Slashdot Mirror
NVIDIA Drivers for 2.6 Kernel
fmileto writes "Kerneltrap.org is reporting that Nvidia has released drivers for the 2.6 series kernel. The driver and install directions can be found on Nvidia's website."
← Back to Stories (view on slashdot.org)
Now if only ATI would do the same.
> For those who've sold their GPL soul
;))
My soul is proprietary, thank-you-very-much. (Under an exclusive license, at that.
These sigs are more interesting tha
Release Highlights
* Support for Linux 2.6 kernels.
* Fixed AGP failures on some VIA motherboards.
* Fixed a problem that prevented X from running on Samsung X10 laptops.
I've been running the 2.6-series kernel since test6 (IIRC), and all the time using the Nvidia GFX drivers for my Ti4600. This story suggests that Nvidia users have had to wait for a new 2.6-compatible release of the drivers, which clearly isn't the case.
Tubal-Cain smokes the white owl.
Someone made an install for the 2.6 kernel that worked great already, so no huge news I guess.
:/ Damn motherboard blew (volt-regulator poped... literally) and its been 3 weeks so far w/ no new one comming in!
I'd download the new ones but my Linux box is broken
BTW, does anyone know how to Overclock a GeForce 5600 in Linux? Haven't found anything yet...
# cd /usr/src/NVIDIA-Linux-x86-1.0-5336-pkg1/usr/src/nv
# make
Unable to determine kernel version.
make: *** [select_makefile] Error 1
#
I don't know it's still have problem with kernel versions
-- There is four mistake in this sentences.
The best part I think is Fixed AGP failures on some VIA motherboards: minion.de has been proposing a good patch for 2.6 kernels for long, while this AGP issue was really boring. I hope it really works better now.
I have been wrestling with getting Framebuffer support to work with the driver that ships with the 2.6.1 kernel, and the XFree86 4.3.0 nv driver. Could never get them to play together. Perhaps this will be the ticket for getting my NV17/GeForce4 card working.
Any hints on what the best kernel framebuffer device/XFree86 configuration is with the new drivers?
is here, courtesy of Andy Mecham, NVIDIA's Linux driver guru.
---
nV News
They won't - there is too much good stuff in there for people to steal. I haven't had trouble with them yet, so if they work who cares.
No, they aren't going to make them open source. If you want open source drivers, write one yourself. What we need now is a good driver for the centrino. And if Intel actually releases it, they should be praised, not critized for not being pure open source.
--- http://davidnehme.blogspot.com
Good question. I often wonder if companies that choose the closed source route are "testing the market". What I am suggesting is that they want to see how little they can give, but still make the sale.
What if this catches on? Then what will we have? Imagine mb chipset manufacturers doing the same thing.
Perhaps nVidea has no intention of providing a open source driver as long as people are happy to pay for a card that gives them SOME way to use it. Not necessarily the preferred way.
This is making my transition to 2.6 on my laptop look much more likely. I was wondering when NVidia would get around to doing such a thing. The fact that it has been relatively prompt seems to suggest that they are still somewhat interested in the Linux market.
The next question is when will they release drivers for Keith Packard's/Freedektop.org's Xserver, because to be honest, I'm very interested in seeing what that can do...
Jedidiah
Craft Beer Programming T-shirts
I tried installing it on a new Debian Woody (stable) installation, but it could not determine the module file name. I tried several options to point it to the kernel source and headers, but it didn't work. Eventually, I installed an older version using http://people.debian.org/~rdonald/index.php.
Does anyone know how to install this new version? I didn't even know it was this new, except for the distinct lack of hints that Google provides so far.
I suffer from attention surplus disorder.
> But when will they make them open source?
Unless there is a huge change in their business model, they will never make them open source.
The code for these drivers contains a lot of information about the underlying design... of both the hardware and the accelleration (sp?) techniques that give this product the edge... exactly the sort of stuff competing companies would love to get their hands on. Remember that a good driver can really have a marked effect on the performance of a graphics card.
Of course, all their competitors have probably already reverse engineered every card on the market, but who wants to make it easy?
I love open source as much as the next guy, but we have to accept the commericial reality that just because something runs on an open source platform, it isn't necessarily open source itsself. Oracle is a good example of this.
Linux on the desktop may change this. When enough people avoid buying Company X's product because their cross-platform support blows goats, they may very well be ready to open the source.
Norman Cook's Ode to Sl
ftp://download.nvidia.com/XFree86/Linux-x86/1.0-53 36/README
/usr/share/doc/NVIDIA_GLX-1.0/).
From the README:
If you do not have a working XF86Config file, there are several ways
to start: there is a sample config file that comes with XFree86,
and there is a sample config file included with the NVIDIA driver
package (it gets installed in
You could also use a program like 'xf86config'; some distributions
provide their own tool for generating an XF86Config file. For more
on XF86Config file syntax, please refer to the man page.
If you already have an XF86Config file working with a different driver
(such as the 'nv' or 'vesa' driver), then all you need to do is find
the relevant Device section and replace the line:
Driver "nv"
(or Driver "vesa")
with
Driver "nvidia"
In the Module section, make sure you have:
Load "glx"
You should also remove the following lines:
Load "dri"
Load "GLcore"
if they exist. There are also numerous options that can be added to
the XF86Config file to fine-tune the NVIDIA XFree86 driver. Please see
Appendix D for a complete list of these options.
The more the 2.6 kernel is accepted by companies the faster it'll get to that "critical point" where distributions will have to start using it not to feel left behind. And for those trying to sell Linux (with services etc) the 2.6 kernel will be an excellent bonus.
I also hope they'll be more stable than the 2.4 + 2.6patch was... I know a fair few people for whom lack of stable videocard support was the factor stopping them upgrade to 2.6.
The last (2.4) drivers that worked for me were 1.0.4496; the recent ones are a disaster, so hopefully they've taken the time to iron out the bugs and this isn't just the latest version of the driver with the (already existing) 2.6 patch bolted on.
