3dfx's downfall?

[Borushka]

Hi, I am new to this forum and I just wanted to state before I am chewed up that, I do not intend to start a flame war, and I apologize to anyone who takes offense to this. If my statements sound general then you now know why =)

For a long time now I have seen the gaming/hardware community slowly turn it's back to the once holy 3dfx. A lot of the time I agree with the arguments made against 3dfx(i.e. lack of 32 bit color in the voodoo 3, missed product cycles, no hardware T&L) But, I would like to point out the few qualities that I consider still make 3dfx products valid contenders in the video market.

1.) Just because 3dfx does not choose to go the same route as fellow competitors, it doesn't mean the feature sets included with the card are any less valid or important to the gaming community. 1a.) T-buffer is just as useless now as hardware T&L when it was first available. But, game developers need an example or a start before they can begin to code for a feature. So in effect the Voodoo 5's T-buffer is really just meant to open the door for the next generation's T-buffer.

2.) I don't know what makes Glide so good to program games on, but the fact remains games are still being made for Glide, and Glide is also open source, so why don't other companies support it in their feature set for games like and based off the Unreal engine.

3.) FXT1 and DirectX Texture Compression are also open source. Once supported by games these technologies are sure to help speed up games and free up some of that precious texture memory on our video cards.

4.) VSA chips and SLI. I know a lot of folks ride 3dfx on how it only takes one competitor's chip
to do what multiple of 3dfx's can. Well, I believe multiple weak chips are just as valid as a strong one. If it was easy to make chips run in SLI, I am sure the twice as strong chip companies would do it. Ii just seems to me multiple weak chips are just as difficult to make happen as a powerful one. To 3dfx I think the multi-chip solution was one made for cost effectiveness or to solve some of the bandwidth issues.

Well I am sure there are more reasons out there, but I just hope I persuaded someone out there that 3dfx is not dead yet. Atleast wait for the next product cycle so the T-buffer effects can have a chance like T&L did. I look forward to reading your arguments.

Borushka

 

[Prodigy^]

they sure aren't it's not as if there are ANY possibility of them going bankrupt until the Rampage is released, so you suckers can keep dreaming

 

[pyr]

sure they arent dead yet but I have no plans to use one of their cards. and since i build a good number of computers that means that Nvidia will be getting a lot of cards sold. Ill look into others when the price and performance are there. I will not buy matrox after the g200 OGL scandal, and ATIs new set looks promising, but still the gforce still beats it. 3dfx just really has nothing that makes it worth buying their cards for. that blur crap looks awful. how can you rail someone in Q3 when they are blurry?

 

[pm]

I think this whole thing of treating any video card - or semiconductor company for that matter - as something "holy" is ludicrous. I guess I can used to the idea that people love their cars, their pets and maybe even their plants, but I really can't understand people getting so possessive about a chunk of refined sand. The zealots on both sides should go get a life in my opinion.

I say this because arguements about 3Dfx dying or whatever seem to always hinge around technology and benchmarks. Make no mistake about it, 3Dfx can survive as a 2nd, 3rd, or even 6th tier player indefinitely if only they solve their business problems. Their fundamental problem to the industry as a whole (and that's who counts, unfortunately, not users) is that they are unreliable suppliers: they miss product deadlines, their quality control is not high enough (V5 5500 two months ago, and others), their supply chain is unreliable (chips from TSMC, package supplies), and now that they aren't #1 and developers don't develop on 3Dfx products their driver support is increasingly looking unrealiable as well.

3Dfx don't really have a technology problem. Whether hardware T&L takes off or the T-buffer becomes the "thing to have" don't matter. Whether their next product has 8X AGP or whether it don't even have AGP multitexturing only affects whether hardcore users will pay big bucks to buy their products - it won't put the company out of business. But if they miss the ship date on the Rampage by more than 9 months (like they did on the V5) they will have a serious cash flow problem and it will only damage their already poor reputation within the industry.

I'm not saying that technology doesn't matter. Benchmarks and feature sets are important, but S3 haven't had either in years and they still seem to survive well. Heck, Trident is still in business and they haven't had a 1st tier product since the late 80's.

Patrick Mahoney
IA64 Microprocessor Design
Intel Corp.

* Not speaking for Intel *

 

[dawks]

1.) T Buffer isnt really anything useful, in its current feature set. Things like motion blur, and Depth-Of-Field are useless. But like you said, its a first generation technology. I can improve 2nd gen, Im open for that. But as far as I know, Its a closed source feature set. If im spending months developing a game, why do I want to program a few useless eyecady effects for a very small percentage of users. No.

2.) Glide Is a cheap way of doing OpenGL. When did glide become Open Souce?? Did I miss that press release?

3.) "FXT1 and DirectX Texture Compression are also open source Yea. All well and good, so theres nothing stopping nVidia, ATI, Matrox, etc... from adding them. nVidia already had S3TC in OpenGL, which puts them one up on 3dfx if/when they implement those two new forms.

4.) Dont really need to comment here. Multiple chip solutions are fine. Maybe nVidia needs to consider this to overcome the bandwidth issue.

Oh well.

 

[Wingznut]

1) Have you seen any games using the t-buffer? Imho, it's not possible to evaluate a feature, without experiencing it.

2) Yes, it was open sourced about four months ago.



Very nice post, Borushka. I for one, am getting tired of the "trendy" bashing of 3dfx. Most of the arguments being general and not very concrete.

Alot of them have to do with whatever feature the V3 line didn't have. Unless you are in the market for a V3, whatever features that card has/doesn't have, is pretty irrelevant.

By that logic, one could say not to buy nVidia, for their terrible OpenGL support with the NV3 (Riva128). Or their multiple problems with their first TNT cards. Well actually, you could probably discount all manufacturers, using this method.