Slashdot Mirror

Nvidia's Cg compilers have been open sourced. Download here.

Making the NForce2 onboard ethernet card work in Linux is easier than you might think.

Either: Go to NVIDIA's websiteand get their driver wrapper, and compile it (if you are using kernel 2.5+, there is a patch for the builder to make it build a .ko module instead)

Or, help start testing the new ForcedEth clean room implementation of the driver. It's not perfect but it works, and it was included in Andrew Morton's -mm patch set in about 2.6.0-test9; so as soon as it's a bit better I'm sure it will be in the mainline kernel.

It's a bit hard to get it going compared to out of the box support, sure, but no different to the rigamarole I have to go through to get almost any network card working on Windows 98.

If you want an excellent source of notes on the NForce2 and Linux, that you are encouraged to update as you find things yourself, check the WLUG Wiki.

1. (shuffle)
2. Hope you find these comments contructive - they are not meant to assault.

   You also have to make it painless to do things like install/remove software and install/remove drivers.

To install software, click on the package you want to install or drop in a CD built for your distribution.

Remove/install drivers: Why bother now? This was a problem a few years ago, but not now. If you have the hardware, it'll be identified. If not, the module (aka 'driver') just won't load.

Nvidia has an installer that works well enough. Use it.

Root/user seperation and the pain of logging in: The password tool in Red Hat / Fedora works well enough; login 1x, and for a few minutes have full root access from the desktop. I'd be surprised if other distributions don't have this feature, though I haven't looked over the past year.

Odd crashes: Agreed. Specific programs bork for no good reason on occasion.

Polish: Agreed, though except for OSX, there seems to be quite a lack of polish for nearly every OS out there.

User adoption: If the OS is installed, people will use it. If not, they won't all the sudden decide to switch unless they highly motivated and willing to commit to deal with any issues that come up. My little sister had no problem, and she's at best "challenged".

Tech widgets: Yep, though they don't deal with those things elsewhere so there's no motivation (or reason) to do it under an X desktop.

At least with an NVIDIA chipset, I know I will have great Detonator driver support.

Ah, Tom's Hardware. Not trying to be negative, but IMHO, they are a terrible source for tech information, and the bulk of their reviews contain startling errors, conclusions that defy reason, glaring omissions, and sensationalized reporting.
The majority of those writing the reviews clearly have no idea what they are talking about, at least regarding the subject they are reporting on. Overall, I would rate them slightly above HotHardware.com.
Tom himself, as far as I can tell, is on the ball and knows his stuff VERY well, but he doesn't write articles much anymore, and obviously doesn't read them either.
It is a common practice among hardware enthusiasts to quote Tom's for the humor value, trying to see if the author of the latest article is even more clueless than he was in his (or her) last article.
To be fair, they do have some excellent articles occasionally, and were the first ones to dare publish information on Intel's unstable Pentium III 1.13GHz processor, but unfortunately these seem to be the exception rather than the rule.

Also, as has already been stated, XGI is hardly a new company. Of course, these bits of SiS and Trident are in completely new territory if they are trying to compete in the high-end gamer's market. Considering that this is their first real foray into that market, I think they have done an amazing job. I'd say give them the benefit of the doubt until they prove otherwise. Remember, even the (once) most respected companies in the field can faulter, and that XGI has something that is even in the same ballpark as the most seasoned of players is an impressive feat.

Oops! Did I just flush my karma cache? Damn!

Oops! Did I just flush my karma cache? Damn!

Oops! Did I just flush my karma cache? Damn!

I dunno...I think they probably should have cut down on the budget for the artists and shelled out a bit more for better looking real women.

I prefer a thumbstick to a keyboard and mouse.

I'm one of (apparently a minority) of people who doesn't buy consoles for this reason alone. If consoles offered me the precision of a mouse I'd be much more apt to jump on board. As it stands analog sticks usually control viewpoint velocity (stick in center = 0 viewpoint movement) making it hard to aim (yes, I'm an FPS player), and those that don't force me to hold the stick in position when its trying to move; either way its no good for FPSes. With a mouse however I just whip it into position where I want to look and presto: snipe.

