Slashdot Mirror
Nvidia Releases Beta XFree86 4.0 Drivers
A lot of folks have been submitting the news from Nvidia that they've released
beta drivers for XFree86. They've got OpenGL acceleration - but are still in beta. You've been warned. *grin*
← Back to Stories (view on slashdot.org)
Finally I can upgrade to XFree86 4. I hope these will make Q3 playable under Linux so I can show off Linux at LAN parties. :)
Also, for all fellow Redhat users out there, the nVidia FAQ indicates that there are now RPMs at the Redhat mirror sites.
I'm downloading them, it's slow and have'nt had a time to look at them. Strange. Anyone knows?
Anybody else notice that in the URL, the Developer folder is under Marketing? Perhaps they're taking this Microsoft partnership a bit too seriously...
---------------------------------
---------------------------------
Visit
2.1.2 Limitations.
No Reverse Engineering. Customer may not reverse engineer, decompile, or
disassemble the SOFTWARE, nor attempt in any other manner to obtain the
source code. No Separation of Components. The SOFTWARE is licensed as a
single product. Its component parts may not be separated for use on more
than one computer, nor otherwise used separately from the other parts. No
Rental. Customer may not rent or lease the SOFTWARE to someone else.
Have fun, guys... I'm sticking with hardware that has free drivers.
________________________________
This isn't an open source driver. They forbid "Reverse-engineering".. you must own an nvidia card to even /use/ the software... the list goes on. "But wait.. I have an nvidia card and I want my driver. It's free!" Yes, you get your driver.. for linux. Only. No BeOS, no herd, no *BSD, nothing. Open Source allows the BSD crew to grab a linux driver, hack it to use BSD, and offer that support. Why can't Nvidia release the source so other (maybe less popular) OS' have a chance?
Blah. Nvidia needs to make a commitment - first it was obfusciated drivers, now just a binary. What next - shall we sign an NDA?
Don't be fooled by the .src.rpm's. They do not contain source code. (did they think they could get that by us?)
Let's hope that ATI releases open source drivers for the Radeon. It looks like that card will be the best thing out there in the next few months. (yes, it beats the GeForce 2 IMHO)
------
Contents of email below:
to: info@nvidia.com
I wasn't sure where else to send this, so I'm sending it to this address.
Thank you for your support of Linux and 3D. nVidia makes great 3D accellerators. I own a TNT2, and have been very impressed by the value it provided me. I have been looking forward to a high-performance driver solution for my card under Linux, and it's great to see your support of DRI. Thank you!
I speak for many Linux users when I say: Can we expect open-source drivers? While the binary-only module that you provide is well-supported in XFree 4.0 on x86/Linux, it does not address the needs of PPC users, Alpha users, *BSD users, and others who can also use XFree 4.0. I would like to note that your competitors (3DFx, ATI, and Matrox) have not only released open-source drivers (un-obfuscated!) but hardware register-level specs as well. Note that even the ATI Rage Pro (a weak card) was consistently out-performing even your GeForce GPU in Linux. While that may have changed as of this driver release, still it was the Linux community who wrote, tested, and finalized the ATI driver (mostly through the efforts of John Carmack). The Matrox G200 handily beat the TNT2 in Linux, thanks to the community. We both know the TNT2 kicks the G200 hard under Win32. My old Voodoo 2 slams all of these cards handily, since open-source drivers have been available the longest for this card. Plus, 3DFx actively supports these drivers themselves.
While I am not a businessman, I don't see how you can lose business by releasing these drivers and specs. Admittedly, some of these users would be a pretty small market, I don't think it costs much to release what you've already developed for another platform.
Your upcoming GeForce 2 sounds like a winner in the specs department, and I'd love to have one. I don't mean to sound ungrateful for your Linux support, but I'm leaning toward the purchase of another kind of card, either a 3DFx V5, or Matrox G450. Neither of these cards has all the specs that your Geforce 2 has (the fillrate plus features; EMBM, Cubic Mapping, 3D Textures, etc) but these companies have open Linux drivers and specs now, and I know I can expect this from them in the future.