"'I pass the test,' she said. 'I will diminish, and go into the West, and remain Galadriel.'"
- JRR Tolkien.
that is more, or less stable and WORKS.
im quite happy with that.
I understand the IP issues involved probably prohibit a source release.
I would just.. let this one go and thank them for at least supporting the linux driver.
Now people can watch porn using cutting edge technology once again!
The nvidia kernel driver has been easy to get working with 2.6 since the 2.5.x days. Problem is, they still haven't fixed the nasty mangled console bug that has been in the last three versions. It is always great to exit X windows and have to reboot before I can go back into X windows or else it will lock up my box. But...even if I exit X windows my console is so mangled I can barely use it any how...this sure has promoted my use of X and only X :)
Unstable Apps: Our Android Apps Don't Suck
For those who've sold their GPL soul to use the binary drivers from NVIDIA (like me) you can get them now.
While I support the GPL and don't particularly like binary-only drivers, I reckon this little phrase has no place in this announcement. To NVidia's credit, they seem to be somewhat serious about supporting Linux in a somewhat timely manner. This sort of allusion won't be a great incentive for other hardware vendors to support Linux at all, they'll just think "whatever we do to be nice to them, those Linux folks will always have something to complain about".
When Linux has 80% marketshare and is a true force to be reckoned with, then perhaps the community will be able to afford sarcasm and get away with it, but in the meantime, there must be other, more constructive ways to entice vendors to embrace open-source.
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
Until now they had to use a third party patch, found here
and still no support for video mirroring output on the s-vid port of my geforce 2 mx400 :(
which means my VOD server for the house still has to run windows... bah!
"why don't you just slip into something more comfortable...like a coma!"
I agree. Everytime somebody on /. mentions Nvidia the very first thing that is replied is when will they be opensource. I've never had a problem with the binaries either.
You had 3D acceleration?
(you can't run the lament screensaver (or glmatrix) without a decent 3D accelerator.. too slooowww..)
Atleast that is what they want people to think...
For the video drivers that might be true. Although I doubt there is really anything really new in there... a lot of manufacturers overestimate their own brilliance.
But for there chipset drivers (e.g. nForce) they are just plain assholes.
Jeroen
Secure messaging: http://quickmsg.vreeken.net/
I know what you mean. I recently put together an SFF Athlon 64 system (Based off of the Biostar iDeq 200P), and I was torn with what to do for a graphics card, because I want to run gentoo as my primary OS. (With windows for the occasional test-compile for work).
I chose an ATI Radeon 9700 Pro because they have better open source drivers (community developed) for they're slightly older stuff, as well as binary support for what's current, but they can't be bothered to release 2.6 AMD64 compiled drivers, and now I'm left wishing I'd bought an Nvidia card just so it would be usable, when I know they have almost no open source support at all..
I want the graphics card companies to realize, when a decently fast graphics card (notice I didn't buy top of the line, so that's a clue to you up and comers) comes out with open source drivers, that's where I'll put my money, and where I'll tell my friends and family to go. You've got to please the geeks, guys, 'cause we influence the purchases of others, as well.
The opensource nforce ethernet driver.. The patched nvnet driver didn't work too well but I've been having lots of success with forcedeth.. Can't wait for it to be pulled into the Linus kernel..
No, they should be critisized for promissing something and not doing it for a year....
Jeroen
Secure messaging: http://quickmsg.vreeken.net/
I made the mistake of getting sucked into the wonderful price point of my nforce2/Athlon system..
The proprietary graphics drivers have been a huge pain.. nvidia support has been terrible when compared to the open source community. For much of 2003 the video would go blank once you exited X11 (forcing a reboot whenever you exited X!).
And now they have gone to a monolithic installer "to make things easier". Right.
The little bit I saved pales in comparison to the time I have spent dealing with this BS.
And how long until nvidia makes my board obsolete?
It's not going to happen. Forget it! nVidia has too much time and "innovation" tied up in the development of the drivers to risk competition benefiting from them being open sourced. I'm sure ATI feels the same way about their drivers too. Hence, ATI also will not release its source code.
Life is not for the lazy.
So you've never ran the 2.6 kernel up until now? Have you never had trouble with the binary drivers because you wait until the binary drivers won't give you any trouble? Sort of self fullfilling don't you think?
GNU/Linux PowerPC users -- such as in Power Macs -- are still out of luck.
Leandro Guimarães Faria Corcete DUTRA
DA, DBA, SysAdmin, Data Modeller
GNU Project, Debian GNU/Lin
Actually in the long run having the source code and/or specs to build a driver/module would be better in that the OS driver spec may change, but all you would need to do is rewrite the driver to work under the new system. (Yes, easier said than done.)
On the other hand binaries are so much easier to deal with though. Install them and they either work, kinda work, or don't.
Maybe those IBM-Linux commercials are right, "the future is open" because with closed source software and companies not supporting every single platform for ever the closed-source software is destand to become unsupported and eventually become stuck in the past. Does Windows XP support all old hardware that may still work in your computer?
Then again maybe I just think too much.
Losing faith in humanity one person at a time.
they CAN'T, because of S3TC support. have to have a license for it, and that implies that it can't be open-sourced.
---
nV News
here it is:
#include "stdio.h"
#define s 1
#define o 1
#define u 1
#define r 1
#define c 1
#define e printf("You won't get any source. Go away!\n -- signed: NVidia\n")
main()
{
s - o - u - r - c - e - s;
}
Enjoy...
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
My next video card purchase will be from a company that makes GPL-compatible drivers. Suggestions anyone?
We're basically screwed. Nvidia's success with binary-only drivers has made ATI do the same thing. Unless the Weather Channel decides to fund development of drivers for the newer cards, Radeon 8500 is the end of the open-source line. And no one else really exists in the high-performance 3D graphics card market.
I hope Red Hat uses some of their new half a billion to do something about this, but they've probably got other goals to hit first.
You had 3D acceleration?
Fer sure, and its definitely been working (smooth 3D viewing of 100,000 polygon meshes). Either that, or the Underpant Gnomes installed a 10GHz processor in my machine while I was asleep...