The console is catching up to the PC graphically as well

Indeed. This is an unfortunate consequence of too many game developers trying to maximize their market share by including all the people who never upgrade. If you keep watch on all the graphics demos that NVidia, ATI, et al. release you know what I'm talking about. Modern GFX cards are capable of doing vastly superior effects than what you see in most PC games. (Example screenshot: Rendered real-time, NVidia demo called 'Dawn')

Windows:

Kerio /or Sygate for firewall (both are good)
Aladdin's free StuffIt Expander (unpacks a lot of different compressed files, including SIT and Gunzip's)
AVG antivirus (free for personal home use)
QuickClear lite (deletes IE cookies/cache/empty's trash)
StartPro (well, it used to be free. Gives you a nice list of programs set to load at bootup, including registry keys.)
Ad-Aware everybodies favorite adware/malware answer.

Mandrake is (of course) easy:

Got the Easy Urpmi and follow the directions to install all the different media sites. Once you do that (its just a cut and paste job) you can fire up rpmdrake and search for software by name/description/type/etc. Mandrake installs with a lot of the right stuff already. I'd recommend maybe installing nano (easy command line text editor if you hate VI/VIM/EMACS/ETC) and of course if you running a system with a NVidia card get the NVIDIA drivers (rpmdrake, but if their not listed NVidia will have them).

nvidia compiler only difference is that it advantages their hardware, ati run slower than it should.
However, since the CG compiler is open source ATI could go out, and add performance enhancements for their hardware.

Strange. They're trying to get a locked-in niche market for their hardware, I guess.

Too bad, since they're using an nvidia chip and nvidia has a stereoscopic display package already.

In reality, it's no more proprietary than many PC BIOSes out there. Just because it's not as well understood does not make it proprietary.

If you were saying this five or six years ago, then you'd have an argument, but with all the New World machines (Blue & White G3 and later) it's been Open Firmware and pretty well documented. Hell, even OpenBSD boots on Apple hardware now and, you know, those kids are not the quickest to support proprietary anything. They don't particularly enjoy reverse engineering hardware because it's more difficult to make your software stable and secure that way.

Apple has generally been quite open about documenting their own hardware for the last few years. The same certainly can't be said for certain other PC parts manufacturers.

I don't think the problem is too much exploitation. After all, that's what the modern entertainment business is built on. Harry Potter, Star Wars, everything else is exploited just fine without the brand losing value. In case of Lara the problem might simply lie in the fact that two primary brand development engines - games and movies aren't so hot.

If I was a brand manager at Eidos, I would concentrate on making the next game really good. Scrap the plans to reuse the engine for several games. Make a couple addons, if you must, but license a decent engine (Half-Life2, Doom3 or something simplier) and make an exciting, interesting and really good-looking game. Today's computers (and tomorrow's consoles) are capable of displaying very realistic and hot-looking chicks. Think of what is possible when you can dedicate 30% of the GPU power to rendering the main character, something that looks like the Nvidia's Dawn or even better. Realistic rendering of Angelina Jolie is possible if Eidos wanted that. If they combined it with decent gameplay that made original TR so cool (come on, this is not art, this is craft, you just need to hire good game designers), the game would sell like crazy and with the right marketing will attract a wide audience, including both young males AND females. Then Lara will be ready for the next round of exploitation all right.

behold the NForce 3 pro. It was the first(?) chipset supporting the Opteron, has many of the features you mention. As for having support for legacy, etc; that's the choice of the motherboard manufacturers. As an example, the Asus A7N8X only has 3 dimm slots, whereas the Gigabyte GA-7N400-l1 has 4.

BTW, for if you really want a legacy free mb, you'll want to check out abit's offering, the AT7. It may not be the latest greatest, and it may not fit your vision of 'legacy-free', nor is it the ideal board you describe.However, its as close as we'll get for now.

A question for you. With all of those onboard components, why do you need 6 pci's(also getting to be legacy at this point)?

-D

It doesn't bother me that much who has the fastest card. All I know is that this sort of competition is great in the Linux arena. With the recent trends in 3d animation studios transition to Linux, they can't ignore the need for high quality drivers.

Nvidia has really polished up their Linux drivers recently, and in response ATI has done the same.

This means Linux is one step closer to gaining a foothold on the desktop. Hopefully this will will spur interest 3D gaming on the linux platform.

