Slashdot Mirror
The Age of Nvidia
EyesWideOpen writes "There is an excellent (and lengthy) two part article (part 1, part 2) at Salon detailing the rise, and... rise, of Nvidia and how the company came to rest atop the 3-D graphics chip industry with a little help from Microsoft. The article discusses how Nvidia was able to persevere in the multi-billion dollar industry while other graphics chip companies, such as 3Dfx which was bought by Nvidia, did not fare as well."
Well, lets see, we could do a long long long analysis of why they're so successful, or we could just state the obvious.
They are the most successful GPU company because they make the best, highest-quality, fastest GPU's, and make a wide variety of them: models designed for gamers, for graphics designers, for businesses. Not to mention, they support a broad range of OS' very well: Windows, Linux, MacOSX, and at one time BeOS. Not only do they support other OS' such as Linux, but their drivers for Linux are actually damn good: benchmarks show that Nvidia Linux drivers operate about 99% as well as Windows drivers.
This isn't like MS where they're on top because of dirty business practices. They're on top plain and simple because they make the best products, from every angle imaginable. Best quality, best performance, best OS-support.
This isn't to say that they're infallible, or always make the right decision. Personally, I think its rather idiotic of them not to support Glide in their GeForce drivers, as Glide offers vastly superior performance in games which use it.
social sciences can never use experience to verify their statemen
Nvidia has also faced recent SEC inquiries on two fronts. First, the company restated earnings for the past three years because of aggressive accounting, and the CFO took a leave of absence. Second, last year several employees were indicted on insider trading charges.
Sig (appended to the end of comments you post, 120 chars)
I've owned about a half dozen different 3d accelerators. I still have a Riva128 based card laying around somewhere.
The 3d market was 3DFX's to lose.
What killed 3DFX was that their good cards rewuired a 2d card to run. They were 3d only. The Banshee, which did incorporate a 2d core was late and always seemed buggy.
By the time they got their act together with the 3000 series it was too late.
At least ATI is starting to provide some competition or the damn graphic cards will cost more than all the other components combined.
Jesus used to be my co-pilot, but we crashed in the mountains and I had to eat him.
Don't take everything in the article as gospel. For example, the article talks about the Voodoo3 as 3dfx's first 2D/3D card -- "at last, 3Dfx had an integrated 2-D/3-D card!", but mentions the Banshee prior to that (which was a 2D/3D card).
Read the article for the point that 3dfx had the market and then went about losing a number of gambles. While that was going on nVidia got lucky and proceeded to execute a _deliverable_ plan like clockwork.
"All the darkness in the world can not quench the light of one small candle."
I have used both NVidia and ATI video cards recently. I found that the 2d image quality on the ATI card (a Radeon 7200) was superior to that of the Nvidia 2MX card that I had. This was only noticable on my apature-grille monitor but it _was_ quite noticable.
I also dislike the fact that NVidia's Linux drivers are closed-source, though of course most people do not care about this.
Oceania has always been at war with Eastasia.
I've had a GeForce 2 MX 400 in my machine for a year or two now, and I can't say I've ever had a complaint with it. The one thing I wish is that nVidia would roll out a card with a tv tuner and remote (a la ATI Radeon 8500DV). My friend has one of these, and after sampling it, I've come to the conclusion that "I want". But I also want a nVidia card. What I would personally like to see should nVidia attempt this is this idea, but on a GeForce 4 MX 440. This way, the price remains low enough to be competitive, and say what you will about the MX 440 not being a real GeForce 4, it's still a pretty nice card for a good price. But for now, looks like I'll have to make do with my (ATI) PCI input card.
http://cchart.yimg.com/z?s=nvda&a=v&p=s&t=1y&l=on& z=l&q=l
Come on,
NVIDIA was able to make the fastest GPUs on the planet because of the engineers they have.
SGI was really not a very good place to be if you were interested in pushing the envelope w/regard to 3D hardware, so a new company was formed, and many extremely talented people from SGI went to work for it. That company was NVidia.
Its sad to see SGI in it's current state, but it is also good to see that SGI's technology, with the proper focus, marketing and pricing, is capable of breaking into almost every segment of the computing market.
Obviously, kudos to the NVidia management team, but lets not forget either the engineers and the company that built the foundatation of 3D graphics on the desktop.
I gots ta ding a ding dang my dang a long ling long
nVidia may not have climbed as high and fast as it did if 3Dfx didn't hand them the entire market on a silver platter. If you would recall, 3Dfx all of a sudden decided they wanted to pull out of the OEM chip market and manufacture their own boards. So they bought STB and called up Diamond and Creative and all the other major video vendors and went something like: "No more 3Dfx chips for you. All your base are belong to us. Ha ha ha."