However, its worth noting that I'm using the drivers as packaged by Gentoo. This may include the patch which another reply to my OP mentions; I wasn't aware of this patch when I posted.
Tubal-Cain smokes the white owl.
That was seriously the only thing holding me back from going all 2.6...hmmm...off to compile!
I wondered this myself, as I'm using kernel 2.6 since test2, also with nvidia support (incl 3d. nvidia logo is showing on X startup). Perhaps it were the gentoo devs who made it possible, as all I did after starting with the new kernel was "emerge nvidia-kernel nvidia-glx" and it worked.
Wonder if Mandrake will pull another microsoft on this one? Tried installing drivers for a dlink nic on 9.2 and got a message that the binary only drivers would "pollute" the operating system or some such. Refused to install them at all - there was no do it anyways option. Anyone else suffer someone elses self righteousness like this lately?
Yes. There's been an unofficial patch available to compile(*) the nvidia drivers against 2.6 kernels since before the 2.6.0_test series.
* Yes, most of the nVidia driver is binary-only, but there is a small module that has to be compiled for your kernel. That is the part that had to be patched.
0 1 - just my two bits
So you've never ran the 2.6 kernel up until now?
See my post below. I've been running 2.6 since it was released, and have never had a problem with the Nvidia binary drivers. YMMV, however.
Tubal-Cain smokes the white owl.
Ok I'll bite,
Maybe he wants to play the linux versions of Unreal Tournament 2003, or Medal of Honor, or Neverwinter Nights, or Quake3, or Return to Castle Wolfenstein, or Enemy Territory, or Savage , or Rune, or SpaceTripper, or Tribes 2 or Serious Sam, or Postal2, or America's Army or another of the various other games on Linux that requires 3D acceleration. Sure, a far cry from what's available on windows, but more games then I'll ever have time to play.
"If they have both, tell them we use Linux. And if they have that, tell them the computers are down." -Dave Chapelle
Whenever you start talking about hardware drivers, especially bleeding-edge hardware in a highly competitive market like video cards, you run into the following problems:
1) There is a natural desire to keep technical details (both in the hardware and in the driver implimentation) secret from one's competitors, so as to build a competitive advantage.
2) You may not own all the technology in the hardware or the drivers, and your licencing agreement with the 3rd-party technology providers may include terms of non-disclosure.
This tends to disincline one from open-sourcing the drivers.
The advantages of having them opened up everybody here is well aware of. But realizing those advantages takes time to sink in at the hardware company - especially when their Linux market is very small (so the perceived risks outweigh the rewards)
As time goes on, and especially as the Linux market grows (to the point where it is providing a signifigant fraction of a company's revenue) I believe the value of opening up the drivers will become more compelling to the driver authors (and more importantly, their management)
Baby steps. Rome wasn't built in a day.
In the meantime, there is value in supporting companies who provide closed-source drivers for products where there is no other alternative. Help them build the Linux experience - both on the technical and social aspects - that will eventually lead them along the path to opening their drivers up.
Would I prefer to see fully open-and-GPLed NVIDIA drivers? You bet your ass. But for whatever reason, they aren't ready to jump off that particular cliff, so I'll support them anyway in anticipation of the day when they DO open the drivers up.
We're fighting 20 years of a culture of secrecy and code-hording here. It'll take time to work through that.
DG
Want to learn about race cars? Read my Book
Yes, build your own card. Because their aren't any. The open source drivers in linux weren't made by the card manufacturers, that is why they will never perform as well as the closed source ones. So if you want an open source driver that performs, you will have to create your own. If and when Linux ever gets the market share to create a demand, I think you will find distributions selling a separate driver CD's, and / or downloads on the manufacturers site that keeps pace with the windows drivers. But I highly doubt you will ever see open source drivers by any hardware manufacturer that faces stiff competition from others. The only periperial card that I have ever bought that had GPL'd drivers by the manufacturer was a multitech modem that contained 4 modems on the board. This was highly specialized for dialin networking and linux has the market share there to make it possible. And modems aren't exactly cutting edge stuff.
Sigh. Like nobody thought of that before.
The problem is the documentation isn't there, and aparently reverse-engineering it is too difficult/dangerous. And yes, there are good reasons for wanting an open-source driver. For example, the current driver is known to do a double memory free on some occasions. Could have been fixed easily if the source was available. And it would have been far better if the NVidia driver could have been developed with all the other drivers during the 2.5 development cycle, because any bugs could have been detected much earlier.
There are very good reasons for wanting open-source drivers, even if you're not an OSS fundamentalist.
Why? Just so you can say you have them? I mean there are people who would go an modify the sources and it would be good.
But I get the feeling NVidia could release any damn source and say it was the video drivers and 90% of people who complain about the drivers not being open-source would be happy.
It could just have a makefile, and spend a lot of time configuring. Then it compiles some huge-ass program, and that program does nothing. Then it installs its own propetory driveres. Problem solved.
This "oh, we love you NVIDIA, we'll buy your products even though you don't support Free Software at all" is counter-constructive to our ends.
But you don't understand : there's no question we don't like their binary-only releases, but they did make a step in our direction. To them, they think readable code gives away their hardware secrets. No need to be nasty to them (and given the level of OSS support for hi-perf video boards out there, we probably ought not to), I'm just saying it's more constructive to try to work *with* them rather than go against them.
They already show an interest for the Linux platform. We just need to lobby them and work out how they could understand that it's possible to make proprietary hardware full of secrets with open-source drivers. If we antagonize them, that won't do any good at all.
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
There are patches out there at minions.de that allowed use of the nVidia driver with a 2.6 kernel but the drivers in there default state from nvidia.com did not work until now.
BTW..... the gentoo sources did contain those patches:)
Yeah, I used page to guide me through making nvidia drivers work on a toshiba laptop running debian with a NVIDIA(R) GeForce(TM) FX Go5200 and a 2.6 kernel. Worked right away too.
