3dfx Interview

okay everyone. i\'m a gamer, your a gamer. as long as i can play my games with nice details (32bit) and with nice speed (30+fps) then i\'m a happy camper.
btw, the GPU is gonna change the whole world...well, not in my book yet. i have to see that shiz in action. will 3dfx live to it\'s hype...it\'s still on the \"drawing boards.\" what about S3s 2000 and the G400? the S3 has potential (have to see it to believe it still) and the G400 - no comments yet.

Yeah but doesnt Darth Vader kick ass much more than some little kid? :)

Also you can argue that Nvidia have been doing the same thing and repeatedly dusting off the same hardware TNT and TNT2 are very closely related, infact the progression from Voodoo 1 to current day has shown more development (from no 2d side to the card at all to twin texturing units to having a 2d side etc) 2.5 years of mastrubating it couldnt be that bad, after all theyve been working on evolving the system they have in the Voodoo series, simply because its not revolutionary doesnt mean that its shite, in addition 2.5 years isnt really that long in some R&D matters, remember how long the N64 took to develop and how kick ass it was when it arrived h/w wise compared to the other console systems.

Ok, maybe that was\'nt nice. But you are off on almost every point.

You said:
Even if the game performance is identical, the CPU itself is doing less so you can do several things at once

Nope. The whole point the boys at B3D were trying to make, is that when T&L is not enable in the software, the CPU usage will be identical to a non-T&L card.

You said:
a card that is ALWAYS fill rate limited is GOOD...That means when 10 guys run on the screen there is NO slowdown at all!

NOT! the term \"fillrate limited\" does not mean the Frame Rate is locked. You are you are confusing the term with somthing else.

You said:
but lightmaps do NOT look better then correctly used real lighting
One of the main points of the B3D artical was this: Vertex lighting looks just plain ugly when rendering large flat surfaces. Run around in quake 3 for a while, the majorty of brushes are large and flat.
Just like #176 (wwerd) said \"Vertex lighting only looks good if you use umpteen polygons\"

You said:
You know what? Who cares! Nothing is perfect
Yeah, but I\'m not going to shell out $300+ just because games that benifit for it are eventualy going to show up.

My feeling is that T&L technolgy is a little ahead of it\'s time. Although i\'m glad nVidia is doing what they can to push the industry forward. I doubt this chip will make much of a splash, or sell many boards.

Vertex lighting only looks good if you use umpteen polygons (a la A bug\'s Life), otherwise I suppose you can see ugly gradients and angular edges between vertices. You can run Q3test with vertex rather than lightmaped lighting. See which you think is bettah.

#174 You poor clueless SOB.

fuucckk - english - Im back to bed :P

\'really pants\'

difference performance results I meant -duh-

\'pants\'

Looking at the graphics cards today even at 640x480 they all give different specs (differences in the architecture I assume) but in essense does the large fill rate of the next 3dfx product mean that performance from one resolution to another will be the same- even now I notice a drop on the V3 going from 640x480 to 800x600 then (moreso) higher?

Must admit the GeForce sounds good, the fill rate argument is a valid one however, remembering that in the interviews one of the Nvidia PR guy said why bother with full screen AA when you can run at 1600x1200 - can it at appreciable fps judging by the released specs? Still depending how the benchmarks pan out when its released I might just swap over to it (or then again wait another generation -money depending natch).

\'pants\'

#164, fill-rate has NOTHING to do with the CPU.

Another thing is when realtime lightning is used it\'s no need to use multitexturing for lightmaps.
So its better to have ability to render four pixels/cycle than two multitextured pixels / cycle.
And another good thing is that GeForce it supports Trilinear/ anisotropic filtering for free.
So when you turn compare V4 AA Trilinear vs GeForce Trilinear 3dfx needs to have 8 times greater fillrate to match it.

Just admit it, they\'ll go easier on you. 3dfx owns j00.

Screw the new Nvidia card... does it run glide? If not, it can bite me.
Don\'t bother hating me... you will never match my old girlfriend.

Dammit, where\'s my beer bitch?

Uh sorry, I forgot to add to the end of my last post: That when 3DFX & NVIDIA say they can give us \"A Bugs Life\" in realtime. they are lying

Since we are quickly aproching the limitations of silicon, and massively parallel computing is far too expensive.
We will probably NEVER see a consumer product that offer \"Hollywood on the Desktop\" at least not in our lifetimes.