And who stepped into their place to fill the void left by them? nVidia.
Now the problem is that STB boards, by my many years experience in building systems, were incredibly low quality. I loved Diamonds, and had never had a problem with them. 3Dfx was gone from the Diamond line now, and I sure as hell wasn't going to start throwing money at STB boards, so I stuck with Diamond and bought a Viper v770...By then many games were supporting OpenGL and DirectX, so compatibility wasn't really an issue, unless I wanted to play Tomb Raider or something, but my old Voodoo would still work with the Viper. As an added bonus, and as you all know, nVidia's chips blew away anything 3Dfx had. nVidia's hardest battle, market acceptance, had been handed to them by 3dfx complete with bow and ribbon.
And then it was with complete irony that nVidia purchased 3Dfx. I love the tech industry.
-R
I've often heard Nvidia boards compared to Corvettes and ATI boards compared to Porsche's. Sure, the Nvidia board will render faster, but let's face it, quite a number of people seem to agree that the Radeon's just look better. I'll take accuracy every time over speed. It's not like the Radeon is a slouch, it's just not the fastest thing out there...
there are no stupid questions, but there are a lot of inquisitive idiots
If ATI cards are so much better, how come every benchmark shows Nvidia wiping the floor with ATI? All a vast "Nvidia-wing" conspiracy?
social sciences can never use experience to verify their statemen
I know a lot of gamers don't care about driver stability since they enjoy fooling around with their system hours on end.
However, rock solid drivers are nvidia's underrated asset. You don't know how much you miss stability until it's gone. Love to see them get more props on this.
I think the author overstates the influence of DirectX (nee Direct 3D) on early 3D gaming. Glide was certainly influential -- not to mention the fact that it actually worked -- but worthy of at least as much credit was OpenGL.
The reason OpenGL was (and is) important is because that's what you had to have if you wanted to run 3D-accelerated Quake. And Quake was the undisputed king of first-person shooters. OpenGL support for Quake required downloading a new executable, but Quake2 shipped with it.
OpenGL's API, designed over the course of more than a decade by SGI engineers, beat the crap out of Micros~1's Hacked-Up Losing Kluge. Only now is DirectX starting to approach OpenGL's usability.
Things are a bit more flexible these days, but back then, if you wanted any hope of selling your 3D card, you had to run Quake. And to do that, you had to support OpenGL. Period.
Oh, and NVidia has always had the best OpenGL implementation out there. Funny how that worked out :-). (Permedia's might technically be better, but have you seen what those cards cost?)
Schwab
Editor, A1-AAA AmeriCaptions
What do you think i'm talking about?
Try looking at the histroy of NVidia and SGI before you post silly flames which make you look like an ass.
SGI were once the pioneers and undisputed leader of the 3D graphics field. They invented lots of stuff, including, as you say, the popular OpenGL API.
What I am saying is at the time of NVidia's formation, SGI was an extremely suckful place to work if you were a hardware engineer - Knowing exactly how to build 3D hardware that would rock the world, but being unable to do it because of, among many factors, SGIs management strategy.
These guys' talents were being wasted, and they saw that with NVidia they could put them to good use.
I'm certainly not trying to belittle SGI's accomplishments in the field of 3D graphics, however i do feel it is unfortunate that SGI-as-we-know-it wasn't able to capitalize on its engineering assets as well as NVidia has to realise some of the vision of Jim Clark etc. w/regard to bringing the benefits of 3D graphics technology to the public.
I gots ta ding a ding dang my dang a long ling long
> we could do a long long long analysis of why they're so successful, or we could just state the obvious.
:-)
Or even better, the not-so-obvious.
> They are the most successful GPU company because they make the best, highest-quality, fastest GPU's
They're the most successful for two reasons. First, unlike 3Dfx, they focused on quick turnaround of incrementally faster processors rather than spending a long dev cycle working on very advanced technology that was too complicated to fit into a 6-month product cycle. The 3D graphics world was *starved* for more horsepower, and quick to jump on the bandwagon of whoever could deliver more faster, rather than the long-term strategy 3Dfx got mired in when their tech missed a whole dev cycle. This was an excellent strategy on nVidia's part, since 3Dfx's Rampage technology was taking far too long to pan out and forced them to release a "stopgap" line of cards that was short on features and performance, in order to try to struggle on until their mythic Rampage chipset could produce working silicon. 3Dfx poured all their investment into a product which would have been groundbreaking, but was so long to market that nVidia was running rings around them with their incremental strategies.