Yes, well, perhaps the title of the story should've been Official NVIDIA Drivers for 2.6 kernel since what we've been using for months has been an unsupported hack. For most of people, including me, it has been working just fine, though, as you suggested, but before this there has been no-one to blame if didn't work. Now it's supported by NVIDIA.
You can go make a comment, why be lazy and assume someone else will do your bidding?
If you want something, you gotta go after it.
But does the IP belong to you, or to God? (or Allah or whatever. I assume that since you believe to have a soul, you also believe in some form of God) I think you got a time-limited lease on it. A run-time licence perhaps? And I don't want to get into the physics of it, being a derivative work of your parents and all...
Kjella
Live today, because you never know what tomorrow brings
I just tried the new drivers.. They failed to work for me.
This is the same problem I had earlier this week when trying to migrate from 2.4.24 to 2.6.1 with the minion nvidia patch (4496). Google searches showed that LOTS of people were having this same problem.
I hope this can be resolved with a simple kernel tweak. Any suggestions?
Jan 29 11:12:43 tool kernel: nvidia: module license 'NVIDIA' taints kernel.
Jan 29 11:13:48 tool last message repeated 2 times
Jan 29 11:13:48 tool kernel: 0: nvidia: loading NVIDIA Linux x86 NVIDIA Kernel Module 1.0-5336 Wed Jan 14 18:29:26 PST 2004
Jan 29 11:13:50 tool kernel: space, but we're in an interrupt or holding a spinlock
Jan 29 11:13:50 tool kernel: 0: nvidia: trying to map 0xd40ff000 to kernel space, but we're in an interrupt or holding a spinlock
Jan 29 11:13:50 tool last message repeated 156 times
Jan 29 11:13:51 tool kernel:
Jan 29 11:13:51 tool kernel:
Jan 29 11:13:51 tool kernel: Badness in pci_find_subsys at drivers/pci/search.c:132
Jan 29 11:13:51 tool kernel: Call Trace:
Jan 29 11:13:51 tool kernel: [] pci_find_subsys+0xe9/0x100
I'm about to assemble a new machine with an Athlon 64 CPU. I know that I could use the 32-bit drivers and have them work, but I'd rather go with the ones designed specifically for my CPU -- unless someone has compared them and found them either identical or nearly so.
STOP MISUSING APOSTROPHES, YOU MORONS!!!
But I get the feeling NVidia could release any damn source and say it was the video drivers and 90% of people who complain about the drivers not being open-source would be happy.
Well not quite. The problem is, without the sources, when the kernel API (or something else) changes, it won't compile anymore. Similarly, try to run the old Qseeme binary on modern Linux distros : you'll have the darndest time getting it to go (if you get it to go at all) because it's linked against libc5. If the sources for Qseeme had been released, someone would have adapted it to glibc for example.
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
Maybe he should have done 5 minutes worth of research with google before investing $400 bucks or so in a video card.
Or maybe his research was limited to the fanboy rantings of how superior the card is based on artificial benchmarks.
Why are the super tech geniouses of linux land always shocked to find the expensive hardware they just bought has no/little linux support?
Anyhow, for ATi to supply good linux support, that would mean true OpenGL support - something even the windows drivers for the Radeon series are pathetically lacking.
The cards are the 'fastest and greatest' only in Direct3D benchmark apps.
I don't need no instructions to know how to rock!!!!
Don't forget patents. NVidia can't release code for things like S3TC and other patented algorithms, but several games/apps require these to work properly.
:(
Sure, one could blame the game developers for requiring patented code, but when you get to the people who want to just use their computers (like me), it all comes down to "does the damn thing work or not?"
On a completely different note, it's a bit sad how (some) kernel developers want to purposefully hard for binary only drivers. Even if drivers have source available, that's *still* a pain for users. What happens when the API changes? (Like it manages to do even in stable series.) Sure, the vendor GPL'd the driver, the driver is probably in some future kernel version the user doesn't have, and the driver on the disk/CD that came with the hardware doesn't work with the user's kernel. Upgrading kernels can be a nightmare (especially when you end up with an update that breaks something which previously worked; also a bit too common).
API and ABI compatibility has many more benefits than just allowing proprietary drivers. Too many developers don't seem to see them tho.
Your example does not compile.
In today's world of video drivers for Linux, I'll take a binary driver with superb documentation over open-source with little or none.
I'd prefer open source with good docs, of course. I'd also like a pony.
"Skill shows through where genius wears thin." -Wittgenstein || Religion: uniting aviation and architecture.
I would say a very long time. Until last year, I was using a TNT with the unified NVidia drivers, and I only switched because I managed to pick up a TNT2 for $10 or $20 (the TNT 1 was Asus's excellent model, but it wasn't good at doing 1600 x 1200).
If they're supporting a card that old, I don't think you're going to have to worry anytime soon. I'm impressed that they're supporting their entire post-Riva lineup with a single driver.
Personally, the Free drivers never worked for me; X just looked mangled. I installed NVidia's binary driver by following the directions (gasp!) and everything Just Worked, and has continued to Just Work. Recently, I used the patches from minion.de to upgrade to Linux 2.6, and yet again it's working fine. I've never heard of the console corruption or X locking problems until reading the comments on this article.
The monolithic installer certainly does make things easier. You don't have to separately install the GL libraries and the kernel module anymore. It's just one file for whatever post-Riva card you have.
I've been able to run OpenGL demos, and the power in my area is less stable than X11 (out when I left for work this morning).
Am I a big proponent of Free as in libre? Yes, certainly. But NVidia released good software which works with the hardware I had already. Same reason I still use my MP3 discman; I don't know of a company that has an Ogg-compatible CD-ROM player (my music collection is too big to constantly fiddle with hardware players, sorry). Broadcast radio in Baltimore sucks and I need music. I prefer the Free alternative when it's practical, but sometimes it's not. You definitely find that out working for a physical plant (specialized software needs).
WMBC freeform/independent online radio.
Agreed. I'm likely to be buying a new system in a few months; are there any options with OSS drivers and at least respectable 3D acceleration?
