NVIDIA's Latest CineFX Card Under Linux 198
Nvidia Lacky writes "Ran across a new article from LinuxHardware.org that goes through NVIDIA's new driver release and also takes a first look at a CineFX-based NVIDIA card, the Quadro FX under Linux. Should be a good read for those that have been frustrated with Linux drivers in the past or that are looking to get a new workstation video card."
Brief summary from the webpage (Score:2, Informative)
Not so useful (Score:3, Interesting)
This type of card is optimized for giant data sets, zillions of polygons or nurbs at once. CAD/3D Modeling, etc. Think of the golum guy in LOTR.
Quake 3 has lots of blocky square guys running around at 300 fps on modern machine. Makes about as much sense as benchmarking it with the original wolfenstein.
This isn't what the card is designed for and it isn't probably going to perform as well as a card several hundred $ cheaper that
good review, bad conclusion (Score:3, Interesting)
The latest drivers fix a lot of issues with 2D performance, which could affect every day usage a lot.
Moreover, they fix compile issues on several kernel versions.
If you're desperate after 3D performance and don't care about anything else period, then stay with the previous drivers. Otherwise everyone is encouraged to upgrade.
Another step. (Score:5, Insightful)
NVIDIA has already been supporting linux lately, maybe poorly to some, but they are among very few who give a shit about it, so let them keep fine tuning, eventually it will be something rather beautiful.
Think of how long it took to perfect windows display drivers, they had what 15 years?
Linux is "new" to most people, it'll come in time and this is proof.
Re:Another step. (Score:2)
I wouldn't put my money on whether XP boots with a monochrome card;-)...
And windows hasn't really been around for 15 years (not as a usuable system anyway, anything before win 3 was Hardware change all the time, it all comes down to who has got the money to implement stuff and pay for licences...
Re:Another step. (Score:1)
Thanks for that useless rant, let me clarify, video cards and drivers are now very high quality because windows has been mainstream for a long time, simple enough for you?
Linux is new to mainstream, if it even is mainstream at all, so it will take time for it to develop, it has nothing to do with how good support is for a 1st generation video card, I don't know where you pulled that from.
CineFX will flop for a while (Score:3, Insightful)
Everyone wants the final specs of Doom 3 first.
I know thats what I'm waiting on.
I don't like ATI drivers but they're hardware is always good. Nvidia is the reverse, I've had a few nvidia based cards fail on me in the past but the drivers were always great.
Re:CineFX will flop for a while (Score:2)
Or not
Re:CineFX will flop for a while (Score:2, Insightful)
Re:CineFX will flop for a while (Score:2)
Some failed outright. Most of them get too damned hot and cause all kinds of stability problems... problems that were solved by putting, oh, anything else in the PCs in question. I'd trade all of the above for a few more cards that don't over heat or require noisy, failure-prone fans.
Nvidia could stand to take a couple of pages from ATI, particularly when it comes to fans and heat sin
Re:CineFX will flop for a while (Score:2)
Besides, have you seen that wart on top of their new card?
It is good *but* can be better (Score:3, Informative)
Re:It is good *but* can be better (Score:1)
It's nice that it can auto-dete
Re:It is good *but* can be better (Score:2)
I'm no stranger to Linux, having used it since '96, and I've been through Slackware, Redhat, Mandrake, and Debian, and a few others...
I don't recall ever having seen it anywhere other than /etc/X11/.
In the "Plug-and-Play" dists, I don't remember looking for it, so it's possible that it wasn't where I'd have expected it, but every time I've had to find it, it's always been right where it's supposed to be.
all seven of us are happy (Score:5, Funny)
The big question... (Score:1)
Too good to be true? (Score:2)
"NVIDIA Linux Update automatically detects the Linux operating system, kernel type, and CPU on a system and sets up the system for optimal performance and stability."
Sounds like the kind of easy driver setup windows users enjoy. I hope Creative follows this format and I can get my sound card working without hours of pain.
Re:Too good to be true? (Score:2)
Re:Too good to be true? (Score:1)
In sharp contrast to Nvidia, Creative hardware is managed by open-source drivers. And it just works. Period.
Re:Too good to be true? (Score:3, Funny)
Re:Too good to be true? (Score:1)
My experiance (Score:5, Interesting)
Re:My experiance (Score:2)
BR It's been that easy for quite a while now.
Re:My experiance (Score:2)
In fact, Nvidia keep the specs of their cards secret, so not only will they not provide free drivers, they stop anybody else writing them.