Thanks for your time, and your Linux support,
"We apologize for the inconvenience."
here
--
A buddhist walks up to a hot dog stand and says ``Make me one with everything.''
Or at least descent into binary-hell. Yes, that's what idiocy nVidia is pushing on it's customers. Binaries. No source. There's no hope you'll find a GPL on *these* drivers.
So now, here's a release of binaries. For x86-Linux only. Nothing for x86-BSD. Or for PPC-anything. With source comes the option to port it. With the source and specs, heaven forbid we decide to IMPROVE their driver.
I reall wish nVidia would wake up. We can all say 'Hooray, drivers at last!' But as soon as we find bugs, we're SCREWED until they finally decide to release their next revision. And given how long these took to arive, and nVidia's track record for maintaining their previous 3.3.x GLX drivers.. Well.. Fuck em. Go buy a REAL video card. One that comes from a company that cares about you, the consumer.
No Reverse Engineering. Customer may not reverse engineer, decompile, or disassemble the SOFTWARE, nor attempt in any other manner to obtain the source code.
This is straight out of the EULA, which really makes one wonder considering that they have tarballs and src.rpm versions of this driver.
So, either they are giving away the source code and some knee-jerk lawer put this clause in with no understanding of what it meant or...
They are providing a closed source driver and don't understand what is to be expected in a src.rpm file.
Either way, you have to wonder, although I suspect (given their history) that the latter case is correct. Anybody who has been able to download from their slashdotted site please enlighten us as to what they are giving us.
Finkployd
I have the installation FAQ in front of me, and it's a long and careful list of things to do. From a quick scan, it looks like people of a nervous disposition should think twice before going down this list - making a quick backup of your current Xfree installation might not be a bad idea, or at least keep the old Xfree86 rpms at hand in case of crisis. Beyond that, it looks like it may conflict a bit with Mesa, so those modules need to be deleted or renamed as well (all in the FAQ).
For a speed comparison under the new drivers, Linux Games has a First Look up which gives me hope that I'll finally see some speed on my TNT2 card!
Cheers,
Toby Haynes
Anything I post is strictly my own thoughts and doesn't necessarily have anything to do with the opinions of IBM.
:wq
Um, I think their marketdroids just put their standard click thru license agreement on without actually burning any cpu cycles on it.
l ish/XFree86_40/
Because there is src rpms available right here:
ftp://ftp1.detonator.nvidia.com/pub/drivers/eng
No reverse engineering required.
Of course they may be obfuscated, still waiting for the download (damn the slashdot effect).
Can they beat THIS???
"In short: just say NO TO DRUGS, and maybe you won't end up like the Hurd people." --Linus Torvalds
There's enough good commentary now establishing that the drivers are a binary release (with a source stub for kernel version compatibility), and a fair amount of annoyance at that fact. One thing I haven't seen addressed is the question of *why* hardware companies like nVidia choose not to release source. I can see that they want to protect their R&D investment in the board's hardware and firmware capabilities, but would simply disclosing the API in the form of driver source really give away that much? I'm curious whether the decision not to do so represents fear and bad habits of closed-sourceness, or whether there a genuine justification (from their viewpoint). And if there is one, whether a method of release might be established that's better than this one (which might as well be binary-only as far as the non-x86 or non-Linux crowd is concerned).
This is a genuine question, not rhetorical; I'm not a video driver programmer so I don't know how much the source gives away about the underlying hardware, but my gut says that it can't be very much. An OpenGL driver is an OpenGL driver, surely.
-- Life is short. Forgive quickly. Kiss slowly. ~ Robert Doisneau
Since chromatic doesn't seem to be around today, I may as well mention a helpful little page he's put together, for those who haven't seen it yet:
/. by PI's Frank LaMonica.
What is Nvidia doing?