One can dream of the day of playing Battlefield 1942 on Linux. I'm using the Liux FireGL drivers on my Radeon 9700 Pro, and so far, they work great for playing RTCW ET.

• In case you don't know, nVidia provides official (but woefully non-GPL) drivers. They also have a message board which I found to be quite informative at times.
• Compile your kernel with MTRR support. It will speed things up a great deal.
• Compile your kernel without AGPGART support. The nVidia driver(s) are faster.
• If you want to try the nVidia driver with a 2.5 kernel, you'll need a patch.
• If you have an nForce chipset, make sure to add "mem=nopentium" to your kernel boot parameters, or else your system will be incredibly unstable. Better yet, ditch your nForce chipset (I did) since the Linux support totally blows, at least for now. Give your old nForce chipset to your wife, girlfriend, mother, Windows box, or whatever.

"But very quickly I had to spend a couple of hours doing things like learning how to set up header files in order to re-compile my kernal to support NVIDIA drivers."

You download a binary and run it, just like you would for Windows.

Then you reboot, just like you would for Windows.

Then it works, which doesn't happen in Windows.

Why would you need to recompile the kernel? That would only be if you had odd hardware (supercomputers or such like) which Windows doesn't support anyway.

If you have Windows on your system you probably don't mind proprietary stuff on you Linux box. Just download the Nvidia driver, run it from a shell (without X running) and start up X. Works everytime.

Although nvidia does suck for releasing drivers under the licenses they do use (apart from their lone GPLed audio driver), their NForce2 chipset has a driver package for sound, network, and GART. Works well with slackware.

Just FYI. ;-)

Everything will be there however, including the heavy (ab)use of stencil buffers. That and high polygon count is what makes Doom 3 look so good.

Actually, the lighting is about half the equation (which is quite good - true stenciled shadows and per-pixel lighting effects ). The other half is normal mapping . The modelers for Doom 3 created creatures for the game in the 1-2 million polygon range, and crunch that detail down into an RBG map used to offset a much lower resolution mesh. It's similar to a bump map (grayscale relief), only it has three components - the RGB values correspond to XYZ offsets, allowing you to cram a lot of the million-poly detail back into a 2-5 thousand polygon mesh.
Not to be confused with true displacement (true displacement is what's used to turn a flat plane into a terrain mesh) - in Doom 3 no polygons are being shifted, which is why in the screenshots the edges of the critters still look blocky.

Everything will be there however, including the heavy (ab)use of stencil buffers. That and high polygon count is what makes Doom 3 look so good.

Actually, the lighting is about half the equation (which is quite good - true stenciled shadows and per-pixel lighting effects ). The other half is normal mapping . The modelers for Doom 3 created creatures for the game in the 1-2 million polygon range, and crunch that detail down into an RBG map used to offset a much lower resolution mesh. It's similar to a bump map (grayscale relief), only it has three components - the RGB values correspond to XYZ offsets, allowing you to cram a lot of the million-poly detail back into a 2-5 thousand polygon mesh.
Not to be confused with true displacement (true displacement is what's used to turn a flat plane into a terrain mesh) - in Doom 3 no polygons are being shifted, which is why in the screenshots the edges of the critters still look blocky.

Everything will be there however, including the heavy (ab)use of stencil buffers. That and high polygon count is what makes Doom 3 look so good.

Actually, the lighting is about half the equation (which is quite good - true stenciled shadows and per-pixel lighting effects ). The other half is normal mapping . The modelers for Doom 3 created creatures for the game in the 1-2 million polygon range, and crunch that detail down into an RBG map used to offset a much lower resolution mesh. It's similar to a bump map (grayscale relief), only it has three components - the RGB values correspond to XYZ offsets, allowing you to cram a lot of the million-poly detail back into a 2-5 thousand polygon mesh.
Not to be confused with true displacement (true displacement is what's used to turn a flat plane into a terrain mesh) - in Doom 3 no polygons are being shifted, which is why in the screenshots the edges of the critters still look blocky.

Out of the box, with the addition of a HD i/o card, probably a good SCSI RAID disk pack.

