The Last Days at 3dfx

sand writes "FiringSquad has a detailed account of what happened in the final days at 3dfx. Every 3dfx product that was released or upcoming is discussed by a former 3dfx employee with inside knowledge on what caused the product delays (including an employee who forgot to fly to Asia to pickup the first Voodoo5 chips). He also discusses money mismanagement and the STB merger. It's a very enlightening article for anyone who's interested in 3D graphics and what goes on inside these companies."

3dfx became a religion

[Anonymous Coward]

3dfx became a religion. I bought an nVidia RIVA 128 and was ridiculed for buying from an unknown firm. That Glide was available for a few graphic cards only was neglected by 3dfx-zealots.

Re:They didn't innovate enough

[ites]

Technology does not solve social issues
and here it looks like 3dfx did not deliver the technology,
but IMHO the problem came because their product became a commodity item.
Frankly, the market for high-end graphics came and went.
Cheap on-board chips work well for 95% of users.
In such a market, only a couple of suppliers can remain
and it will be those with the lowest margins and costs,
not those with the best technology (which means creative people and higher margins).

Re:Businesses come and go

[PainKilleR-CE]

but today there is little need for faster graphics.

The need for faster and better graphics is exactly why 3dfx died. nVidia caught up and passed them while they were making mistakes like telling people they didn't want or need 32bit colour in 3D games or making 2d/3d cards that didn't hold up to their 3d-only boards.

Surprising this has not happened with soundcards

[ninjadoug]

Makes me wonder how Creative have managed to stay top of the soundcard pile. They seem to have been making consistently the best cards, (apart from a brief time when Gravis Ultrasound marketed using the Demo scene). No-one has really compared for the non-professional market. In fact I cannot think of any other Tech product where the first company to make something is still regarded the best. (Intel excluded) So from a business perspective maybe innovation is not the key but improvement of existing technolagy. Sad but true

3DFX and Real3D

[BobWeiner]

I used to work at a company called Real3D. The company was thoroughly mismanaged -- despite having an excellent engineering group. It's a similar tale to 3DFX, only R3D never quite penetrated the market. Eventually the company folded, all the engineers were laid off, and most of them have gone to work for ATI. Whatever was left of R3D was eventually consumed by Intel.

I remember walking by the manager of engineering 's office -- he was busy day-trading stocks all day. Our marketing department kept trying to add new features to our board (feature-creep-itis), trying to scramble to catch up to the competition. The introduction of new features really pushed back our schedules in a big way.

Poor management and poor marketing are what really killed R3D.

Re:3DFX and Real3D

[Anonymous Coward]

I used to work at a company called Real3D. The company was thoroughly mismanaged....all the engineers were laid off, and most of them have gone to work for ATI.....

Posting anonomously here to protect my buttocks... Once R3D got to that stage it degenerated into trying to extract income from other 3D companies by threatening them with patents. I believe that 3Dfx was one of the companies targetted.

In particular, they claimed to have invented MIP mapping/trilinear filtering when in fact prior art predated the filing date of the patent. It's enough to make you froth at the mouth.

Re:3dfx became a religion

[Czernobog]

Seeing as this is slahdot could you please explain to me what is so bad about 3dfx becoming a religion, when GNU/OSS/Linux and the like have reached a level of fanaticism amongst the developers and users that is hardly matched by any other social/technologic/scientific/religious movement?

In other words, what makes you think that OSS is more valid a subject of religious following, than a company making products, that up to a point in time reached new heights in performance in previously unexplored ways?

NVIDIA's G4? ATI's 9600? HA! I'm still using my V3 3000.

Re:Surprising this has not happened with soundcard

[PainKilleR-CE]

Makes me wonder how Creative have managed to stay top of the soundcard pile. They seem to have been making consistently the best cards, (apart from a brief time when Gravis Ultrasound marketed using the Demo scene). No-one has really compared for the non-professional market.

Both Gravis and Aureal made better sound chips than Creative, and better cards were made from the chips. Both companies lost to Creative the same way, too: Creative brought massive lawsuits with little merit that lasted so long the companies went bankrupt paying the legal fees to defend themselves.

In other words, Creative managed to stay at the top of the soundcard pile by legislating anyone that looked competetive out of existance.