Yes, this thought is depressing but... I still have the October issue of playboy featuring Killcreek to look forward to! yay!

Ok, that\'s it for today. I swear I won\'t add to this thread again today.

aleks..i was wrong, you\'re right...i\'m drunk..i just spilled my life story to some chicks..i feel like shit....sorry about taking oyur phrases out of context...im an nvidia fanboy...im pretty stupid when it comes to computer hardware...if i insulted you, i apologize..i hate flamewars..who gives a fuck about video cards....jesus h fucking christ..theres so much more to the world....g\'night


did i mention im drunk?

BTW, if your curious. It took 270 SGI Origin servers to brender the moive \"Antz\". The 270 Origin servers had a combined total of 700 MIPS CPU\'s. (Needless to say, MUCH MUCH more prossessing power than 157 puny PetuiumIII PC\'s)

It only took 997 hours (41.5 days) to render from start to finnish on this monster rig. But during production, they had to render the entire movie over 15 times.

#153 Buying 157 PC\'s would solve that problem. =)

No seriously you could go out and buy 157 PentiumIII class PC\'s at 2000$ each. 157 x 2000 = 314,000$

That\'s still pretty cost effective considering the budget of most hollywood movies.

#150 means nothing, 100 and 200 is the shit.

#147, where the hell is the fun in that? Come on, join in the convo and flame/be flamed. :)

hawkeye nice try saving face there, but you never admitted that you were wrong to flame me.

This statement

Yes, when it was released, it decimated all the rest, but who cares?

is pure bullshit. By your reasoning, the TNTU is shit because the GeForce or Napalm will be better, in fact by your reasoning every card in the world is shit because eventually there will be something better. And you flamed me because I dared say that the Voodoo2 was the best Quake2 card at its time. Geez.

About your price/performance concerns, wouldn\'t you say that the Voodoo3 2000 is a no-brainer over a TNTU just because it gets a pretty similar score in most games, and it\'s 2x cheaper? Somehow I don\'t see you saying that, seeing as you think everyone who has something nice to say about 3dfx is a \"fan boy\". So what if a card is $100 more than another. The TNT2U is the best card for Q3Test right now, PERIOD. Sure the TNT is good too and it\'s better price/performance, but it\'s just not the best card for Q3Test right now is it?

Fuck I hate making such obvious arguments for such close-minded people. And you call it \"disjointed ramblings\", the least you could do is say \"I was wrong, I\'m sorry\", you\'d get much more respect that way.

#142 The point is there is NO computer, not even the fastest SGI computer, can render \"A Bugs Life\" in Real Time. I\'m saying you do, but I\'ve talked to many people who think that there are these \"uber-fast\" graphic workstations, where you can just load some 3D renderer and playback \"A Bugs Life\" in real-time. SGI\'s are fast, but nowhere near that that fast. Each frame can take several seconds to several mins on the fastest of machines.

The reason why 3dfx, Nvidia, etc constantly bring up this movie in their advertising, is only to emphasize that their card are going to exceed anything we have ever seen.

Of course 1600x1200 is still no where near FILM quality, even with AA. And then there are advanced rendering functions such as \"raytracing\" that no 3D accel card will be able to touch for along long time.

Still the nextgen cards should bring us a step closer to pre-rendered CGI.

But truthfully, it think will still have to wait many years, before any computer will be able to render the \"A Bugs Life\" movie in Real Time.

oh steve..
chk blues.. there is some thing 3dfx related..
http://www.qamex.com/full-list.asp?sitesection=unreal&articleid=1569
UT SHOTS with V2
heh

What about a voodoo3 2000 pci
$90 after rebate..
MORE bang for the damn buck

#141..fine, if we can compare SLI to a single TNT, then we can compare that stuff to SGI workstations and stuff too. Right? I mean they\'re available. The computers that did \"A Bugs Life\" must be available somewhere too. Lets compare those to a TNT. See what I\'m getting at? If SLI cost the same amount as a TNT, then fine, compare them. Until then, its not a fair comparison. SLI cost about 200-250 bucks, a tnt cost around 100-150. Performance versus price shows the TNT is a better \"bang for the buck.\"