So by all means praise Nvidia for making the best of a bad job with their proprietary
Re:My experiance (Score:2)
http://www.nvidia.com/view.asp?IO=freebsd_1.0-320
An older version, but its there. I'm guessing it'll be updated soon.
Re:My experiance (Score:1)
Yes. For one of them, at least. [nvidia.com]
Re:My experiance (Score:2)
Re:My experiance (Score:2, Insightful)
But free driver releases aren't considered worthy of a Slashdot story - only proprietary ones with their cumbersome workarounds
Re:My experiance (Score:2)
Or is there something in the nVidia drivers that needs to exist in kernelspace?
Re:My experiance (Score:2)
Hmm, Matrox? Bear in mind that Matrox release high quality documentation for all their cards, have donated hardware to driver writers before and so on. That's why the Matrox drivers are generally high quality, get all the cool new features from X and DirectFB first and (more importantly) work out of the box on new systems.
To be honest, if I was going to build hardware specificall
NVIDIA and Linux (Score:2)
Now, here I am weeks later with no sound, under (updated) RH 8.0. Could someone please point me at a useful resource for diagnosing why sound isn't working, and fixing it? TIA!!!
I wouldn't mind upgrading to RH 9, except nForce 2
Re:NVIDIA and Linux (Score:2)
If not you're more or less shit out of luck
Should work smoothly under RH though as it's the "supported" distro IIRC...
Re:NVIDIA and Linux (Score:2)
You'd think. Not so far, though...
One error I did find in the readme.txt for the nforce2 platform drivers is that they claim:
rpm -rebuild foo.rpm
should work, however from what I can glean:
rpmbuild -rebuild foo.rpm
is actually correct.
Re:NVIDIA and Linux (Score:2, Informative)
It took a few minutes of searching the web, but the sound fix is on Mandrake's site.
Oh! after a little googling, I found NVidia has mandrake specific drivers that apparently work like a rested engineer.
http://www.nvidia.com/view.asp?IO=linux_nforce_1.0 -0248 [nvidia.com]
Anyway, here's the patch that worked on my system incase the new NVidia drivers work like a monkey on acid:
http://www.linux-mandrake.com/en/errata.php3#nforc e [linux-mandrake.com]
Hope this help
Re:NVIDIA and Linux (Score:1)
Re:NVIDIA and Linux (Score:2)
Re:NVIDIA and Linux (Score:1)
Re:NVIDIA and Linux (Score:2)
What I was really hoping for is some unified explanation of the Linux sound architecture. That would explain the relationship between /dev/mixer, the actual audio device, and the overall capabilities or lack thereof (3D sound, surround sound, etc.).
I guess, though, if I find the right combination of incantations it'll "just work". ;-)
Re:NVIDIA and Linux (Score:2)
I think my best strategy is going to be to let RH 9 settle out for a bit and wait for A7N8X support, as well as functional dual boot.
Thanks for the post though!
Article Text (Score:3, Informative)
The New Utility
Let's start this review by pointing you to a Bjorn3D article that talks about the Linux advantage. Their article basically takes you through a press briefing that the media received from NVIDIA. It's a good article that talks about NVIDIA's stance on Linux and what they hope to accomplish in the Linux sector. It would have been the intro here but they did such a good job....
Now let's get into the hard stuff and show you the goods. The new NVIDIA installer, based on the Loki installer, is designed to make driver installation painless. Let's walk through a standard installation:
Grab the Linux driver from NVIDIA's website here: http://download.nvidia.com/XFree86/Linux-x86/1.0-4 349/NVIDIA-Linux-x86-1.0-4349.run
Run the utility by typing "sh NVIDIA-Linux-x86-1.0-4349.run".
You'll be greeted with the screen shown below. This screen simply displays the license and asks you to either accept or deny it.
Once you've accepted the license, the utility will check for a kernel module that has been pre-compiled. If it does not find one, it will then ask if you want to check the NVIDIA ftp site for a module.
If you say "Yes" to the above question, the utility will go check and will return with either a module or a statement that you will need to build a module from scratch. Since we were working with a custom compiled kernel, we got the later.
When you select "OK", the utility will then go to work and compile the new module and proceed to install the rest of the driver package.
That's it. You will finally receive a screen stating that installation was a success and now all you need to do is configure your XF86Config file.
Now that you've seen how easy this installation can be, we'll now show you all the options that the installer has:
glacier src #
This program will install the NVIDIA Accelerated Graphics Driver for
Linux-x86 1.0-4349 by unpacking the embedded tarball and executing
the
Below are the most common options; for a complete list use
'--advanced-options'.
