| |||||||||||||
|
| |||||||||||||
Reverse engineering: more than NVIDIA deserves?Reverse engineering: more than NVIDIA deserves?Posted Feb 18, 2008 20:22 UTC (Mon) by tetromino (subscriber, #33846)Parent article: Reverse engineering: more than NVIDIA deserves?
You know, before complaining about NVIDIA's lack of cooperation, we should look at the respective track records of NVIDIA's closed-source driver team and of the free software community. And if you take an honest, hard look, you will see that NVIDIA's closed-source Linux drivers beat *all* existing free drivers, for any piece of graphics hardware, in terms of how much performance they can extract from the hardware, their reliability, the speed with which support is added for new hardware, and the rate at which new features are implemented. I hope that one day, we the community will finally start making drivers that actually work on all hardware that we claim them to work on, that really do properly support all features that we claim to support, and that don't cause our machines to lock up for no good reason. But until that day comes, instead of complaining about NVIDIA, we should be thanking them for releasing an excellent (yet unfortunately closed-source) driver.
(Log in to post comments)
Reverse engineering: more than NVIDIA deserves? Posted Feb 18, 2008 20:37 UTC (Mon) by jospoortvliet (subscriber, #33164) [Link] I think you missed the point of the article entirely...
Reverse engineering: more than NVIDIA deserves? Posted Feb 18, 2008 22:11 UTC (Mon) by einstein (subscriber, #2052) [Link] I wouldn't say that he missed the point, but that he's reflecting some good old fashioned practicality into what was a purely philosophical discussion.
He IS missing the point Posted Feb 19, 2008 8:22 UTC (Tue) by khim (guest, #9252) [Link] I have nothing against nVidia and binary drivers per se: you want/need to use them - to do. No problem. But to spend a lot of resources to produce open-source drivers for nVidia looks like useless waste of time: by the time free drivers will reach maturity (year from past experience) nVidia hardware will be obsoleted anyway. And for the next 2-3 years closed-source proprietary drivers are Ok. So the only way to keep Nouveau relevant is to waste efforts forever and keep adding support for newer and newer models. And this looks like stupid thing to do. But of course there are no rush: if/when AMD's drivers will reach maturity Nouveau support will dry up. I think it'll happen automatically if free AMD drivers will be any good...
You're missing the point Posted Feb 19, 2008 9:37 UTC (Tue) by liljencrantz (subscriber, #28458) [Link] Strongly disagree. nVidias drivers are flaky and unstable in my experience. Using them makes my laptop significantly less likely to come out of sleep mode working. There are various minor graphical glitches when using Compiz, like window borders sometimes getting the wrong transparency level and the mouse cursor getting the wrong color. Sometimes when switching resolution, the screen starts to flicker, and doesn't stop until you reboot you system. On my previous desktop machine, the computer would lock hard when using both video outputs at once. Because my desktop machine had an old graphics card and my laptop has a low volume laptop GPU, these issues are not prioritized by nVidia and will very likely never get fixed. All in all, the driver is crappy, and because it is closed source, I have no other option than to sit around and hope nVidia will accidentally fix this in a later release, which seems somewhat unlikely.
You're missing the point Posted Feb 20, 2008 6:04 UTC (Wed) by drag (subscriber, #31333) [Link] > All in all, the driver is crappy, and because it is closed source, I have no other option than to sit around and hope nVidia will accidentally fix this in a later release, which seems somewhat unlikely. Well, with hindsight, what you should of done was never purchase the hardware in the first place. With these sort of things your not only paying for the hardware, but your paying for the software. That is without the drivers the hardware is useless (or nearly useless). Nvidia, by keeping stuff closed, is using their software as part of the product you are purchasing. One is nothing without the other. That is because you bought Nvidia's hardware you are paying for them to continue to develop closed source drivers. (In comparison how much have you paid to develop open source drivers?) So the other course of action, right now, is sell your Nvidia stuff and buy something with Intel on it. So if you have problems with their driver support yourself or somebody can actually do something about it. If you cannot do that because of the money. Oh well. Better luck next time. If you cannot work with Intel hardware because of performance issues, then you'll never be able to work with the open source nvidia drivers either. There will never be the developer support or community support nessicary to get those drivers up to the same level of Nvidia's own drivers in a timely fasion. So your screwed coming and going.