There's a lot of information there about their (lack of) support for free OS's, issues with the DRI, etc. It was put together shortly after the Nvidia and Linux Troubles article on
Have to add, of course, that if ATI's new Radeon hardware really does outdo the GeForce 2, and ATI releases programming specs for that puppy-- Nvidia is toast.
iSKUNK!
However.
If I had the source, I could recompile it for a different platform, and possibly port it to a different OS.
If I had the source, Alan Cox and John Carmack would have the source too, and that would most likely be a good thing.
Saying "YOU don't know what to do with the source" does NOT mean that source availability would not improve the driver.
---
You missed a little detail: their driver is a kernel module. Binary-only modules are apparently legal.
The binary-only driver will work fine on a BSD box just as long as you have the Linux kernel module loaded... hmmm...
----------------------------
The FAQ says riva 128 is supported by xfree 4.0 w/o these drivers, and while I know it's not a high performance chipset, I've got one kicking around. (in the box I'm typing from, actually) Anyone know/run one of them under xfree 4.0 and care to comment? or should I just run 3.3.6 and utah-glx? :)
bash: ispell: command not found
This sig left intentionally blank.
This nVidia driver release is probably the most amazing thing that has happened to me this week. First I get DSL, now this. Okay, I'm getting to my point. I would first like to congratulate nVidia on a driver release that is fast (though not completly stable, but hey, it's beta.) and very usable. This is a major step in the right direction for the Linux movement, though it does have its faults. First, it is not Open Source. I personally don't care, but I know the OSS community in general does. They think it is a bad thing for binary-only things to be used on Linux. While they may believe it's true, I also think they care about the quality of the Linux environment. Face it. Very few people use Linux because of its apparent freedom. People mainly use Linux because it is a very high quality environment. In that end, most care more about the quality of the environment than the freeness of it all. The nVidia drivers immensly increase the quality of the Linux environment, and in that end, it is a Good Thing. True, it trades freedom for that quality, but in the end, few people hack their video card drivers, but many people need high quality 3D acceleration. It is part of a broader trend of getting Linux accepted into the mainstream market. True, some think that Linux should stay a hacker-only system, but in the end, that too is detrimental to the quality of the system. Without the mainstream acceptance of Linux, 3D acceleration would have been unthinkable. Even three years ago, did anyone even think that someday the top consumer 3D company would write drivers explicitly for Linux? I doubt it. In addition to drivers, Quake, Corel Office, all the apps that are being ported, and partially KDE and GNOME, are due to the increasing acceptance of Linux. Although I doubt nVidia cares about the few of you who will boycott the GeForce based soley on the fact that there are no OSS drivers, I do think that it is important to encourage them and congratulate them on this release. (IE: Lay off the flame mail.) If they want to do more in the future, than that is their decision. Encouragement is good, but "OSS DO OR DIE" is bad.
A deep unwavering belief is a sure sign you're missing something...
AWESOME.
In quake, texture-intensive levels that really sucked in windows (q3tourney4) now run as fast as everything else.
My only concern is the mouse input. has anyone found a good bunch of settings that make the mouse input smoother? Things I've tried: in_dgamouse, sensitivity, and xset m X Y. anyone have the magic combination?
So someone should be able to port it to *BSD.
Well, the linuxgames article said they had a blank screen using the new drivers with just a normal Riva TNT, the reviewer chalked it up to his configuration error.
I have a Riva TNT, also experienced the blank screen, and from examining XFree86.0.log I found this interesting tidbit:
(**) NVIDIA(0): VideoRAM: 0 kBytes
I, uh, have a 16 meg card, setting the VideoRam parameter in the device setup seems to have no effect, whatever happens there is always 0 KB of video ram detected. Er, Slight bug huh NVidia? Maybe if they were open source drivers we could fix it instead of having to wait for you.
-- iCEBaLM
I am ashamed at all the negativity on this board. There are almost not positive comments and most are flames directed at the closed-sourceness of the drivers. Even ACs who otherwise would have been ignored are being moderated up because they are against nVidia. People are posting "no source is bad" without even backing up their reasons. I for one would like to put in a positive comment.