SGI's always been about moving massive amounts of data internally; your (and my) multi-ghz systems are still spending the vast amount of time stroking off while waiting for disk reads, memory copies, that sort of stuff.

I remember getting my shiny new Gefore3 and running the Zoltar demo for the first time. Amazing detail and quality and what not, but it actually pops up a, well, popup, saying 'please wait while we transfer an ungodly amount of data to your video card!'

What's the point of having a whomping video card when it takes a good thirty seconds to a minute just to transfer the data required to render a head and neck?

I think he's taking the Ghandi approach to teaching them IP-grubbing corporate motherfuckers at nVidia (Who make excellent drivers) a lesson--Force himself to suffer and maybe they'll take sympathy and throw him a bone after a while.

But even that isn't so simple anymore. Consider the nForce2 Audio Processing Unit. The tests I've seen indicate that it is the least CPU consuming audio solution available. You can even use 6 channel digital output in games with it. If I understood correctly it combines the channels in game in to one dolby digital stream.

NVIDIA SoundStorm
Firingsquad's Audigy 2 - nForce 2 comparison

Another possible reason is that he's running a Via motherboard which requires him to bring his AGP strength down all the way. This is mentioned in the nVidia driver manual.

Speaking as someone who just install Debian and the NVidia drivers, you're full of it.

Just grab NVIDIA-Linux-x86-1.0-4363.run and run it. Answer a few questions and either reboot or insmod nvidia, and launch X. Boom NVIDIA drivers

It took me absoluetely no time in order to figure that out, and in fact I dont even see (not that I looked very hard) where you can download the GLX and kernel drivers anymore.

All in all if you dont even know wtf your doing it takes 3 seconds to install.

Not with NVidia. Take a look here:
Nvidia Linux
They support Linux IA32, Linux AMD64, Linux IA64 and FreeBSD. They support their chipsets, etc. I also think a good portion of Intel's stuff is "offcically" supported under Linux like their graphics chips and audio. And even if they do not release a driver, they provide the specs which is more usefull to Linux then a proprietry driver in most cases.

For those of you with nforce chipsets who have non-nvidia graphics cards, this link will give you agpart support. It's the new release of the linux drivers. It's not linked to from the nvidia.com main page, for some reason, but it works.

They don't support FreeBSD (unlike nVidia)

Futuremark Statement
For the first time in 6 months, as a result of Futuremark's White Paper on May 23rd, 2003, Futuremark and NVIDIA have had detailed discussions regarding NVIDIA GPUs and Futuremark's 3DMark03 benchmark.

Futuremark now has a deeper understanding of the situation and NVIDIA's optimization strategy. In the light of this, Futuremark now states that NVIDIA's driver design is an application specific optimization and not a cheat .

The world of 3D Graphics has changed dramatically with the latest generation of highly programmable GPUs. Much like the world of CPUs, each GPU has a different architecture and a unique optimal code path. For example, Futuremark's PCMark2002 has different CPU test compilations for AMD's AthlonXP and Intel's Pentium4 CPUs.

3DMark03 is designed as an un-optimized DirectX test and it provides performance comparisons accordingly. It does not contain manufacturer specific optimized code paths. Because all modifications that change the workload in 3DMark03 are forbidden, we were obliged to update the product to eliminate the effect of optimizations identified in different drivers so that 3DMark03 continued to produce comparable results.

However, recent developments in the graphics industry and game development suggest that a different approach for game performance benchmarking might be needed, where manufacturer-specific code path optimization is directly in the code source. Futuremark will consider whether this approach is needed in its future benchmarks.

NVIDIA Statement
NVIDIA works closely with developers to optimize games for GeForceFX. These optimizations (including shader optimizations) are the result of the co-development process. This is the approach NVIDIA would have preferred also for 3DMark03.

Joint NVIDIA-Futuremark Statement
Both NVIDIA and Futuremark want to define clear rules with the industry about how benchmarks should be developed and how they should be used. We believe that common rules will prevent these types of unfortunate situations moving forward.

About Futuremark\x{00AE} Corporation
Futuremark\x{00AE} Corporation, formerly known as MadOnion.com, is the leading provider of computer performance analysis software and services. Futuremark\x{00AE} is known around the world for its benchmark products, including the 3DMark\x{00AE} Series and PCMark2002 (with more than 30 million copies distributed worldwide) and value-added services powered by a database of over 5 million real life benchmarking results. Futuremark\x{00AE} has offices in Saratoga, California and Helsinki, Finland. For more information, please visit http://www.futuremark.com.

