Nvidia Removed Linux Driver Feature Due to Windows
tomshardware.com
tomshardware.com
Keeping support for four monitors in the linux code probably meant adding a bunch of #ifdefs in various files.
They just decided it wasn't worth the effort to maintain.
We'll never know. That's a binary blob on your machine.
> That's a binary blob on your machine.
On my machine? It's less likely than you might think.
There is probably a lot of use cases for many monitors that don't really require Windows (perhaps pro power-users, like traders, who need lots of browser windows open at once), so they need to ensure that their premium cards retain their edge even on Linux.
That makes much more sense.
That's a good point; I cannot say. Perhaps they think they are dealing with people who think 3D performance is necessary, or dealing with people (IT organizations?) that prefer 1st party drivers to hardware savings.
I don't think Microsoft is really involved in this though.
People buy hardware to use their advertised features. If you are the type of person that only needs 2D rendering and video then you should use an Intel GPU. I would assume people who buy a state-of-the-art dedicated GPU want to use them for actual 3D acceleration.
Think full-screen apps that call APIs that aren't going to work across screen boundaries.
I have the 560 Ti and it's capable of playing most current games comfortably. The newer cards seem to be able to do 4 monitors, but the 500 series is being left behind by nvidia as they push the 600/700 series with their shield and shadowplay features.
Then again, I find the simplicity argument somewhat convincing as well. NVidia's unified driver model must be hellishly complicated anyway, anything they can do to keep control of the complexity is probably something they'd greet with open arms.
Verified on three machines with identical hardware.
AMD doesn't have decent drivers even for Windows. (By decent, I mean less than 1 graphics-related crash on average per year.)
Counter anecdote mode: engage!
As an owner of many Nvidia cards, I find they crash infrequently, and when they do it's because:
* I was running Windows 8 Preview and pre-release Nvidia drivers
* There was some sort of hardware problem with my systems
I have used four ATI/AMD graphics cards over the years. Not all of them were negative experiences, but none of them were relatively more positive than using Nvidia hardware.
Before that, I had an ATI HD 5870 for over a year. I had about 3 crashes on the desktop, some that lost me work. The anger from losing work is what I remember most strongly, and made me vow not to buy ATI next time.
Before that, an 8800 GTX. I had 0 crashes from the desktop over several years.
I was running Nouveau drivers on the 670 for no more than 60 seconds before it crashed.
I think it's more about resources -- NVIDIA and ATI have lots of money that they put into driver improvements and maintenance, and the radeon and nouveau maintainers primarily have free time after work, occasionally with one or two full-time contributors. If we are serious about open-source driver platforms, we need to figure out how to allocate comparable resources. Some easily accessible tutorials on 3D driver development would probably be helpful.
Neither company ever plans to see an open-source driver rival their proprietary within a few years of a card's release, despite their "commitments" and "assistance". ATI's officially stated strategy is that they provide documentation so that the OSS drivers can provide basic compatibility and functionality, and that their proprietary driver should be used by anyone who has actual performance needs. They're basically offloading the onerous maintenance of previous generation hardware on the community.
The point is that nvidia is not especially bad. They've graciously given us the only consistently decent video driver on Linux for 10+ years now, despite its closed nature. Let's try not to be too hard on them.
I don't want to give the impression that I'm ungrateful for the help that ATI offered the implementers of drivers for its cards, but I think when we take all the facts in balance it comes out in a wash or even with nvidia slightly ahead: nvidia has been consistently producing a high-quality, near-performance-parity driver for Linux for a very, very long time now, and it's almost always compatible with new kernel and Xorg versions prior to their release. ATI has given docs for basic functionality in the free drivers, but they've utterly failed to produce good, performant drivers for Linux machines, closed or free, and in fact they are now on the verge of being two Xorg releases behind.
nvidia may not have "committed to help the open-source community" until recently, but they've done a great service for Linux by being the only vendor to provide consistent, solid, high-performance drivers for Xorg and the kernel. Intel has shaped up recently and gone full-bore OSS, but nvidia has a much longer track record of good [closed] *nix support.
i can tell you, however, that the specialist cards typically are only in super-computers. consumer GPUs are typically faster, and significantly cheaper than their HPC counterparts.
There's no reason at all to doubt his status as a NVidia employee.
So, no, it's not clearly an Nvidia employee. That's why I said
> I see no basis
We still don't know whether that's an official position or the speculation of a volunteer/low level employee.
Why would you assume that the NVidia logo was a default avatar when most posts on the forum appear at first glance to have a question mark as the avatar?
How are you still unwilling to conclude that he's a NVidia employee when a post of his about the status of NVidia's internal bug tracking system is corroborated and followed up by a post from a moderator whose signature identifies him as another NVidia employee? Yes, Sandip might be only a low-level employee who has nothing to do with NVidia's business strategies with respect to Linux, but he's clearly not just some random poster. (And for you to suggest that NVidia might let a volunteer have access to an internal bug tracker suggests you know nothing about how NVidia does business.)
edit: basically, you read 5-6 things into my words that simply aren't there. I avoid aggressive bickerers like you because I end up spending too much time clearing things up. You're primed to take the least charitable interpretation possible. I have better things to do than to deal with someone like you.
Anything is a better thing to do. I've wasted too much time in the past arguing with perpetually perturbed people to let it happen again.
I'm currently running an MSI 660 PE Ti with 3 screens in TwinView mode with driver version 319.49. This gives a total desktop of 4960x1920 in landscape-portrait-landscape layout.
I would like to get another 660 Ti and add another 3 screens in TwinView. Is that even possible?
Does anyone know how to use Intel + Nvidia on Ubuntu 12.04 (or later)? I was able to do this in the past with Windows but I can't get it working with Ubuntu.
"We are altering the deal, pray we do not alter it any further".
Closed source, closed negotiating position.
When you buy one of these cards, you are already assuming the position to get kicked, you are just hoping not to get kicked too hard or too many times in a row. Doesn't seem like a sensible thing to do really.
The Intel graphics are not so fast, but at least a big chunk are open source.
don't get me wrong, open source is great. we need it now more than ever. but this is exactly the wrong sort of attitude to take.
If you're interested in compute just pick up a 5850 or 5770 from a friend, they're cheap (if you can find them) and they run well with the open source drivers, and have support for at least 3 monitors. My 6850 has 2 DVIs, an HDMI and 2 miniDPs, I can drive 4 monitors at once with it.
In my experience, nvidia has been providing a superior linux experience in the trenches for the past couple of years. The 'amd is good for linux' seems to be an emotion from yesteryear.
Seems like your distaste for a company/policy is running this comment.