:)
A) This is good for Linux. The OSS die hards might not like it, and it is unfortunate that the Alpha people can't use it, but overall it is good. It furthers Linux in the home market and the desktop 3D workstation market. It make linux a higher quality, more usable environment.
B) It is good for Linux users. Now people with the fastest cards (GeForces) can lay the smack down on people puttering on with G400s and Voodoo 3s.
C) It shows that Linux is being treated equally among OSs. nVidia wouldn't release their source to Microsoft, and they aren't doing it for the OSS community.
I really don't care whether or not nVidia releases sources. Some people may, and if nVidia does, good for them. In the meantime, those people should congratulte nVidia on the release, and gently encourage them to release more source. (Hey I could benefit. BeOS needs GeForce specs!) Ultimately, however, it is their decision, and it is up to them what they want to do with their work. I do think, however, than an overly negitive response (as opposed to a positive, but gently encouraging response) could clam nVidia up from releasing sources. I doubt they'd be turned off to the Linux market, because SGI and nVidia have their own plans for Linux, but they may become even more closed and not port to other OSs (ahem, BeOS.)
PS: What is wrong with you people? Do any of you care about speed? Voodoo 5 has already shown to be only moderatly faster than a GeForce but you'd prefer an open Voodoo, rather than a closed GeForce 2? Doesn't anyone care about SPEEEED!??
A deep unwavering belief is a sure sign you're missing something...
"Patience is a virtue, afforded those with nothing better to do" - I don't remeber
"Patience is a virtue, afforded those with nothing better to do." - I don't remember
Okay, whilst I welcome this as much as anyone else (I own a TNT2u, and I didn't pay 180UK pounds for it just to be forced to boot to Windows everytime I want to use it properly...), I have to say that this sucks. It sucks because:
a) It's a binary only release, with no specs for the chipsets, etc
b) It's a beta.
a) sucks for all the obvious reasons - the "community" can't make bug fixes or learn from it, only x86 Linux is supported so everyone else is SOL, etc.
b) sucks for a slightly less obvious reason. Both of the previous driver releases, for XFree3, were also beta "development only" releases. When are we going to get the real thing?
Yes, I realise that this isn't easy, and that they have a decade or two more man-hours in Windows driver development, and that Linux is a niche market, etc etc, but damnit, I want real drivers! Failing that, I want the chance to fail to write them myself!
Thank you, NVidia, for a really good graphics card, but I'm afraid that my next card is going to have to be from a company that shows a little more commitment to my OS of choice.
Cheers,
Tim
It's official. Most of you are morons.
Sure, it's good that NVidia is paying attention to the Linux market - course at 12,000,000 and growing exponentially they don't really have much choice - but they'd pay a whole lot more attention if you sat this round of NVidia products out and went with Matrox, which reportedly has the best rendering quality in the field, is far from shabby in performance, and has made the fullest release of technical specs so far, which can only result in the best possible drivers. Matrox needs to be rewarded for this, and NVidia needs to learn why secret specs as bad.
--
Life's a bitch but somebody's gotta do it.
I don't think that nvidia is evil, but stupid I wouldn't bet against. They could have had these drivers written for them, free! Why wouldn't they want that? I had to sign a non-disclosure once, to get some schematics (computer industry, but a different part of it), and when I saw them, I saw why. They must have been ashamed of their work, and afraid that the world would find out just how bad their design was. It was full of text-book no-no's. I figure that you only hide things you are ashamed of. Any competitor who's serious could buy a board, take the chips apart in a clean room, and so on... I can't imagine that they really think that they're protecting themselves.
So, I'm not going to buy an nvidia board until I see open source, accelerated drivers for it. Not because of ideology, not because I hate nvidia, but just because I want my machine to work, today and next year, and the year after that.