\x{00A9} 2003 Futuremark\x{00AE} Corporation. 3DMark\x{00AE} and PCMark trademarks and logos, Futuremark\x{00AE} character names and distinctive likenesses, are the exclusive property of Futuremark Corporation. DirectX\x{00AE} is a registered trademark of Microsoft Corporation. All other trademarks are property of their respective companies.

About NVIDIA
NVIDIA Corporation is a market leader in visual computing technology dedicated to creating products that enhance the interactive experience on consumer and professional computing platforms. Its graphics and communications processors have broad market reach and are incorporated into a wide variety of computing platforms, including consumer digital-media PCs, enterprise PCs, professional workstations, digital content creation systems, notebook PCs, military navigation systems and video game consoles.

NVIDIA is headquartered in Santa Clara, California and employs more than 1,500 people worldwide. For more information, visit the Company's Web site at www.nvidia.com.

Certain statements in this press release, including any statements relating to the Company's performance expectations for NVIDIA's family of products and expectations of continued revenue growt

So under "right to privacy", we are never allowed to include someone's home in a photograph? I guess that makes this product completely illegal.

From the above-mentioned website:

The EarthViewer3D free trial has been temporarily disabled.

Streisand's suit complains that the photograph is of extraordinary clarity and violates her right to privacy, as it shows details of the property that one would not ordinarily be able to see from the road or the beach.

So under "right to privacy", we are never allowed to include someone's home in a photograph? I guess that makes this product completely illegal.

Or is it only rich and famous Hollywood stars whose homes are covered by "privacy" acts. After all, who would want to look at a picture of my shaby old 200K home.

Streisand only seems like she is for the "little people" when it benefits her---either by raising publicity for her or by making her feel better by "fighting the EVIL REPUBLICANS". Strange that it is *HER* that is fighting this environmentalist's work and not some land developer or corporate polluter. But you can be sure if she wins, every land developer and corporate polluter will be using her case as a precedent. After all, don't they and their workers deserve privacy as well?

Brian Ellenberger

" If you were smart, you would use Cg which is ATI and Nv compatible, but no luck, it's Win32 only. No Linux or MacOS X version. "

Really!?

Here.

Check out the hair on this NVIDIA shot

and compare with the hair in this shot off the ATI.

The zoom on the NV example isn't quite as close-up, but you can still see how the hair feathers away at the tips, while the hair on the ATI is all jaggy and uniform thickness right out to the tips. Like some kind of Raggedy Anne yarn hair.

So does the ATI not support alpha-to-coverage? Or did these guys making the wrapper just not translate it properly? The hair looks pretty bad without it.

I seem to remember when I saw the NV guys give the demo that they made a point of how expensive the hair was to render. Each hair is a separate GL_LINE_STRIP, there are thousands of them, and if you turn off the complicated blending and smoothing on each one of those little hairs, that could maybe add up to a large savings. I don't know, but I wonder if that would account for most of the speed-up they see running it on ATI.

Still a neat hack. It would really rock to have complete NV<->ATI compatibility dlls that would work for all apps and not just this demo.

While they may have made some slight improvement to one aspect of image quality by improving normalization, which I guess makes the lighting a little more accurate, I really doubt the improvement is all that noticeable. Maybe I could tell given a side-by-side comparison, but I doubt I would notice if only shown a version normalized one way or the other. On the other hand, that ugly hair is pretty obvious. To me that makes the NVIDIA sreen shots look better.

If anyone wants it, I've mirrored the OpenGL wrapper here (78k).

You can get the dawn demo from here

direct link

"I'm really interested in what kind of chipset NVIDA will be offering for AMD64-based CPU's!"

There is the nForce3 Pro.

"You are probably going to have to write your algorithms in assembly (or even possibly in hex machine code) to absolute addresses in the GPU buffer. Then what are you going to do when the next generation of GPU comes along -- start that coding all over again?"

You can use Cg which is a higher level shading language for GPUs. No need to write in assembly.