--info
Print embedded info (title, default target directory) and exit.
--check
Check integrity of the archive and exit.
--extract-only
Extract the contents of
run 'nvidia-installer'.
The following arguments will be passed on to the
utility:
-a, --accept-license
Bypass the display and prompting for acceptance of the NVIDIA
Software License Agreement. By passing this option to
nvidia-installer, you indicate that you have read and accept
the License Agreement contained in the file 'LICENSE' (in the
top level directory of the driver package).
--update
Connect to the NVIDIA ftp server 'ftp://download.nvidia.com'
and determine the latest available driver version. If there is
a more recent driver available, automatically download and
install it. Any other options given on the commandline will be
passed on to the downloaded driver package when installing it.
-v, --version
Print the nvidia-in
Re:Article Text (Score:2)
1. Install the nvidia drivers (last 2 revisions do this).
2. Launch gkrellm and keep an eye on the cpu meter.
3. Open any window and slide it around really fast. You'll see the cpu meter spike to 75% or more and stay there until you quit jerking the window around.
Now really, does it need that much cpu power to move a window back and forth? I'm on an athlon xp2100 here! Ridiculous.
tux (Score:1)
UT, UT2003, Quake III Arena, shall I continue? (Score:4, Interesting)
Re:UT, UT2003, Quake III Arena, shall I continue? (Score:2)
Re:HL, War3, BF1942, SimCity 4, Freelancer,... (Score:2)
Re:tux (Score:2, Interesting)
Frozen bubble on the other hand is a very nice game, as is nethack. Hmmm... I notice a pattern.
Nice, but not really a positive thing. (Score:2, Interesting)
Some people seem to believe that these drivers are Free Software (well, they usually say "open source"), simply because you compile the glue between the binary and the kernel, but this is not the case. The NVidia drivers are proprietary software, and it's a problem when a task can only be done using non-free software. Sin
Re:Nice, but not really a positive thing. (Score:1, Interesting)
Re:Nice, but not really a positive thing. (Score:2)
The reason for binary-only drivers suggested by another poster to this story - that they allow Nvidia to maintain a nonexistent distinction between 'consumer' and 'professional' hardware - is, if true, an example of market segmentation, a monopolistic practice. That is not capitalism or at least not well-functioning capitalism.
Re:Nice, but not really a positive thing. (Score:2)
Re:Nice, but not really a positive thing. (Score:1)
In the worst case it would be possible for Nvidia to release drivers which perform the basic 3d acceleration functions - those which are implemented wholly in hardware on the card. This would keep the free s
Re:Nice, but not really a positive thing. (Score:1)
Isn't it possible that, say NVidia has developed some amazing way of compressing polygons to speed up data transfer to their card and do not with competitors to know about it?
At home I use Linux, and I am very happy with it. However I believe that some flexibility in the open source approach is good. If somebody wishes to reveal their code, great. If not, let them play, too. Live and let live, as they say. A lot of companies are worried that the R&D money they invested in designing costly software i
Re:Nice, but not really a positive thing. (Score:5, Informative)
Here's the problem: nVidia and ATI make professional and consumer versions of their cards with the same hardware. The only difference is a resistor telling the BIOS which one it is. The _drivers_ are what tell the card to use certain features. While I'm not sure exactly how much is BIOS, and how much is driver, I'm betting having access to the driver source gives you a way to enable those professional features on the consumer cards.
Hence, not only are there no vendors like you want, but it becomes increasingly unlikely that they will pop into existence. Sorry.
-Erwos
Re:Nice, but not really a positive thing. (Score:4, Interesting)
Yes, and the real problem is that these companies insist on this dichotomy. I'm pretty sure NVIDIA could raise the price on it's consumer chips by 10% and eliminate the "professional" line with no loss of profit. The beauty of this would be (among other things) that consumer apps could use useful pro features like fast line drawing ;-) and fast pixel reads, which are disabled in consumer drivers.
It has always irked me when chip companies do totally artificial things to boost prices on some part of their line - like making 486 chips with a math coprocessor then disabling it to make "SX" chips. Silly practice.
Re:Nice, but not really a positive thing. (Score:2)
Re:Nice, but not really a positive thing. (Score:2)
Perhaps, except that if software actually used the pro features, those other cards would run like crap by comparison.
Also, you are presuming that NVIDIA wouldn't make up the difference in volume, as the thousands of firms using pro cards switch almost exclusively to NVIDIA. That wouldn't help those competitors much, would it?