I'm not criticising nvidia, nor suggesting that releasing these binary drivers is bad. But I can't get excited over this. It's not good enough news to get me to trust them to deliver working software, when I need it. If someone with the resources of Microsoft can't deliver that, little nvidia never will.
See what I've been reading.
Isn't that how it works though? The EULA forces you to If you won't agree to the restrictions, you are violating the license, just as signing a non-disclosure agreement removes some of your "freedom of speech" because you can't talk or write about what you've learned.
There is also fair use in the purchase of copyrighted material. Can you imagine buying a cookbook and not being able to use any of the recipes because reverse engineering of its methods was prohibited ?? There is no aspect of copyright law that protects against reverse engineering of the methods used in that copyright (only against reverse engineering measures designed to protect copyright itself, such as DMCA).
I cannot imagine a judge upholding that an EULA actually does prohibit reverse engineering. It is a pretty absurd notion, and makes copyrights into patents. That is a large portion of the reason that copyright does NOT protect against reverse engineering - there is an entirely much more stringent form of protection for that - the patent. You actually have to apply for a patent, and be reviewed. I can scribble my name on the wall and copyright it. It is taking a triviality and trying to make it into a patent. It is just plain wrong, and I find it hard to believe any copyright scholar would argue the point. There are some rights that cannot be placed in EULAs by law, and I think reverse engineering is one of them.
Section 10.0b, US copyright law
In no case does copyright protection for an original work of authorship extend to any idea, procedure, process, system, method of operation, concept, principle, or discovery, regardless of the form in which it is described, explained, illustrated, or embodied in such work.
Consider, who is most likely to what the highest horsepower graphics cards out there? Well, that would be those people who play lots of games, like lots of high-tech toys, and basically are "in the know" when it comes to computers. In some senses, that is US my fellow slashdotters.
Well, let's see here, it also happens to be the case that we are also the life, liberty, and the open source way kind of people. Hence by not playing straight with their drivers (and bowing to M$'s demands), they are estranging their main purchasing base.
Now, this does assume a high degree of overlap between those who want the card but won't buy the card because they are TRUE BELIEVERS, but I think it is a fair statement. Even if it isn't, we're smart enough to realize that even if I'm driving a Ferrari, it needs to have good fuel to make it run. That is, no matter what the card, if its software component is stuck in, oh say, reverse, then it is not worthwhile.
Regards.
Unless I downgrade from the 2.3 kernel I'm using at the moment, I can't use these drivers, because they don't support 2.3. That stinks. That's why people are pissed off. That's why I plan on buying _anything_ but nvidia in the future.
:-(
Yes, the fact that they've released drivers is good, but you have to understand that Linux isn't just another operating system, that you can release binary drivers for and have them work - it's not like windows in this sense. There is no ABI for Linux, so there's no way that you can write a driver that will be guaranteed to work on more than one version of the kernel. It could easily break between minor versions (even of the stable kernel), and there's no way that it would work between major revisions. Now, if they'd released source for the kernel module at least, then that problem would go away, because people who wanted to use their nvidia cards under 2.3 would port the driver themselves, but with binary only, that can't happen.
It's pretty much the same with the X side of things, though the X developers go to more trouble to maintain binary compatibility. But even with that, there's a very good chance that things will fail at some point, and people will have to wait on nvidia for fixes. That's not good enough, not for an open source world.
And the final nail in the coffin of these drivers is the fact that they're ia32 only - there are a lot of PowerPC users out there, and the numbers are growing. They can't use these drivers, and they probably have absolutely no chance of getting working drivers for their platform. Again, if this was a source release, the problem would dissappear almost overnight.
Ultimately, the problem is that you can't treat Linux or any open source system the same way that you'd treat a binary-only OS. You can't treat the code the same way, because it behaves very differently, and you can't treat the users the same way, because they behave differently as well, and they expect different treatment. Nvidia might have produced some quite nice drivers for their cards here (and they do seem pretty fast, certainly compared to the original release), but they've screwed up in the long run by caring more about their paranoia than their customers. I don't know whether nvidia will lose out in the market in the long run, but I know they've lost out in the open source market because of this.
himi
Still pissed off . . .