"I mean forget about X-86 as a requirement, are you going to listen to your users whining that your software requires a particular model graphics card?"

This is less a problem that you think. For example $499 Geforce FX 5900 Ultra and $79 Geforce FX 5200 can execute exactly same shader programs. FX 5200 is just slower.

Many complain that GPUs have limited functionality. That's true, but the progress has been enormous:

- Just 2 years ago GeForce3 was state of the art. It could execute only ~20 pixel shader instruction and the presicion was limited to 8bit integer per channel.

- GeForceFX can execute 1024 pixel shader instruction with static flow control and 32bit floatin point presicion per channel.

- Within 2 years we will have a GPU which can execute virtually unlimited amount of instructions with dynamic flow control and 64bit per channel floating point precision.

You will still need CPU for desktop work, but GPUs will be used for many things like scientific computing, simulation and off-line 3D rendering (making CGI effects for movies).

In some situation GPUs can be used already . In some cases QuadroFX (which uses "old" NV30 core) can render broadcast quality scenes 20 times faster than fastest CPU available.

Now I gotta Find out how to install Drivers.

Go here

Here are a few examples:

• Opera 6.0
  I can't get my hands on a license without downloading the software :(
• RealPlayer
  The same thing. They obviously don't want us to read it unecessarily :)
• Flash
  You may not alter, merge, modify, adapt or translate the Software, or decompile, reverse engineer, disassemble, or otherwise reduce the Software to a human-perceivable form. That's more clear cut, and straight out said than in most licenses :)
  ...
  You may not modify the Software or create derivative works based upon the Software.
  ...
  You may not export the Software into any country prohibited by the United States Export Administration Act and the regulations thereunder.
  Which I guess means that Cubans, Lybians, Iranians, North Koreans, Syrians and some others can't use the software, but I may well be wrong as for which nationalities this is.
• Java
  Except as specifically authorized in any Supplemental License Terms, you may not make copies of Software, other than a single copy of Software for archival purposes. Unless enforcement is prohibited by applicable law, you may not modify, decompile, or reverse engineer Software.
  ...
  You may not modify the Java Platform Interface ("JPI", identified as classes contained within the "java" package or any subpackages of the "java" package), by creating additional classes within the JPI or otherwise causing the addition to or modification of the classes in the JPI.
  ...
  You acknowledge that the Software may automatically download, install, and execute applets, applications, software extensions, and updated versions of the Software from Sun ("Software Updates")
• NVidia drivers
  ... SOFTWARE designed exclusively for use on the Linux operating system may be copied and redistributed, provided that the binary files thereof are not modified in any way (except for unzipping of compressed files).
  ...
  Customer may not reverse engineer, decompile, or disassemble the SOFTWARE, nor attempt in any other manner to obtain the source code.

As you see, the licenses limit the terms of use , modification and redistribuation of the software in ways which are to me not acceptable.

As far as drivers, I'm primarily thinking of a Windows-only box, since I want my dual-boot machine (the one with the F15-sounding fan in it) back on Linux full-time. Nvidia released fairly new drivers for linux. Do therse have missing features as well? I was thinking of making a little 10GB partition for Linux just for grins, but if I have to wait for the maturity level of chipset drivers to go up I guess I could.

-B

Sadly, they're also walking the Voodoo path with their new "strategic alliances" under which publishers will develop games using Nvidia specific features. It's a pathetic tactic from a company that built their success on strong engineering rather than lame marketing ploys.

Check it out. Your next workstation?

I believe that America's Army, Splinter Cell, Unreal Tournament 2 and Unreal Championship were all built on the Unreal 2 engine. Unreal 2 was not released for months after the other games were.

Nvidia Games: America's Army
UT2K3, UC, Splinter Cell

Admittedly, it is difficult to discern whether some of these were written with the original Unreal engine, but I think they all used the most recent one.

This will be resloved with the new nVidia GeForce FMXP 41000x It's supposed to be 3rd quarter this year. It's supposed to totally smoke anything ATI has right now, with photographic quality and with at least 6 times the Q3 fps. And it has outputs to a film projector. And built in lasers. And 7.1 Audio onboard. And a soda/Pizza exhaust port.

....and it's supposed to be profitable.