You're missing the point Posted Feb 20, 2008 15:42 UTC (Wed) by liljencrantz (subscriber, #28458) [Link] Your points are mostly valid, but Intel doesn't sell any desktop cards, and back when I bought my hardware ATI where nowhere near open either, so I really had no option there. I really wanted to buy a laptop with an Intel GPU, but the exakt same laptop that I bought but with Intel graphics cost a full $600 more than the one I bought. They where selling the laptop I bought at almost half price because they had introduced the exact same model but with Vista certification. Thanks, Micorsoft!!! While I strongly wanted to support Intel, $600 was simply too much, and I caved in. I have some regrets.
You're missing the point Posted Feb 21, 2008 2:17 UTC (Thu) by pr1268 (subscriber, #24648) [Link] Saving $600 is worth a few regrets.
You're missing the point Posted Feb 20, 2008 22:20 UTC (Wed) by jayavarman (subscriber, #19600) [Link] My experience has been quite different. On my laptop the nvidia driver works pretty much well and stably. I just close the lid, the laptop sleeps and then comes back again. I can get several days of uptime. The power management seems to also work quite well, although the chip/driver is a bit too laggy to increase the frequency when needed.
Reverse engineering: more than NVIDIA deserves? Posted Feb 28, 2008 21:34 UTC (Thu) by alext (guest, #7589) [Link] Not really, he is saying that Free Software developers can't develop as good drivers with no documentation of the hardware as people who have are informed about every essential nugget of configuring and operating the boards to the max. If the NVidia drivers weren't the best you really would hope they would fire the engineers! I've thought from the start nouveau was a mistake and more so every time AMD or Intel release spec's for hardware. My next machine won't have NVidia in it unless they are free by then and I think this whole affair is just making people realise that others deserve their support where they didn't know different before. I didn't. So NVidia need to react or start to lose share. There are practical reasons, if I can have a machine that requires no special treatment (even with the helpful Ubuntu restricted drivers tool for example) to run at its best I will choose it over one with just the hiccup of an extra task. That's the practical reality.
Reverse engineering: more than NVIDIA deserves? Posted Feb 18, 2008 22:09 UTC (Mon) by ikm (subscriber, #493) [Link] I second that. It's just awkward to state that Nvidia is "giving us nothing". Their drivers are unmatched. Yes, they are closed source. And, as such, yes, if you've got a problem with your
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 9:27 UTC (Tue) by yodermk (subscriber, #3803) [Link] > It's just awkward to state that Nvidia is "giving us nothing". Depends on who the "us" is. In the context of this article (and, just about every article on LWN), "us" is the Free Software community, those who care about openness because of its many advantages. In that context, nVidia has truly given us nothing. They sell hardware that works with Linux (because they made it work), but that does not advance the cause of those who believe they should be able to know about the internals of the hardware they buy.
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 16:20 UTC (Tue) by ikm (subscriber, #493) [Link] Is it right to draw the line the way you did between open-source developers and mere users who just need to get some work done today? Furthermore, isn't it just impolite to throw stones that big into a company which actually does provide linux support, though in the only form they can?
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 18:12 UTC (Tue) by vmole (guest, #111) [Link] The "mere users" made the choice to buy unsupported hardware. Why should I change my principles to support their bad habit? The way to help people addicted to tobacco is not to provide them with free tobacco, but to help them stop smoking.
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 19:05 UTC (Tue) by ikm (subscriber, #493) [Link] I think you don't have to change your principles in order to recognize that these people can be a part of community. What alternative would you suggest for people who need high-end 3d under linux now? To stop having these needs, right? Or to go develop their own drivers? Or maybe even go develop their own hardware? It's ok to have principles, but some grip on reality would be nice as well. Particularly regarding the fact that people with different principles and different needs can still coexist in the same world without making it look like a checkered board.
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 19:40 UTC (Tue) by vmole (guest, #111) [Link] You did suggest that I change my priniciples. You said I shouldn't say bad things about NVidia because they won't provide the information needed to support their hardware properly. (And actually, I don't say bad things about them; I simply don't buy *any* of their products, and, if asked, recommend that others avoid them as well.) And sorry, this is harsh, but if you don't care enough about free software to avoid products that only have proprietary drivers, then you're not part of my community. Yes, that means you can't play Duke Nukem Forever. Such is life. Now, don't get me wrong. If people want to spend their time working on Nouveau, I won't recommend that they be stopped, somehow. But I think it's perfectly reasonable to ask why they are supporting the one common 3D HW companies that is not making any effort at all to support the free software community. And no, I don't count closed source drivers that only work with certain kernels and certain architectures as "support".
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 20:34 UTC (Tue) by ikm (subscriber, #493) [Link] 3d is not only for self-indulging, as you're suggesting. And I'm certainly not a fanatic. I just happen to have work to do. You have suggested nothing in how to accomplish that. Suit yourself.
