non-profit. If you want to make profit, you start a business. That's why we have terms like "business" and "non-profit" to describe these kinds of entities. For the most part, don't expect non-business things to come out of businesses. (There are exceptions, but ATI isn't one of them.)
Their are wonderful OSS drivers for 2D features, it is hardware-accelerated 3D where things fall short. HA3D means AGP or PCI-Express, not PCI. Even with support there is no way the OGP is going to release anything usable on that front for YEARS.
, 39352584-2,00.htm
On the other hand, Intel has been providing specs and source code for their integrated graphics chipsets. This includes hardware accelerated 3D, though the chips aren't up to the nVidia and ATI top or upper-mid range. Hardware T&L is missing, for one thing. However, their next refresh of those chips should get much closer and should still have excellent OSS drivers.
Intel offers much more hope than OGP ever will, as noble as that effort is.
http://www.die.net/doc/linux/man/man4/i810.4.html
http://www.zdnetasia.com/news/software/0,39044164
-Charles
Learning HOW to think is more important than learning WHAT to think.
The hidden premise here is that somehow documenting the interface will make it easier for competitors to 'steal' some advantage. That's so obviously wrong in so many ways it's shocking someone would assert it in good faith.
What are they going to do? Copy the interface so their card will be compatible with the other cards drivers? Well, yes, I suppose someone could do that. Wouldn't necessarily even be a bad thing (standard interfaces are generally considered a good thing, even ad hoc standards.) But this is a far cry from somehow "stealing" the actual video card technology. That technology is, in many cases, patented, rather than protected as a trade secret, so the competition can (and you can bet, has) gone and read the patents right off anyway. They just can't legally imitate it too closely. And to the extent there are things in hardware that *are* trade-secrets, a disection of one of the cards would be a much better way to get at them. Looking at the external interface is the last method one would use to try to disect the inner workings of a device. Note that refusing to disclose the interface doesn't do jack to stop the competitors from disecting the hardware.
No, I'm sorry, that whole line of argument is utter nonsense.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Friends don't let friends enable ecmascript.
How is documenting "register F7C0 does foo" going to reveal how the chip is actually architected?
How is not documenting the registers going to prevent NVidia from putting an ATI chip under an electron microscope to analyze their circuits?
Face it. If you're one chip fab competing against another one, documenting the externally-exposed registers for programmers is NOT going to deter your competitors in the slightest, nor is releasing binary-only drivers. Remember, decompiling code for reverse engineering IS legal (just don't copy & paste the code , recompile, and call it your own, that's copyright infringement) and decompilers are readily available, so there is NO advantage ATI has over Nvidia, or NVidia has over ATI by not open sourcing drivers.
The only thing that they are doing is alienating potential customers and slowing down the progression of open source. NVidia does have an advantage because while ATI's drivers totally suck, NVidia's drivers actually work so most of us accept NVidia's drivers. Of course, if I were wanting to upgrade from xorg 6.9 to 7.1 right now, I'd be pissing and moaning about binary-only releases right now (the fact that I am STILL running a piece of shit ATI card now is immaterial ATM).
The Christian Right is Neither (Christian nor right). See: Matthew 23, Matthew 25, Ezekiel 16:48-50
"I plan to be very very careful when buying ATI again."
After learning exactly how "awesome" ATI's driver support was when I tried to setup 3D with my Radeon 8500 (and also Xinerama, etc) to play WoW under Cedega, in both 32-bit and 64-bit modes, I switched to nVidia and haven't looked back (yes, nVidia's drivers ran with Cedega and WoW in both 32-bit and 64-bit Linux installs perfectly well).
So, I suggest to you, to never buy ATI again. Saying you'll be careful when buying ATI again, is like saying you'll be careful when shoving a live scorpion into your pants again. ATI is shit. Regardless of what their hardware might do, if you don't have drivers to make it do it, it's the same as not having the card!
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
If they make the specs public then people will know when the company hacks the drivers to cheat at benchmarks. I'm sure most companies, to some extent, mess with the card to make the benchmark programs happy. It's just too important for them not to manage that kind of publicity. So if the specks were public then they might get caught by someone saying. "HEY your card doesn't do what the benchmarks say it does".