The ATI DRI drivers don't support S3TC for this reason. In many ways this does cause a problem with some applications that rely on S3TC (like UT2003), but for many purposes S3TC isn't needed and the DRI (open) drivers work very well without it.
In my opinion, ATI have done a pretty good job with supplying documentation for their older cores. Pretty much everything up to R280 is supported under Linux with open-source drivers. That includes 3D. Most other features like dual-head and tv-out work as well. Video capture works too, with the GATOS variant of the DRI drivers.
Red Hat's CEO seems to think that Linux isn't for the desktop, so I doubt they'll be making a big investment in anything related to the higher end of consumer video cards.
WMBC freeform/independent online radio.
Find a country where the law (1) forbids manufacturers to keep secrets from customers and (2) specifically permits the use of reasonable force in certain situations; or (3) forbids the release of closed-source software. {This may require the purchase of a small uninhabited island, but you probably can pick up a used one on ebay. Alternatively, there was talk about some Central / South American countries outlawing closed-source software: see if you can find one that actually has}. Go there and take a graphics card with you. Request driver source code from manufacturer. Point out that local law says you have a right to that information. If driver source code is not forthcoming, disassemble binary driver under reasonable force provisions of local law. Publish source code as discovered by you on secure web server located in that country and with all logs diverted to /dev/null. Get general feeling that this is like DeCSS all over again. Wait for authorities in USA to determine that your liberated source code is not illegal. Sue prosecutor.
Je fume. Tu fumes. Nous fûmes!
My Radeon 9800 Pro works rather nicely on my gentoo box, although not quite as fast as in windows, it still gets the job done at a satisfactory level (enough to play UT2k3 at a more than reasonable framerate).
Oh I dunno about assholes... As a proud owner of a NForce2 board, yeah I'm frustrated as hell with their crappy sound drivers (well, tecnically I use the ALSA drivers, but they're based on the nvidia-released ones and still aren't any good). But I mean, they *did* give us the nforce agpgart stuff for free. I guess they were bitches about the nforce ethernet driver (which was trivially reverse-engineered), so feel free to be angry about that. However, on the sound front, from what I've heard SoundStorm is at least as full of proprietary badness as their video drivers, probably Dolby IP among others. And the unofficial word is that they *are* working on a binary driver, which is probably the best they can do.
That doesn't provide framebuffer support. That provides NVidia driver support for XFree86. The 2.6 kernel doesn't seem to have a fully functional framebuffer driver for nvidia cards (I couldn't get it to work either, but it works for me in 2.4). I don't know how to get XFree86 to try to use the framebuffer driver from the kernel, but I've heard there's no performance gain over using NVidia's drivers directly.
Developers: We can use your help.
A Tale in the Desert http://www.atitd.com works well on it :-)
Tried installing drivers for a dlink nic on 9.2 and got a message that the binary only drivers would "pollute" the operating system or some such. Refused to install them at all - there was no do it anyways option.
/sbin/lsmod command as a user to see for yourself.
First RPM doesn't give such error messages, nor does RPM prevent you from installing the RPM you desire (Remember: Unix assumes that root knows everything, and never limits root from doing anything either)
Secondly, you are possibly confusing the above error with something not done by mandrake, but the kernel. The kernel automatically complains about non GPL/BSD modules being loaded, however this DOES NOT prevent the modules from being loaded. Issue the
Lastly, as another poster already pointed. Mandrake sells a commercial version of their distro that automatically uses such binary only drivers. (Their 100% FLOSS distro does not ship with them but like *any* other distro, can use them.)
Sunny Dubey
This is good news. Now we just need to get lirc support and ivtv support without nasty hacks, and I'll be good to go.
Waiting for this driver was one reason I haven't yet upgraded. The patch didn't work for me either. But the other big problem I've had is getting apps to use the new sound API. Almost everything seems to still be written with the older API, and for some reason I can't get the kernel option for the new driver to expose the old compatible API to work. Maybe I just need to try the latest kernel version...
Developers: We can use your help.
The fastest consumer graphics card available WITH OSS drivers (not binary-only) is the ATI Radeon 9100. You can also get the FireGL 8800 which is a little bit faster...
Allah means "the God".
Assumption is the mother of all fuck ups, or in this case the grandmother on the fathers side of ignoring religions of millions of people in Asia.
${YEAR+1} is going to be the year of Linux on the desktop!
Linux has a coherent design philosophy, which you must understand before attempting to change it. One cannot randomly load modules built vs. different kernels, and you can't load proprietary modules w/o tainting the kernel, and it tells you as such. You sound like the type of complains about those RIAA "whiners" as you download gigs of MP3s.
I want to delete my account but Slashdot doesn't allow it.
Nvidia drivers make the center of my screen fuzzy, (3d is fab though) I'm wondering if they are worth it, if I have to move all of my small text to the periphery of the screen to be able to read it.
The days of the digital watch are numbered.
and what hardware is that, sunshine?
decent (like 'rtcw @ 1024x768+, 4x AA, 8x Anisotropic') only comes with binary drivers
either from ATI or nVidia.
cheers.
``If a program can't rewrite its own code, what good is it?'' - Mel
NVIDIA never wrote the BeOS drivers anyway. They released some, although I beleive not all, specifications to Be, Inc. engineers who wrote the first driver, then some dedicated ex-Be engineers updated them using details from the XFree86 driver.
i ve r/
http://www.bebits.com/app/1622
There is a driver based off of the Matrox open source driver for in the OpenBeOS CVS that supports more cards. I haven't used this myself, but you might have more luck with your specific chipset.
http://web.inter.nl.net/users/be-hold/BeOS/NVdr
Good luck!
Andrew McCall
> You need to free up some time brother because I finished all those games before breakfast. Sheesh.
Yes, and he's busy writing free software or working or getting an education. What a shit, huh. Working... psh. (sarcasm)
And he probably doesn't maintain a Linux box just for games. He maintains for working and sometimes feels like fragging something. If you want to get real computer work (typesetting, modeling, programming, etc.) done, you use *nix. (Yeah yeah I know, Windows is great. Whatever.)