Reverse engineering: more than NVIDIA deserves? Posted Feb 21, 2008 19:02 UTC (Thu) by amikins (guest, #451) [Link] ... Wait, what? You -ARE- aware there's REAL WORK that requires 3d graphics acceleration to be anything resembling efficient? There's modelling objects or products, either prior to production, or as the actual production in the case of media. Architects, engineers, animators.. All these people can reasonably require extremely high end 3d graphics to get their work done. Then there's simulations. Astronomers, physicists.. They, too, can need 3d modelling in order to get their work done; it isn't all pure number crunching in the background. You have to -show- what's happening, too. Oh, and then there's the game developers, too...
Reverse engineering: more than NVIDIA deserves? Posted Feb 21, 2008 19:52 UTC (Thu) by vmole (guest, #111) [Link] Of course I'm aware of that. But I'm also aware those users, who have money to spend, did not use their buying power to influence one of the 3D companies to provide free drivers. They have taken advantage of the hard work of others to get off the expensive SGI workstations.
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 19:23 UTC (Tue) by mrfredsmoothie (subscriber, #3100) [Link] The only form they CAN? You know for a fact that NVidia is legally constrained from releasing free drivers for their hardware?
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 20:41 UTC (Tue) by ikm (subscriber, #493) [Link] Of course not. They are just unfriendly people who despise linux crowd because they think they are too geeky.
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 7:19 UTC (Tue) by rsidd (subscriber, #2582) [Link] I totally agree, and there's another thing: the number of Linux users is small but growing; but the number of Linux users who care so much about free software that they will not use the NVidia driver must be quite microscopic. (I suppose those are the same users that won't use Adobe's flash player, or Adobe Reader -- which is still the only real option for filling forms. Personally I don't know any such people, though I'm sure they exist.) Vendors who want to pre-install Linux are unlikely to be put off by NVidia's proprietary driver. I don't foresee the Nouveau driver being real competition any time soon, and I don't think abandoning the Nouveau project will punish NVidia in any realistic way. Meanwhile, I agree with the parent -- till some time last year, a Linux user who wanted high-end graphics would have had reason to feel grateful to NVidia, which had both high-end hardware (unlike Intel) and high-quality free-beer drivers (unlike ATI).
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 8:40 UTC (Tue) by michich (subscriber, #17902) [Link] Vendors who want to pre-install Linux are unlikely to be put off by NVidia's proprietary driver.They should be. It's one thing to download a proprietary driver and use it on one's machine, but distributing it together with Linux is another matter. The 'mere aggregation' clause of the GPL does not apply here, IMO.
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 9:36 UTC (Tue) by yodermk (subscriber, #3803) [Link] > I suppose those are the same users that won't use Adobe's flash player, or Adobe Reader Vastly different context than kernel modules/closed hardware. I use Skype, Flash, and (rarely but occasionally Acrobat Reader). Although I would be grateful for adequate Free alternatives, there is nowhere near the urgency of replacing those as there is of replacing closed kernel modules. Closed kernel modules have a good list of practical disadvantages, including being tied to the vendor when you want to upgrade the kernel or x.org. If there's a problem, *only* the vendor can fix it, not the hundreds of awesome kernel hackers. And there are potential legal problems. (Yes, I use the closed nVidia driver on my laptop now, but I'm not happy about it. I'll switch to Noveau when it's ready and I'm certainly buying ATi on my next computer.) > Vendors who want to pre-install Linux are unlikely to be put off by NVidia's proprietary driver. Haven't you read the comments from the Dell Linux guy? Dell now *actively* prefers hardware with open specs, *even for systems they pre-install Windows on*!!!
Reverse engineering: more than NVIDIA deserves? Posted Feb 22, 2008 15:23 UTC (Fri) by wilck (subscriber, #29844) [Link] > Haven't you read the comments from the Dell Linux guy? Dell now *actively* > prefers hardware > with open specs, *even for systems they pre-install Windows on*!!! Please have a look at Dell's offerings. Except for the low-end, all laptops have NVidia boards. Even the "Open Source PC" comes with an NVidia board.