Re:Nice, but not really a positive thing. (Score:2)
I'm a 'dirty hippie' (not even close BTW) because I want fully accelerated OpenGL on consumer hardware?
You need a good smack with a cluebat.
Re:Nice, but not really a positive thing. (Score:2)
Re:Nice, but not really a positive thing. (Score:3, Interesting)
It used to be that the Quadro lines used the same chip as the GeForce line, and that swapping a resistor (and the BIOS) would turn a GF card into a Quadro. Starting with the GF4/Quadro4 line, this is no longer true. I have a Quadro4 and the chip is most definitely different than the GF4 chip. Further, people have attempted to "hack" a GF4 into a Quadro4, an
Re:Nice, but not really a positive thing. (Score:2, Interesting)
Turning a GeForce 4 into a Quadro is also easy, however the antialiasing won't be quite the same...
I don't see why they used those benchmarks... Measuring the speed of quake for a workstation card is kind of stupid imo.
Re:Nice, but not really a positive thing. (Score:4, Insightful)
As you say, NO vendor provides such drivers. You know what? As long as zealots such as yourself keep making these sort of declerations, hardware companys will continue to avoid Linux/BSD/etc.
Think of the message you are sending to vendors.
Zealot: Hey Support Linux!
Vendor: Urm, okay
Zealot: Great!
Vendor: Here you go!
Zealot: Hey, IT'S NOT UNDER THE GPL! BOYCOTT!
Re:Nice, but not really a positive thing. (Score:4, Insightful)
Re:Nice, but not really a positive thing. (Score:2)
Actually, I do want to be free to bring my car to any mechanic, or walk over the Brooklyn Bridge any time of day, or read my books anyway I please.
Even if I may have to move to Paris, Freedom to do it.
Re:Nice, but not really a positive thing. (Score:1)
1. Graphics accelerators are primarily designed for Direct3D command interface first. Some OpenGL drivers might simply convert OpenGL commands into Direct3D commands.
2. Vendors may use proprietary optimisations in the device driver. This might include triangle stripification, deferred rendering or caching.
3. Vendors may also use cross-licensed technology from other sources. Any NDA would prevent the disclosure of this information.
The only way to have true free
Re:Nice, but not really a positive thing. (Score:1)
My new cards will all be ATI from now on, unless NV make their driver open source soon.
Re:Nice, but not really a positive thing. (Score:2)
The ATI Radeon 9700 and 9800 lines are superior to the Nvidia Geforce FX from a hardware point of view, so that's a good reason to buy them. But drivers continue to not be a good reason to buy ATI - they have a poor history on support and their Linux support is certainly not better than that of Nvidia.
Re:Nice, but not really a positive thing. (Score:2)
ahhhh, you gotta love that crack pipe. (Score:1)
Re:Nice, but not really a positive thing. (Score:2)
An OpenGL driver is a huge amount of code. We're not talking about a network card driver, that just bangs some registers, and can depend on the support of the kernel networking subsystem. An OpenGL driver is all of OpenGL. The NVIDIA driver encompasses everything from the header files, to the GLX module, to the kernel driver, to the X 2D driver, to the OpenGL library. This adds up to several megabytes of code. An OpenGL application interacts with the NVIDIA driver the
Re:Nice, but not really a positive thing. (Score:2)
Re:Nice,but not really a positive thing.-Patronizi (Score:2)
Re:Nice, but not really a positive thing. (Score:1)
How pointless.
Re:Nice, but not really a positive thing. (Score:3, Insightful)
I was an Amiga user in the 1990s.
That period of being an orphan, in spite of the beauty and the speed, held enough heartbreak and frustration for a lifetime. Really, it was enough, and I can now say with utmost confidence and conviction that it will never, ever happen to me again. And the strength of this conviction is equal to what you will hear when RMS says that he has the source code to his printer driver.
If something is not mine, then I simply won't have it. You call it pointless
Re:Nice, but ATI is the answer.. (Score:2)
What's the big deal? (Score:2)
The current setup for nVidia's drivers basically consists of installing the kernel module and using "nvidia" as the driver under XFree86 instead of "nv". The whole process takes about 3 minutes from download.
Shouldn't you guys be reporting on smart bombs or something?
Competition? (Score:2, Interesting)
3D accel on ATI rage 128 locks up hardcore! (Score:1)
My experience... (Score:5, Interesting)
From my personal experience, I was trying to install RH6/7 about a year and half ago. My video card is ATI All-In-Wonder Ultra Pro AGP. Its a 32 MB card, with TV-Tuner/TV-Out capabilities. It was fairly new at that time as Radon cards were just arriving.
