NVIDIA PR Responds To Torvalds' Harsh Words

[johnc]

They're too busy scoring big wins in the mobile sector to worry about the relatively minuscule number of open source desktop users.

[V!NCENT]

Linux user responds to nVidia's empty respons to Linus' hasr words below.

Still haven't heared nVidia do something about their SUCKING.

My stance regarding nVidia is therefore still:
FUCK YOU, nFailia!

[hal2k1]

You need to read the legend before coming here to correct me.

Pfft. You were wrong under any interpretation of the feature matrix. Those items which are still TODO are that way because there is not information available to wite a driver to. It is NOT because open source developers are incompetent, slow or lazy. Please get your facts straight.

There are many features that the open source Radeon driver does not support yet.

There are a few, yes.

It seems you do not see those features listed as TODO and this means it needs those features to be compatible with OpenGL spec or provide the features the Windows drivers includes.

WTF? TODO items are still TODO because there is as yet no information published (mostly due to legal concerns) on how to interface to those parts of the AMD/ATI GPUs. Hopefully when legal review is complete a way can be found to publish the programming specifications for these still TODO features.

As it stands right now the open source Radeon driver is not ready for production systems. It is OK to use for simple 2D tasks, but for OpenGL will have a problem.

Actually, as it stands right now, the open source Radeon driver for Linux is perfectly stable and compliant with OpenGL 3.0. I don't know where you get information from, but wherever it is they are woefully out of date. As for 2D acceleration ... that is what the Linux desktop actually uses, and the open source Radeon drivers for Linux perform better (at 2D) than the fglrx closed binary blob does.

Gamers or users running 3D applications should still use nVidia graphic cards to ease playing their games or using 3D applications in Linux since nVidia's close source drivers are ready for production if you like them or not.

Gamers are a tiny market, games are a trivial use of computer hardware, and they have virtually nothing to do with Linux. If you want to play games, why don't you buy a game console?

[V!NCENT]

I think there are two camps here (open and blob) which use the same language but a different dictionary. Let me explain:
Open Good: works stable, secure, gives enough features to be perfectly usable "It just works", good integration with Linux kernel mechanisms. Perfect for production usage on the desktop. High speed desktop effects. Everything might not work in the driver, but is useable nontheless because for example video can run perfectly, without tearing, in software. There are no fancy gimmicks and uber framerates, but that's less irritating than the shortcommings of the blob (like slower boot, no KMS, less security, upgrade limiting in time, lifespan, etc.)

Blob Good: gigantic full spectrum latest standards and protocol support, massive optimal compute power, full video acceleration, shiny graphics, gimmick support (3D bullshit and "EyeFinity" 3+ screen support, etc.

So when we are discussing "superiority", first ask yourself superiority in WHAT.

What's superior for me is the open definition of Good. Blobs can be better at their stuff that I couldn't give a rats ass about, because if I supstract $99 for an Apple TV and $250 from an Xbox360, from a $400 Crysis nVidia card, I'm still left with $50 spare for actually buying games and renting movies, or modding my Xbox in the case I'm not into buying anything. And even then I don't have to upgrade the console as much as I would have to upgrade my GPU. Capiche?

[tecknurd]

Au contraire, here is the feature matrix for the open source radeon driver:

http://www.x.org/wiki/RadeonFeature/

Green means "done".

Yellow "WIP" means work-in-progress. Most of the time this means new work is required as such a feature does not exist anywhere (e.g. Video Decode {XvMC/VDPAU/VA-API} on the 3D engine, this doesn't exist anywhere, not even on Windows).

Red "TODO" means no documentation has been provided for that feature (e.g. Video Decode {XvMC/VDPAU/VA-API} on UVD hardware).

There is very little missing from this feature matrix because of the reason you imagined, to whit: "open source community is not fast enough to keep up with the changes that need to be done". This is for the most part simply untrue, that is NOT the reason why missing features are missing.

ORANGE or WIP "means that someone has started on the initial implementation." This means that someone started writing code for the feature, but outcome of it is unknown. The feature with this status is unstable for production.

YELLOW "means that it is implemented but has some known bugs." It explains it self. The feature with this has bugs which means unstable for production.