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 20:35 UTC (Tue) by AJWM (guest, #15888) [Link] > the number of Linux users who care so much about free software that they will not use the > NVidia driver must be quite microscopic. (I suppose those are the same users that won't use > Adobe's flash player, or Adobe Reader -- which is still the only real option for filling > forms. Personally I don't know any such people, though I'm sure they exist.) Well, you may not personally know me, but I avoid NVidia hardware so that I don't have to care about their proprietary driver. (An ATI 9250, with a free driver based on specs they released before they went closed, gives me a perfectly adequate 30 fps in FlightGear FlightSim.) I don't use Adobe Reader either, there are fine free PDF readers out there, and I have no need for filling out PDF forms. I'll admit to using Adobe's flash player -- but note a key difference between using Adobe's no-charge downloadable software vs NVidia's you-pay-for-it-by-buying-the-card software. When Gnash works as a Firefox plugin for YouTube videos, I'll switch.
Reverse engineering: more than NVIDIA deserves? Posted Feb 20, 2008 1:37 UTC (Wed) by BeS (subscriber, #43108) [Link] >When Gnash works as a Firefox plugin for YouTube videos, I'll switch. Good news for you: Gnash works as a Firefox plugin for YouTube since version 0.8
Reverse engineering: more than NVIDIA deserves? Posted Feb 20, 2008 7:05 UTC (Wed) by AJWM (guest, #15888) [Link] Cool. I think I last looked at it at 0.7.1. I'll give the latest a try. Thanks.
Gnash and YouTube Posted Feb 20, 2008 22:27 UTC (Wed) by zlynx (subscriber, #2285) [Link] For values of "works" that include non-working, randomly positioned play/stop buttons, 100% CPU usage, bad network buffering. I used to use Gnash but if it's installed when I load my 80+ entry Comics tab, Gnash tries to play *every* flash animation loaded even if it isn't visible on the page or on the tab. At 100% CPU *per animation* load climbs to 20+ before it falls over from memory exhaustion.
Gnash and YouTube Posted Mar 1, 2008 1:19 UTC (Sat) by zenaan (subscriber, #3778) [Link] So did you submit the URL to the Gnash guys, so they could debug the problem?
Reverse engineering: more than NVIDIA deserves? Posted Feb 19, 2008 10:07 UTC (Tue) by debacle (subscriber, #7114) [Link] I probably will never know how good or bad proprietary drivers are, because I refuse to use them. Of course, I'm not thanking NVIDIA for their drivers, just because I don't care about non-free software. If I would like to use non-free software, I wouldn't use Linux in the first place.
Reverse engineering: more than NVIDIA deserves? Posted Feb 20, 2008 13:12 UTC (Wed) by jmayer (subscriber, #595) [Link] <SPECULATION> Some things make me think that eventually binary only drivers will be banned completely from being loaded into the kernel. If that day comes, having the most stable and most performant driver will not help the least and with OSS drivers being available for Intel and AMD cards it might be that kernel developers will just decide that there is sufficient choice to go forward with the ban, whether it hits Nvidia users or not. </SPECULATION>
Reverse engineering: more than NVIDIA deserves? Posted Feb 28, 2008 9:32 UTC (Thu) by forthy (guest, #1525) [Link] Well, one can speculate that, but how can this come true? NVidia then might not work on a kernel.org kernel, but a simple patch would remove this limitation. The legal question about the NVidia kernel module is about the same as the ndiswrapper discussion: Both are designed to load a Windows driver module (or something very similar to such a module) into the Linux kernel. Both wrappers are available under the GPL, and therefore compatible with the kernel license. The modules they load are not designed for the Linux kernel, and therefore not a derived work. One can argue that creating such an interface is a Bad Thing(tm), like a plug-in interface to GCC has been considered as such, but that's it (even though I agree to the argument). And the kernel is "tainted" when a Windows driver runs in its space; but for a license that excludes any warranty whatsoever, the consequence of "tainted" is not that important.
+1 Posted Mar 2, 2008 8:43 UTC (Sun) by gvy (guest, #11981) [Link] Worse, newbie hackers tend to break basic functionality for fancy stuff no one sane really needs to get things done! Intel driver was clearly broken some half a year ago, judging on numerous frustration reports in mailing lists I follow; Driver "ati" (at least radeon part) broke for me with 6.7.x -- no it's not fun to get arbitrary 96dpi on 133dpi display at all. So before moaning and shooting with dirt it's really worth looking without those special glasses, who's actually doing better even in presence of specs for hardware which does require some expertise to drive. And that expertise seems to be slowly disappearing in xorg project with older developers getting silent...
+1 Posted Mar 9, 2008 16:23 UTC (Sun) by nix (subscriber, #2304) [Link] The authors of the ATI and Intel drivers are not `newbie hackers' by any definition, and breaking basic functionality in development branches while restructuring things is entirely expected.
|
|
||||||||||||