Second, much like Enron, nVidia (allegedly) inflated their financial statements in a very unethical manner in order to draw in more investment due to steadily rising stock prices during the investment bubble. Honesty is punished by investors if it isn't all wine and roses; inflated financial statements draw more investment. In the case of Enron, the house of cards collapsed. In the case of nVidia, the tail wagged the dog--inflated financials drew more and more investment, which funded more and faster product cycles, which allowed nVidia to really pull ahead of 3Dfx, just as 3Dfx fell further and further behind thanks to their Rampage sinkhole. The high investment due to questionable financial statements is what allowed nVidia to fund its whirlwind snowjob, culminating in the purchase of its beaten and devalued old rival. There's been an SEC probe into these purported financial improprieties, and from everything I've seen, it looks like nVidia's creative accounting was their source of power, funding their product cycles--kind of like winning by cheating. No, *exactly* like winning by cheating...
This demonstrates a few principles we already know from much practical experience. In computers, short-term strategies which produce small gains *now* are much more likely to be successful than long-term strategies which would pay off big, but not in the near future. IA-64 is a prime example of this--Intel's roadmaps when Itanium first shipped showed it being adopted in droves by this point in time, yet it hasn't been; if an when it succeeds, it will be because of Intel's unusually deep pockets, but meanwhile x86-64 Yamhill has been developed "just in case" AMD's Hammer architecture captures the low-end-server and mainstream desktop markets, markets which Intel had *insisted* would eventually have IA-64 trickle down to without any interim architectures. This same principle was seen in the software world, with for example every single version of Windows that was built atop DOS rather than NT.
The second principle of success which nVidia's strategy illustrates is a financial one, illustrated well by Enron. People invest more money with companies which are already financially successful than with ones who really need the money, so that inflating the bottom line is rewarded immensely--and punishes companies which are honest, by giving fu7nding to their competitors. With Enron the bubble burst. With nVidia, the bubble carried them to the top, and funded dev cycles which neither 3Dfx nor Matrox nor for most of that period ATI could compete with. It's a gamble, and the dice rolled in nVidia's favor. That doesn't make their alleged financial improprieties right, but it makes them (if true) a *major* factor in nVidia's success.
> models designed for gamers, for graphics designers, for businesses.
3Dfx did the same, so nVidia is in no way unique there. In fact, high-end graphics maven Quantum3D was a 3Dfx spin-off intended by 3Dfx to be a major user of 3Dfx's highly scalable chip architectures (8-way Voodoo 2's and 16-way VSA-100's, for example, which *killed* everything else at the time for the high-end). For mainstream businesses, 3Dfx had their line of STB boards (following their STB buyout, which many see as a huge mistake, since they got into the board business instead of concentrating on just chips). And for gamers, obviously, the famous Voodoo lines. Low-end-professional 3D graphics wworkstations were the only market not really targeted, since Quantum3D boards compete in a higher-end space than Quadros did.
> Not to mention, they support a broad range of OS' very well: Windows, Linux, MacOSX
As did 3dfx, but 3dfx bettered nVidia in this respect by releasing a large chunk of code. nVidia has on the other hand been excruciatingly secretive with almost all code.
> This isn't like MS where they're on top because of dirty business practices.
Then why did there need to be an SEC probe into their financial (mis)statements? Again, if not for the funding attracted by reputedly "too optimistic" financials, nVidia could never have pulled off the quick incremental development cycles which kicked 3Dfx's ass.
> Personally, I think it's rather idiotic of them not to support Glide in their GeForce drivers
This is the one thing I agree with you about. Glide is rightfully dead--its limitations are well-known, and today DX and OGL are the clear choices. However, a "legacy Glide module" would have been *very* nice, as almost all older games with Glide support work much better in Glide, and some older games *only* work in Glide. This is precisely why I bought an old Quantum3D Voodoo2 X-24 dual-Voodoo2-on-a-single-card board as a secondary adapter for my gaming rig--it's the only way to have full compatibility with many older games. If nVidia were unwilling to spend their time writing it, the Open Source community would likely be glad to do it for them since many are avid gamers and fans of old classics--but nVidia refuses to release any code, even the obsolete Glide code.
Now, let me go play a round of Turok in asskicking Glide mode, courtesy of my dual Voodoo2 card, in honor of the dead.
Chasing Amy
(We all chase Amy...)