--
My very own DeCSS mirror.
They cover this in the first part of the license... They are not _selling_ you anything, they are licensing it to you.
/usr/X11R6/bin. You use it, and reverse engineer it, and contribute the code to XFree86. And you post the code on the web. Of what, exactly, are you liable, if anything. You accepted no agreements. In fact, to gain access to the driver you need not accept ANY agreement.
Also a total crock that would never hold up. If you stop licensing it, can you give it back to them ?? Are there any continued costs of usage ?? There are NO licensing fees, a critical aspect to licensing. You do not pay per unit time. You pay to be able to use software ad infinitum. The software company meets NO criteria for ownership even if their software EULA says so. These agreements are not worth the electrons used to print them. They can not and would not hold up in any court.
Consider this. You buy an NVIDIA card, but never read the licensing agreement. You instead scan the media that came with the card (that you bought), and download the XServer module and dump it into
The EULA is quite simply the result of software barons trying to extend their protection by bluffing people who don't know better.
I downloaded/installed the drivers... found all the XMMS plugins crash when I try to use them. Hmph. Anyone else expeirencing this? Pulsar -fps gets 70fps and I'm running in 32bit 1280x1024. Oh ya, glplanet seems to not work either.
Ian
Go with Matrox.
The Utah-glx project will provide open drivers.
Also, if you're the type of slashdot reader that opposes closed APIs protected by copyright law, then you probably don't want to support 3dfx, because of their attempts to monopolize the 3D card market with Glide.
... at least in the EU. While you can sign a contract that give up your right to reverse engineer a program, the contract will not be legaly binding.
They now have their development environment setup so that their windows and linux drivers use nearly the same code base( methinks ), so that whenever they fix a bug/add a feature to the windows or linux driver the improvement shows up in both. I think this is very ingenius.
Yes, I agree. But as often as the cries of "open source!" have been raised here, it's not the source code that really matters. It's the programming specs. Source means you can get a really good driver for XFree86. Specs means you get a really good driver for XFree86, BeOS, OS/2, GGI, SVGAlib, and any other OS/graphics system people care enough about. After all, if you want to write a driver for a system other than the one supported by the released source, you have to grok the specs from that source first. Published specs save you that big first step.
Come on, do you see Intel telling amd "here are the circuit layouts for our IA-64 architecure"
No, but Intel does publish the register and instruction interfaces. They tell you how to actually use their hardware.
Imagine if that weren't the case. Say that x86 Linux systems couldn't use gcc, because no docs are available to write up a machine specification. Instead, they bundle icc (Intel C compiler), which is available in binary-only form from Intel's FTP site. All binaries on the system are built with it. You can't program the CPU directly (in assembler) because the information necessary to do that is protected IP.
Does that not strike you as an utterly preposterous situation?
Should the register interface to a video card be any different?
I think a lot of people have lost sight of what sheer folly it is to sell computer hardware components without making available information on how to interface with them at the hardware level. That's not protected IP. That's something one ought to be able to take for granted!
Yes, it's true that Nvidia's register specs may contain trade secrets. It's actually the case that their driver implements a lot of functionality that would normally either be in firmware or silicon (Nvidia cards don't have a clean "register interface" in the usual sense, or at least they haven't). But what does that mean? That means they did not design the hardware correctly. They got careless. And 9 out of 10 says it's because they've been used to shipping only Windows drivers, where binary-only is not only a non-issue, it's the norm. There was no concrete reason to keep the interface IP-clean, or even publicly documented, so both those goals were dropped as soon as it was convenient.
in the meantime I'll be enjoying my nvidia cards, closed-source drivers and all. If nvidia is willing to maintain linux drivers for you, what are you complaining about?
Alpha and PPC users have plenty to complain about. BeOS and OS/2 users have plenty to complain about. GeForce on Linux users might very possibly have something to complain about when the GeForce 4 rolls around, and Nvidia wants to boost sales.