My other car is first.
I know you were just trying to be funny, but for those who think you may be serious...
No you don't need to get the 2.6 kernel to run these drivers. They still work with your old 2.4 kernel as well!
I've never had a problem with the binaries either.
I have.
Perhaps I'm a corner case use with 1920x1200 DVI and using OpenGL, but earlier drivers would crash X on me.
Recent nVidia binary drivers, like 4496, have not crashed in months of use.
"Provided by the management for your protection."
I don't have this problem. I can exit X gracefully and still do anything that I want, including restarting X. Occasionally I have a problem where the console doesn't show anything that I type (I get the bash prompt, any stdout output, but nothing that I type -- including carriage returns -- show up on screen), but this is rare, may have a different cause, does not carry over to other virtual terminals and goes away if I merely log out of the session. Perhaps I'm not enabling a certain feature that leads to this problem?
STOP MISUSING APOSTROPHES, YOU MORONS!!!
The Linux drivers NVIDIA released are actually newer than the Windows 2000/XP ones! I call that pretty darn good support!
I'm all for Open Source, but there are probably far too few 3D/OpenGL engineers who have the time to work on and release quality Open Source 3D/OpenGL drivers. NVIDIA has practically their whole driver engineering team working for us. I consider the closed part just an extended piece of 'firmware' for the (closed source) video hardware. The 'loader' and glue code are open source.
It would take a couple of man years to produce quality drivers that even come close to what we have now, and by that time the current crop of 3D hardware cards will be thrice obsoleted (hi Matrox!)
Better to spend our resources improving other things (like GNOME, D-BUS, whatnot) than to duplicate driver magic, just for the sake of being open source.
Now, if you're a PowerPC user, I take everything I said back *grin*
-adnans
"In short: just say NO TO DRUGS, and maybe you won't end up like the Hurd people." --Linus Torvalds
Good to see that nvidia is still working hard for my money. I'm glad to keep paying them.
From my experience with different video cards and 3D FPS in linux, nvidia is head and shoulders above the rest, and they have been very responsive to the community.
Is nvidia perfect? well, no, they could do a better job of playing nice with the vendors package management schemes, but their cards and drivers are rock solid and give the best bang for the buck.
2.6 kernel includes nptl - redhat's nptl is merelay a backport of what's in the 2.6 kernel
The NVIDIA drivers worked fine in 2.6 before with a simple patch (changing the location of an include or two).
This barely even requied work from nvidia... just an ifdef or two.
I was using nvidia's drivers on 2.6 weeks ago.
This may be true, but I would bet on market forces here. If company X decides it can get away with binary modules, and that module has serious issues, who's gonna by the hardware? With a video card, at least there's the fallback option, but with a northbridge driver, or with an ATA driver, you won't be able to boot your PC.
Some products, such as video, possibly wireless, or whatnot, can probably get away with binary only drivers and survive in the Linux space. Other hardware will just fail. Until Linux on the desktop is a little more popular, it's not going to be an issue though. I do expect some companies to do whatever they can to scam the community. I don't think this is what ATI and Nvidia are doing.
Bah
You use riva frame buffer right?
This has been around for a loong time, before the last 3 releases. It's not directly NVidia's bug, it's related to the riva frame buffer conflicting with the NVidia drivers.
Turn off frame buffer support and use a plain console and you will be fine.
Is this a big issue? I've never encountered this problem with any of my machines so far. I can exit X and go back in, etc. No problem.
A person with a decent disassembler and knowledge of reverse engineering can get to your secrets even if they are binary only, so that argument is not valid. IIRC, Nvidia stated that their driver include some third party code that they cannot open source.
I've been using 2.6 with and Nvidia card and Nvidia GLX drivers for a while now. (since 2.6test10)
I haven't been able to get framebuffer working though. X and OpenGL under X worked like a charm all along. Now someone else posted that there was a 3rd party patch, so I'm assuming that's how Gentoo gets it to work, but fact remains Nvidia cards have worked with 2.6 for a while now.
Though if this somehow fixes framebuffer (somehow I don't think that's what's wrong) I'll be a wee bit happier. In reality though, I could care less whether framebuffer worked as long as X works ok.
Oh. Wow. A post below just let me to this link:
D 10 7171185522684&w=3D2
http://marc.theaimsgroup.com/?l=3Ddri-devel&m=3
Apparently, you *can* patch the opensource DRI Radeon drivers to support S3TC after all.
Now that NVIDIA display drivers are there for 2.6, there's only one thing holding me back. I've got a Nforce2 motherboard. Somewhere around 2.4.23 (or was it 2.4.22?) support was added to the kernel for my onboard 3com card.
I know that the initial 2.6 didn't have this support, and I want to know if it's been added yet or not. I know I can go to kernel.org and see the changelog for the latest version of 2.6, but where can I get a changelog of ALL the changes since 2.6 started?
Cogito ergo sum in Slashdot.
Here's a question for you: do you need to use the nvidia driver? If so, why?
I'm not sure. This will be my first non-headless (headful?!) Linux box. What alternatives do I have, and what are the tradeoffs?
"People who do stupid things with hazardous materials often die." -- Jim Davidson on alt.folklore.urban
Should have also mentioned that these were Video drivers as the most important drivers for me are nforce.
Use the "nv" driver that comes with X (NVidia's driver is called "nvidia"). You might as well, since you won't have to do anything special to get it going (except writing "nv" in your XF86Config file). If it doeesn't work for you (I think it may not do the extremely high resolutions) then install NVidia's drivers, and pray that they're bug free (they weren't when I tried).
First off, this is a display driver that was just released, not the nForce drivers.
Secondly, I moved to 2.6.1 a few weeks ago and my hard drive speeds are fine:
Here's my hdparm settings. Perhaps yours got messed up somehow?