1) RH installation was in graphics mode, but after that X wouldn't start.
2) After subscribing to about 3-4 mailing list and about 1-2 months of digging i found that the Chip ID in the card was not compatible , and all that was required was overriding the chipID with that of a previous version of the card.
3) Then after about anout 1-2 months of digging i was able to get h/w based 3D accl. I recompiled my kernel/ X atleast 20 times during those day.4) Another month to get the TV tuner working.
I am not a Video Card hacker, but i can compile and install X, Kernel modules etc. But hadn't been for the good folks at http://gatose.sf.net (not goatse :-) ) I wouldn't have beeen able to get it to work.
So it took about 6 months from the release of the card for me to get it fully working.
Re:My experience... (Score:3, Interesting)
Re:My experience... (Score:2)
gatose.sf.net is nothing but hardcore, hacking and reverse engg. of the cards, so its understandable that it takes a while for them to figure stuff out and support it. I can't imagine how may cards they must be buring each month
Al though i hear that ATI now has native linux drivers for some of its cards, not the entire range though.
Re:My experience... (Score:3, Interesting)
Re:My experience... (Score:2)
Re:My experience... (Score:1, Informative)
Seriously, though, your complaint seems a bit flat, as NVidia has done a huge amount of work on fixing this. I haven't had an NVidia driver issue with any of my systems in recent memory. In fact, we've been getting superb performance and easy installs for well over a year now.
Re:My experience... (Score:2)
[/humour]
Re:My experience... (Score:2)
Really I am not looking back 1.5 years ago to see how things go when I get a new card, which hopefully will be one of the new FX cards.
Is the FX actually out yet? If it is not then Nvidia have released a driver that enables use of the card before it even hits the shelves...This can only be a good thing, well one would think so
StarTux
Yes, but why no APM (Score:3, Informative)
KNOWN LAPTOP ISSUES
o Power Management is not currently supported.
*sigh*
Re:Yes, but why no APM (Score:2, Informative)
What they are refering to is laptop style suspend-to-memory features. However, with a simple 1 character source patch to the nvidia driver everything works fine.
I'm currently using my Dell C840 laptop with GeForce4. Suspend, AGP, everything works. See (suspend to RAM): http://jaeger.morpheus.net/linux/c840 [morpheus.net]
Re:Yes, but why no APM (Score:2, Informative)
You are talking about the old 4191 and earlier driver versions.
The new release cannot use the hack you are referring to.
Surprise, Surprise... (Score:1)
It seems that, once again, the only UNIX that's figured this sort of thing out is Darwin.
All hail OS X.
Seriously though, does anybody have a good understanding of what the real issue is here? Is the problem the Linux kernel itself, or nVidia not putting development money into it? The fact that there is no legit OpenGL implementation, or the fact that nobody see's it as a priority?
What's the biggest obstacle between Linux and proper, mature, accelerated 3D from a major vendor?
Re:Surprise, Surprise... (Score:1, Interesting)
Hardware specs. The manufacturers won't give them, or in some cases can't give them because they have contracts with other companies for licensing hardware designs or patents. TV-Out on ATI cards for example, is unsupported because ATI can't give specs because they have a contract with Macrovision. See this thread on gatos-devel where I explained this problem in more detail [sourceforge.net] (Note that I mention Macromedia ther
Re:Surprise, Surprise... (Score:2)
kernel headers problems (Score:2)
Re:kernel headers problems (Score:1)
Re:kernel headers problems (Score:2)
Do apt-cache search kernel-headers to see which kernel headers are available.
How does this compare to the current ATI support? (Score:3, Insightful)
I currently have a GeForce2GTS and have tried many of the drivers from NVidia with sporadic stability (yes, I've posted to their forums... no, they haven't been able to help).
I'm seriously considering a Radeon 9100... does anyone know how stable the current drivers from XFree are on this card?
Secondarily, how does the performance compare between the above two cards (under Linux)?
Yes, I do occasionnally use 3D in Linux (I don't use MS-Windows at all)... bzflag and UT mostly.
Let them know you are using it! (Score:2)
SuSE 8.1 or below should note (Score:3, Informative)
SuSE 8.2 users should be fine
StarTux
Any distribution? (Score:2, Insightful)
Re:3 minutes after posting... (Score:4, Insightful)
Very few of these projects have the funding to run a webserver that can withstand a slashdotting.
Re:3 minutes after posting... (Score:2, Funny)