The real kicker, however, is that if Nvidia did things correctly, all those users could be made happy at no additional expense-- if only Nvidia would allow it!
A line needs to be drawn somewhere. What should be free, what shouldn't be free? The guys at nvidia put a lot of hard work into their designs and can't afford to give them away. They have to make _a_living_ off of this and I don't think they'll jeopardize their livelihood in the name of open-source.
iSKUNK!
> That implies that having two monitors is extremely important.
:)
If you want more games for Linux, it is, in a round-about way.
1) As a 3d graphics programmer, having 2 monitors hooked up via the voodoo has been a God-send. 3D output on the voodoo connector, source code on the vga connector. Very productive.
2) All the artists where I work, have dual monitors (even dual 17" work quite well.) Why? Because 3D Studio MAX pretty much requires it to get any "serious work" done. (Hey modelling game objects is serious work
Of course if we had a good modeller on Linux that would help. Anyone know of any at the same quality as 3D Studio Max ??
> most of the dual headed cards out there only worked in WinNT
For PC's, yes. Usually the modellers are running NT since most of the "high-end" modelling tools have been written for NT.
Cheers
What makes a video card fast is 90% done by what's in the core of the chip, not the driver. For example, Voodoo1 vs. Voodoo2 - Before 3Dfx released a Voodoo2-aware glide for Linux, someone discovered that you could get the voodoo2 to work with the V1 drivers with 95% of the performance of the Voodoo2. (Which was about 2-3x that of the V1). All that was needed was to change a byte in the card-detection routine. Yes, the V2 had some additional features that needed driver modifications, but it's not like those were anything revolutionary - higher end boards had been using the same techniques for years.
Even if NVidia has something super-special that their drivers reveal, that's what patents were made for! If it's patented, it can't be stolen and used by another hardware vendor even if the source is available.
retrorocket.o not found, launch anyway?
I'm pretty sure you need to apply some kind of license to be able to download XServer module for nVidia powered card. And just that license especially denies your right to reverce engineer it - if you don't apply it you cannot legally download module. Of course IANAL.
Except that you can download the rpms without accepting the license. Of course, the page says that by downloading you are accepting the license, but that is also a load of bull. No price, and no restrictions on access, and no signature equals no license. They make NO attempt to make you read the license, and place NO restrictions on license acceptance in order to receive the rpms.
Again, this is quite simply nonsense being pushed with the idea that you will not see through it. Besides, the drivers are copyrighted, and copyright NEVER protects against reverse engineering of the methods that are copyrighted (only copyright protection mechanisms may not be reverse engineered, and only in the US).
This entire change in public perception on the actual value of copyright is ridiculous. For example, Microsoft claims they have the right, through copyright protection, to mandate certain things occur on first bootup. The judge thinks that is ridiculous. Copyright provides NO such protections. The software industry stands to make a TREMENDOUS amount of money if things like UCITA and DMCA are upheld and/or extended. They would like to make money by taking rights you currently hold.
I speak from a users perspective. I can't edit the source to KDE to fix bugs, and only a few users can. Not programming skill entirely, but the few weeks that it would take to get up do date on the KDE source are not worth it for me. Similar thing with drivers. Most users cannot fix these bugs, and in the end wait of KDE or Utah GLX to release a fix rather than wait on nVidia or Microsoft. It has not clearly been shown that bugs are fixed faster in Open Source drivers. Sure, some execeptions exist, like the ATI drivers, but in general, OSS software (At least major stuff like Utah GLX or KDE or Mesa, etc) are just as bloated and buggy and take time to fix than it's closed source cousins. (The kernel is a notable exception mainly because it has an overwhelming amount of support.) OSS software just costs less, and if there really is a small showstopping glitch that you can fix, you can do it on OSS software. However, those cases are extremely rare to the end user.
A deep unwavering belief is a sure sign you're missing something...