Re:3dfx started to fail for this reason

[MentalPunisher2001]

Nope.
Until the Riva 128/Riva TNT arrived on the scene, 3dfx was the ONLY way to go.
Trust me - I even had a Rendition Verite card.
Don't even mention ATI's rage pro (or MY rage at the lack of decent drivers for it).
After the TNT, Voodoo 2 SLI was STILL faster.
The Banshee gave 3dfx a 2d/3d solution, but it was inferior to the TNT AND the Voodoo 2 (without SLI).
Later, 3dfx created the Voodoo 3 - in its many flavors, at different clock speeds.
NONE could render in 32 bit color.
Nvidia came out with the TNT 2 which COULD render in 32 bit color, and was slightly faster anyways (my V3 topped out at 200MHz, a lot of TNT 2 cards went even faster - and could use asynchronous memory/GPU speeds (yes, I know the term GPU was non-existent at the time - but it is now).
That was the time for 3dfx to shine with its Rampage product.
Nvidia released the Geforce - bringing geometry acceleration to the masses.
3dfx brought the Voodoo 4 and 5, which were 32 bit enabled. However, they did not have geometry acceleration, and used a more expensive multiple chip architecture to achieve semi-competitive performance. They were behind the times in an industry where you cannot afford to fall behind.
That was the end - Rampage never saw the light of day. Even the Voodoo 5 6000 (or Voodoo 6 6000 - I forget) vanished.
3dfx was good, but NVidia made some bets which paid off.
3dfx was used to LOOONNNGGG product cycles.
Remember how many years the Voodoo graphics chipset (original) ruled the 3d scene??
Remember how long the V2 SLI obliterated the competition??
Nvidia changed everything with their 6 month product cycles - less profit, but more progress.
Had 3dfx encountered stronger opposition in the Voodoo Graphics days, we might not be speaking of the company in the past tense.

Sorry if this is double posted - my login didn't work right.

Re:Competition

[jayayeem]

Price/performance ratio is not the only level of competition for graphics cards. It does happen to be the criteria I'd use if I were in the market.

Some buyers (office managers) will buy based on raw price.

Others (gamers) will buy on raw performance.

One company may eventually fill all three niches (and any others I may have missed) but I don't think it is the inevitable outcome.

As for DirectX, it is the minimum point of entry to the graphics market today. If you don't support it, and support it in the segment you compete at, you are dead. One reason 3dfx died, IMHO, is that it tried to compete at the high end, but its Direct X support was decidedly low end.

I'd read the article if it weren't blocked by my proxy.

Re:Reads like a 5th grader's essay.

[waltc]

Agreed.

Additionally, many of the details were wrong--for instance the author mistakenly thinks the V3 started shipping after nVidia shipped the TNT2. In fact, I bought my V3 in Februrary of the same year in which nVidia shipped the TNT2 in June. 3dfx had a big lead in performance which they let slip throught their fingers...it was some 18 months after the the V3 that 3dfx shipped the V5---Had the product shipped on schedule it would have coincided with the GF1 shipment from nVidia--and the V5 would have decimated it in performance. Being 6-7 months late there really hurt 3dfx. It was critical.

I agree that the company was mismanaged, however, as you say, this article tells us very little and seems to have been written by someone who worked at the company on a daily basis but was actually privy to information only on a scuttlebutt basis--could have been a Q&A or marketing person, for all it sounds like.

The most egregious errors were about "fear" and "sage", etc. These were if they were anything, concepts--certainly not products in development. 3dfx never got past the V56K stage--my own feeling is that the initial talks with nVidia were on track when 3dfx pulled the plug on the V5 6K. There were some very early crude and buggy prototypes of Rampage, I believe, but the rest of them were purely concept.

There's a lot to this story and some real insight would be interesting to read--but I doubt the officers in the company are going to air their dirty laundry in public--especially the details about the nVidia deal.

Why would slashdot think this was an "interesting, behind-the-scenes- look"? Beats me.

Ahh... the Voodoo

[neonstz]

I remember buying the Orchid Righteous 3D with a whopping 4 MB of RAM back in 1996. The graphics was just incredible. The bundled glide-version of Fatal Racing (or something) was very good, and even got LAN support for multiplayer action. I remember testing glQuake for the first time. I even cracked the first beta-patch for Tomb Raider. Tomb Raider in 640x480, perspective corrected, mip-mapped polygons. It was better than anything else.