"The more corrupt the state, the more numerous the laws"-Tacitus
The article is interesting, but it has several mistakes. First of all, T-Buffer tech was introduced on the V5, and the article mistakenly stated that the V4 came prior to the V5, when they were released at the same time. These are somewhat minor quibbles, to be certain.
If any of you remember, the purchase of STB befuddled everyone, and for good reason; STB's products were a mix of Nvidia and 3dfx chips, and OEM's had the freedom to pick and choose what they wanted to buy. Furthermore, 3dfx had great co-branding with companies such as STB, Creative Labs and Diamond (I still get all twittery when I remember waiting to get my hands on a Diamond Monster Voodoo II). In one fell swoop, 3dfx destroyed what was best about STB, and it's co-branding with other manufacturers.
The smart money left shortly thereafter.
The ensuing fiscal mayhem following the purchase of Gigapixel was a financial blow (coupled with late product releases) that they simply would be unable to recover from.
Had the Voodoo 4 and 5 been released on time, they simply would have crushed the TNT 2 Ultra and put them in a much better position to pay off all that enormous debt. But, the card was late, and it had to compete against a far superior offering from Nvidia, which was the Geforce.
And the smart money that left a long time ago was not wondering if, but when.
So, not any single decision led to the downfall of the once dominant player, but many. Not listening to the market (we don't have 32-bit support for color in games since people don't really need it...take 16-bit or else!). Excess execute hubris such as the purchase of STB and Gigapixel and the foundering on product release dates. Trusting on name brand and uncompetitive products all eroded the company to nothing.
In terms of Nvidia, their executive staff has always been able to seize on opportunities, and possess a remarkably clear vision of where they want their company to go in the marketplace. Their purchase of 3dfx's IP (which also included Gigapixel's IP) for only 70 million was absolutely brilliant, as was the absorption of 100 of 3dfx's top engineers ensures that Nvidia will be able to utilize all the fantastic goodies 3dfx had sitting in the R & D lab.
It's also really great that ATI is able to mount such a good force of competition in this arena; along with maybe-will-runs such as Matrox and 3D Labs...all this competition keeps em on their toes.
Have the Bit Boys ever gone into tape-out? Or did they soak up the former executive staff from 3dfx?
This is not ment as a flame, but a correction.
They're the most successful for two reasons. First, unlike 3Dfx, they focused on quick turnaround of incrementally faster processors rather than spending a long dev cycle working on very advanced technology that was too complicated to fit into a 6-month product cycle
Actually, it was the exact opposite of this. nVidia never produced a video card that was faster than one of 3dfx's top model while they were still in business. In matter of fact, it was the technology in the early nVidia video cards that were driving their sales. Check out tom's hardware archive and read through the articles of the past and you will see that the benchmarks show 3dfx clearly winning the frame-rate race, but it was nVidia's 3D image quality that was coming on top each time. It was only when 3dfx went for an entire year without ever coming out with a new chipset, did nVidia finally catched up in speed.
The second principle of success which nVidia's strategy illustrates is a financial one, illustrated well by Enron. People invest more money with companies which are already financially successful than with ones who really need the money, so that inflating the bottom line is rewarded immensely--and punishes companies which are honest, by giving fu7nding to their competitors.
If you are talking about the whole SEC thing, that was a recent occurance that began at the start of last year. There was inside trading going on and that was what the investigation was about. However, in the beginning nVdida was a privately funded company. They went public AFTER 3dfx went bankrupt. Shadey handling of money had hardly anything to do with their triumph over 3dfx.
As did 3dfx, but 3dfx bettered nVidia in this respect by releasing a large chunk of code. nVidia has on the other hand been excruciatingly secretive with almost all code.
Quiet the opposite, nvidia has created a site to drive development of 3D software http://developer.nvidia.com/ They encourage open source and try to get the entire community in helping establish and designing standards. Which brings us to the last point...
Glide support work much better in Glide
Glide is the farthest thing you can get from programming standards. I cringe everytime someone spews the statement "Glide is better." Do you honestly have an understanding of what a proprietary API is? Glide works fast on a 3dfx card, because it's the "language" that voodoo "speaks". It's like running a windows application through an emulator on a Mac and whining that the G4 architecture isn't as good as x86 because it doesn't run as fast as on a native machine.
nVidia is a good company. They come this far through the work and sweat of a very talented group of designers and programmers. Don't try to smear that with the shadey business practices of certain individuals that unfortounetly worked at the company.
This guy is an obvious M$ troll please don't feed him.
http://Lenny.com
4 great justice!