I'm not sure what they all mean, but I'll past them all anyway. I think the first four are the ones you want to take a look at:
I remember when legal used to mean lawful, now it means some kind of loophole. - Leo Kessler
Drats, I just got a used 9100 because I thought it was the fastest card with Free drivers (used because I'm not giving ATi any more of my money until they start to release Free DRI drivers). I have my principles and I'm standing up for them. I was one of those people who wouldn't buy a portable music player until it supported Vorbis and had a Free sync client. This is why I have a Neuros (I'd have tricked my brother into buying it from me and then gotten a Karma if he had a Free sync client, but sadly it lacks one). I haven't used proprietary software in about three years now.
I have this fear that I'm going to be stuck with the Radeon 9100 / FireGL 8800 as the fastest graphics card I can get for many years to come. At least the pro-audio people woke up and release specs (I can use a 96 channel RME Hammerfall just as well with ALSA as I can with Windows or OS X).
HAL 7000, fewer features than the HAL 9000, but just as homicidal!
I'm reminded of the following FSF essay when I read your popularity-based definition of "support":
No they won't, they'll recognize that we're willing to stand up for ourselves and not capitulate to someone offering the latest temptation away from freedom. You don't bargain with someone by giving them everything they want on their terms. Since when is it our job to welcome any proprietor that comes down the block? When nVidia is willing to deal with me in terms of software freedom, I'll be happy to recommend their hardware. I'm sure lots of other free software users would too. We can have popularity too, but it's better to have popularity on the basis of software freedom.
Freedom-talk won't dissuade nVidia or any other proprietor. Software proprietors want your money and they want to control how you can use your computer. As long as you are willing to surrender these things you will be quite popular with them.
This is the real problem--you have hit the nail on the head here: you are chasing mere popularity. The GNU system was started in the pursuit of freedom. When you placate proprietors you might become more popular but you will never get freedom.
I could not care less if nVidia "embrace[s] open-source" because I want software freedom, something the open source movement doesn't value. Reading the next few paragraphs of that same FSF essay is instructive.
Digital Citizen
Your comment is hardly insightful. There was plenty of free software available (including support for high-end video cards) before nVidia distributed their non-free software. There will be plenty if they leave. We owe nVidia nothing and we should reject their offer to push us into dependency.
Digital Citizen
I have nvidia cards and their 3D accelerated drivers in every linux workstation I use.
I can't remember any of them ever crashing with the nvidia cards. I have seen kernel messages from time to time about sleeping in illegal context etc, but no crash. (On the other hand, I see severe instability with the ATI cards I've tried, such that I just turn off DRI if using an ATI card, so that the machine will reliably stay up)
My main desktop at work has a riva tnt2, and it's mainly mozilla and vnc all day every day, with 3D screensavers when I'm away, while my home workstation has a geforce 2, and that's where I do a lot of q3a, ut2003, movies & music, and of course 3D screensavers when I'm away.
Uptimes on both boxes are basically as long as I feel like waiting before upgrading the kernel, usually 60-90 days at a stretch, but bottom line is they reboot when I reboot them, period.
One thing I tend to do is use intel chipsets and CPUs, that might be a factor. Also, I use in-kernel agp, rather than nvidia agp. The nvidia readme mentions several known issues with certain hardware or BIOS settings, be sure and double check those.
First: to fix it you have to download the correct kernel source *from Mandrake*. I will have to locate the URL for this, I found it in a web search looking for help in getting the driver to work. Try googling for "Nvidea mandrake 9.2 kernel source"
/usr/src/linux location (did not compile the kernel at all) and this time the nvidea driver compiled and installed flawlessly.
The message is not from Mandrake, but from the kernel. And I think it is misleading.
I have Mandrake 9.2 and I ran into the same problem. The 9.2 install did not come with the source or header files for the Kernel.
If you download the Nvidea compile & install shell script, it complains about the missing ones. I then downloaded what appeared to be the correct version from kernel.org and put the header files in. The Nvidea driver then compiled but the installer complained that it failed. It said to look in some log file, that file contained module load complaining about a lot of missing symbols, and then adding the misleading (and IMHO somewhat nasty) message about this not being a GPL component. I think it always adds this when there are missing symbols.
I wasted a good deal of time actually compiling the kernel source I downloaded. I discovered the the Nvidea driver works perfectly with that, but I lost a lot of Mandrake stuff, in particular Superdrive, and my network interface refused to work.
Then further searches on the web revealed lots of people who figured this out and said where to download the correct kernel source. I got that, and simply stuck the header files in the correct
I'm sorry, I don't understand your meaning. nvidia makes great video cards, and provides well written, up-to-date linux drivers.
Exactly what is your point?
If your doing 3D visulisation of data, that can be a lot of polygons.
I have tools that are available for both windows, and Linux.
If I load a protein molecule into them, Windows burps a little, and then smoothes out.
Linux doens't burp at all, and give a consistant 0.001 fps, if using software renderes.
Hell, I've seen some people consider a GeForce FX 5700 just to get something that will allow them to rotate big [0] molecules in nearly real time.
Who said it was games? Some people use this stuff for work. When you pay through the nose for enough CPU to work out the gemoetries in near real time, it bites a lot of ass that your display sub system can't keep up.
[0] Ok, obscenly big - something like 30 000 amino acids, so 350 000 atoms, each being a sphere. A lot of vertices.
I am going to download the drivers anyways.
Why? I'm sure they keep count of the linux driver downloads. If they see a boost in the numbers, maybe we will get quicker updates.
I have an ATI card, but that won't stop me from buying an NVIDIA card in the future if they provide some serious 3d support ala Windows.
Does this sound dumb? I know it's fudging the numbers, but with computers and software there is always that chicken and egg problem we have seen so many times before.
Discuss.
--
This is slightly off-topic, but could someone please enlighten me as to what's wrong with so-called "karma whoring"? I'm sure I must be missing something here, but as far as I can figure out, a Karma Whore is someone who posts helpful, informative comments. Why should they be modded down for that?
So does ATI but ATI apparently works with the free software community. Unlike nVidia, ATI chose to put themselves in a position where they could provide real support to the free software community; you can find drivers for some very impressive ATI cards in XFree86. Other manufacturers work with the free software community too by shipping specs or software we can use. I'd much rather buy from them and recommend that others buy from them.