RED "means that someone needs to write the code. The required knowledge to write the code may or may not be known. Please ask on #radeon if you want to get your feet wet on this." This means nobody has been scheduled to handle that type of feature and nobody knows if it can actually be supported for each Radeon model.

Radeon graphics does support video hardware decode since the first Radeon model in Windows, but ATI at the time wrote it so horrible that it was unstable in Windows. AMD not providing any documentation for UVD just prove that I am right that not even AMD is willing to be completely open source to the community. This may change, but I still doubt it.

You need to read the legend before coming here to correct me.

There are many features that the open source Radeon driver does not support yet. It seems you do not see those features listed as TODO and this means it needs those features to be compatible with OpenGL spec or provide the features the Windows drivers includes. As it stands right now the open source Radeon driver is not ready for production systems. It is OK to use for simple 2D tasks, but for OpenGL will have a problem. Gamers or users running 3D applications should still use nVidia graphic cards to ease playing their games or using 3D applications in Linux since nVidia's close source drivers are ready for production if you like them or not.

[scjet]

Lenovo Thinkpad T420. It's about a year old I think.

a year yes, your just under the wire on that.

But this: http://ubuntuforums.org/showthread.php?t=2008824 and 1000's of other users, wth many posts similar are gambling whether or not this or that newer model can switch it properly ?
Linux on a new Laptop, especially a powerful one with an expen$ive Nvidia-gpu, should NOT be a gamble for new users.
Until Nvidia provides a "proper" Optimus driver solution for Linux, this will always remain an Nvidia/Intel/WinBlowz party only. They are basically saying NO to a powerful Linux Laptop-Worstation choice for Linux users.
.... hence, Nvidia <- fu_k you!
A couple years ago, only a handful of these incorporated Optimess, but those days are long gone, because all the newer laptops,... will have it, whether we want it, or not.

I actually just picked up a cheap-used Dell Precision M90 purposely (with nvidia Quadro-FX2500M) -Lol. Everthing is just there though -Linux, (or any of the *BSD's), X with 3d glx nvidia driver... all working smoothly.

[kotakotakota]

Ya?, well then post the make and model, how old is it ?.... please.
Because ALL "newer", and future laptops with nvidia have Optimus (hard-wried), which is "useless" for Linux right now.

Lenovo Thinkpad T420. It's about a year old I think.

[scjet]

Just a note, some BIOS do let you choose to disable Optimus and manually specify which GPU you want to use. I did that on my Thinkpad and works great.

Ya?, well then post the make and model, how old is it ?.... please.
Because ALL "newer", and future laptops with nvidia have Optimus (hard-wried), which is "useless" for Linux right now.

[kotakotakota]

You "think" ??? -go think with your own $1799.00 bub. Seriously, unless its a couple years old, OR, it's the newer i think? -lol Nvidia GPU/AMD CPU Laptops?, then nope.
So, sadly enough, we're all stuck with an utterly USELESS "new" Optimus Laptop,..., for Linux.
Even "Clevo",... and/or any of the very few remaining others are building them all "with" Optimus now. - NO choice, 'cause it's hardwired, no thanks to Intel/Nvidia/Microsoft.

So ya, I aggree with Linus: "Nvidia, fuck-you !"

Just a note, some BIOS do let you choose to disable Optimus and manually specify which GPU you want to use. I did that on my Thinkpad and works great.

[V!NCENT]

I think NVIDIA has picked up on these market trends and is trying to diversify its way out of the gaming niche. They wanted to go the x86 IGP route but Intel said no... so they're looking at using the ARM architecture instead (Project Denver). If they can pull that off they'll be in a good position. They've focused heavily on the mobile space (and have established themselves as a big player in that market). GPGPU computing appears to be an important part of their strategy and it looks like they're going to dabble in cloud gaming as well (a bit risky).

The interesting thing is that Linux will probably be the cornerstone of all of those avenues.

By looking into that, I can conclude that nVidia indeed jumped to the conclusion way earlier than I did (and appearantly my subconcious knew that too, giving me the idea that I came up with it myself, oops ).

But that better be fully open documentation, because otherwise that succes will be shortlived when consumers find out that buying nVidia means upgrading whenever nVidia thinks it's time to upgrade while the consumer might not want that yet. That would result in "never buying nVidia again!". Or at least having open source support ready at blob support discontinuation of the product.