Anyway, Glide was a very lowlevel API which basically just provided a polyfiller. A very fast polyfiller. With perspective corrected texturemapping, gouraud shading and z-buffer. Rewriting a software-rendered game to make it run on the Voodoo was very easy.

Re:GLIDE

[Zeinfeld]

We're all really fortunate that we avoided the nightmare of being locked in to a proprietary market controlling API from 3DFX. Luckily, we are in a new enlightened age where most games run on an open, freely shared API fostered by a community of the best minds from every segment of the industry. There's no limit to what can be done with our newfound freedom using APIs like Direct3D...

That is precisely the point. Given a choice between having the software standard set by a hardware company and a software company the market has always chosen the software company. It happened on glide and it happened on Windows.

The reason is very simple, the rival hardware companies are not going to allow their business to be subject to a competitor's control of the interface layer. However 'good' Glide was there was no way that it was in the interests of nVidia et. al. to support an interface controlled by 3Dfx. So it made perfect sense for the rival manufacturers to support DirectX.

OpenGL suffered from the same problem since regardless of the number of times SGI claimed that it was an 'open standard' the field was tilted from the start in favor of a rival hardware manufacturer that had a very different interest.

DirectX won because of elementary market dynamics and also because Microsoft presented DirectX as a gaming platform and not as a 3D platform. This was the critical wedge between the game companies and the OpenGL scene. DirectX has features like audio synchronization built into the core. There is simply no comparable standard for audio interfaces - the last attempt I am aware of was Jim Gettys work following on from the X Consortium.

Three or four years ago The Motley Fool chose 3DFx as a pick for the Fool portfolio. I dropped in on the discussion board and saw all sorts of chatter about how glide was going to rule and so competitors to 3DFx wer dead. I could see then that it was not going to happen and so decided to pass on the investment, just as well I did since it quickly became a dog.

Basically the only reason why the market ever opts for hegemony is to save itself from an even less tollerable hegemony with interests directly opposed to the stakeholders. That is why it decided that Microsoft was better than IBM and 3Dfx. Compaq, Gateway and the rest could see that Microsoft was an indirect threat while IBM was a direct one.

Lessons from "Been there, done that"

[Eric Green]

When I read the 3dfx article, I winced. The story was so familiar. Let me summarize it:

1. Bunch of cool geeks come up with great idea, and start a company.
2. They don't have lots of money, so they release a limited compromise version of the product. Even this version is really cool.
3. To handle all the sales and manufacturing tasks necessary because of the new best-selling gear, the vulture capitalists call in "professional" management, which might be someone who last ran a garbage truck company, or a carpet cleaning company.
4. The new management has no understanding of the market, so they look at what the biggest company in the market is doing, and say "We want to do that!".
5. In the meantime, this bunch of cool geeks is working on the great idea that will be the company's next product. There still isn't lots of money or people.
6. The new management says "Don't do that, we need this other product first!".
7. The new management strips the engineering team of the cool product. The engineering team pleads for more good people, but the new management says "that'll hurt our margins too much."
8. The other product flops.
9. The cool product has gotten obsolete while the other product flops.
10. Management panics and hires lots of dunderheads.
11. Meanwhile, management decides they can make more margin in another business, and put all the money from the sales of the first product into that new business.
12. The new business flops.
13. The cool product is being worked on by a few die-hard engineers, but is starved of resources. Delivery date gets shoved off further and further in the future.
14. Company realizes mistake. Hires a bunch of bodies off the street to work on cool product. Unfortunately, the bunch of bodies are dunderheads and make the project even *LATER*.
15. The first product becomes obsolete.
16. Sales plummet.
17. Company dies.

I've seen this happen so many times. The point:

1. *FOCUS*. Choose your niche. Don't try to be all things to all people. Decide exactly what you're going to do, and do that one thing. Only. Diversification is for the Fortune 500, not for a high-tech startup.
2. *EXECUTE*.
3. Hire good people when they're available, not when there's a project for them. A good person can always fit into an existing project.
4. Throwing bodies via mass hiring at a project dooms it to failure, because most of said bodies will be dunderheads. Most good people are available only for short periods of time. You have to hire them when they're available, not when it's convenient.
5. Professional management is important, but unless you have a visionary at the top who understands the market, the company will lose its focus, thrash around, and die.
6. *DELIVER*.