Considering that I don't think they're actually helping me at all right now, I won't miss them if they take their non-free software away. If they delivered free software drivers, this whole conversation would change. I'd not only thank them for supporting our community but I'd buy nVidia hardware and I would recommend nVidia hardware to my friends and clients. And then if nVidia stopped supporting the hardware I wouldn't be left with software I couldn't inspect or improve. I'd be left with software I could improve either by doing the work myself or hiring someone to do it for me. If my particular nVidia card was popular, there's a good chance someone else will improve the software and I could run their improved version if it was still free software.
Digital Citizen
Do you mean the binary drivers? Some people, such as Linus have issues with using binary-only drivers with the kernel as this 'taints' the GPL kernel. Nvidia is not releasing the source b/c this would make it extremely easy to cheat in games, ie, turn on wireframe mode, etc.
I know, I have it running on the 2.4 kernel right now.
2.6 needs new modutils. The package is called module-init-tools and it is in testing. Maybe someone has backported the package for woody. Google for that.
Because they're usually posting information that is available if you RTFA.
"Windows Me offers tremendous reliability and stability improvements..." -- Paul Thurott
This is what some people seem to ignore when they say "who cares, if it works nobody needs the source code".
There are more operating systems out there than Linux, and these OSes (as well as Linux) run on many more platforms than x86.
Closed source drivers = if you want an nVIDIA card and 3D support, you're at the mercy of nVIDIA.
PPC/ARM/MIPS/blah Linux? Any other OS? Forget about it.
Help savingAmigaOS and a free PowerPC market
You need to emerge either development-sources or gentoo-dev-sources (same as development, but with patches that will eventually be in the gentoo-sources ebuild)
/dev/pty support turned on manually, or your consoles won't work.
The 2.6 kernel needs
Other than that it's pretty straight forward if you've ever done a manual compile before. I don't know if genkernel works with 2.6 yet or not, I haven't tried it. (and have no plans to)
Either that, or the Underpant Gnomes installed a 10GHz processor in my machine while I was asleep...
/proc/cpu
do a:
#cat
I'm sure you'll find those underpant gnomes are sly devils.
If you're like the current pool of moderators and some of the posters on this thread, perhaps you'll give nVidia your money in the hopes that they'll do the right thing with it. Why not? It's just a small hop from giving your money to become dependent on them. Treating corporations like charities is a sure-fire way to make you their friend.
In other words, if the preeminent license of the free software community were rewritten to flush all our software freedom down the toilet, we would gain...what? What could possibly be so compelling that we should be willing to trash 20 years of giving all computer users software freedom?
Digital Citizen
are you trying to use the rivafb framebuffer driver? if so, don't. it doesn't play nice with nvidia's X drivers. use vesafb instead.
Xfce: Lighter than some, heavier than others. Just right.
and the source still hasnt been leaked from nvidia??? we had the source to halflife 2 months before it was finished, and thats a full game, this is small enought to be emailed via hotmail! geez, where are the blackhats when you need them, someone on a 2800 baud modem copuld download the source in less time that it would take them to read this article!
Noone writes jokes in base 13!
Really?
:o (didn't want to step on your point m8 ;) )
Because ehm...
I kinda finished most of 'em (+ a bunchload of Win-only games)
I was actually wondering:
Games are a tad slower on lin than their win-variant. Is that because of drivers, lack of focus for Linux-Desktop in the past or the whole kernel - xfree - desktop - layering thingie?
With the 2.4 kernel I couldn't get the NVIDIA driver to work if the kernel was himem enabled - will this work with the 2.6 kernel?
On the third hand, a lot of people dive directly into the comments and really do find short summaries like that very useful. And people don't skip the article just because they're lazy either; you might want to go directly to the comments on this story to see comments from people actually trying the drivers, for example, rather than read a pre-chewed press release that will not really tell you how well it actually works.
So no, I don't think most karma-whoring is a bad thing either.
Trust the Computer. The Computer is your friend.
Not sure about the drivers or lack of focus, but it IS NOT the "layering thingie". A common misconception is that the layers slow things down. Unless your running X over a network there is no lag . Things don't become slow just because its modular and layered (unless your a bad programmer).
Regards,
Steve
Which means the mods aren't RTFA-ing if "whoring" "works".
... a patch which is automatically applied when you're using Gentoo, and therefore causes many users to not realise they're even running a patch. :-)
Karma: It's all a bunch of tree-huggin' hippy crap!
You're new to this whole technology thing I see...
The OpenGL screeensavers that ship with linux use them -
The various multimedia players use them -
The OpenGL games use them, for instance ut2003 and the like -
Actually, every app that uses X11 ends up using the nvidia drivers once they are installed, though not all apps utilize the hardware-accelerated OpenGL.
I was really upset when I first tried to get ethernet working on my nForce 220 board. It used to run fine for a bit then the whole system would hang for 3 to 10 seconds especially if the network was active. I said I'd never buy another nVidia product if I could help it. I put in some realtek card and was on the net in 10 minutes with no problems since.
That said I appreciate the fact that nVidia is supporting Linux. If they were not then I'd really have to buy different hardware. The only thing I'd really like is to not have to recompile the module every time I recomiple the kernel (which I do quite frequently of late as I'm working on in-kernel code). If I could link the nVidia module into bzImage I'd be really happy.
PS: 1) yes "experimental" module versioning is on 2) yes maybe I'm just missing something.
Yes, portage automagically applies that patch if needed, I've been using it too. I've applied it myself with other distros, but with portage I've watched it automagically work. :)
I touch computers in naughty places
The problem with MythTV and Freevo is they require you to install LAME (which encodes the audio stream to MP3). There are patent issues with MP3 encoding.
A MythTV developer recently talked about either making the LAME dependency optional by either using Vorbis, or simply not compressing audio. The latter option is quite feasible - sound is much less of a disk hog than video - only about 500 MB/hour.