• Please review our updated Terms and Rules here

The Geforce 3: why?

H

hunterjwizzard

Veteran Member
Joined
Mar 21, 2020
Messages
890
When I got into retro gaming, the first thing I bought was a Voodoo3 3000 on the basis that I couldn't see what top end 3dfx glide games looked like otherwise. These cards still command a high price-point for that reason.

Then there's nVidia's offerings. My understanding is most of the geforce cards contain all of the hardware to support rendering modes from the previous cards. So for example a Geforce 6800 AGP has all the same capabilities as a Geforce 3 Ti, but more RAM, faster GPU, etc.

So, why then do Geforce 3s command such a high price today? Sometimes as much as a Voodoo3 ?
 
hunterjwizzard said:
So, why then do Geforce 3s command such a high price today? Sometimes as much as a Voodoo3 ?
Click to expand...

"Because people are dumb" is the only reason I can think of. My vague understanding is that a GeForce 3 is effectively architecturally the same a GeForce 4 but is no faster than a GeForce 2, which is why it wasn't a particularly long-lived product(*). I guess there are bozos out there that really want to experience hardware anti-aliasing on the slowest possible card.

(*) Maybe that's it. The didn't make them very long before the Geforce4 came out, so its "rare".
 
Eudimorphodon said:
(*) Maybe that's it. The didn't make them very long before the Geforce4 came out, so its "rare".
Click to expand...

Maybe you're right and it is jus the "collectability" aspect of it. There are tons of people who simply collect graphics cards for fun, possibly even more than those of who build retro PCs for fun.
 
Eudimorphodon said:
My vague understanding is that a GeForce 3 is effectively architecturally the same a GeForce 4 but is no faster than a GeForce 2,
Click to expand...

The reason the GF3 was often slower than a higher end GF2 card, like the GTS or Ti is because of the hybrid architecture. The Geforce 3 was the first card with programmable shaders, vs. fixed function pipelines. It included some Geforce 2 technology because the new programmable shaders weren't fast enough to emulate the old fixed function pipelines at a respectable speed.

Depending on the game, the Geforce 2 and 3 traded places. Games that used the older Direct X 7 and prior APIs were generally faster on a Geforce 2 Ti, Pro or Ultra. Where Direct X 8.0 titles only worked on the Geforce 3 card.
 
GiGaBiTe said:
The reason the GF3 was often slower than a higher end GF2 card, like the GTS or Ti is because of the hybrid architecture. The Geforce 3 was the first card with programmable shaders, vs. fixed function pipelines. It included some Geforce 2 technology because the new programmable shaders weren't fast enough to emulate the old fixed function pipelines at a respectable speed.

Depending on the game, the Geforce 2 and 3 traded places. Games that used the older Direct X 7 and prior APIs were generally faster on a Geforce 2 Ti, Pro or Ultra. Where Direct X 8.0 titles only worked on the Geforce 3 card.
Click to expand...
Yes but won't those same games work exactly as well on a geforce 4 or higher?

I'm trying to understand if there is some technology that makes the 3 more desirable than 4+.
 
It's nothing to do with the tech, simply a collecting thing. There are people who want "one of everything", and there are less GF3 than GF4. So if you want one, you're paying more.
 
hunterjwizzard said:
Yes but won't those same games work exactly as well on a geforce 4 or higher?

I'm trying to understand if there is some technology that makes the 3 more desirable than 4+.
Click to expand...

There are some games that use rendering modes that work better on fixed function pipeline cards, and not so great on programmable/unified shader cards. One such example is Half-Life, it uses immediate mode rendering where draw commands are blasted to the frame buffer as fast as possible to render a frame, then blown away for a clean slate for the next frame. This is how games used to be programmed before the advent of programmable/unified shaders, which use a retained mode rendering method that is more efficient. In retained mode, a render list is created and only updated when it needs to be, saving tons of processing power and bandwidth vs the old brute force method.

GPUs with programmable/unified shaders hate immediate mode rendering and generally have performance penalties.
 
Some GF2 cards are faster than GF3 models except when you enable AA. GF4 cards other than the entry level 4200 are getting pricey as well.

If you collect old AGP video cards, you probably want the harder to find GF3 series so you pay more. I don't think I have many GF3 cards, at least a GF3TI model. Generally, I just collected a couple of each series and moved on unless they were super cheap (I should have hoarded more I guess). Most people just snag the last generation 8x AGP cards and call it a day but I like variety.
 
Plasma and GiG are right. Twenty years ago I got the Radeon 9800 and it was one heck of a video card. I never really got into Nvidia until the 1080 with the exception of the 6800 AGP. I still have a pair of 7850's which I ran in Crossfire on a 990 platform with the Bulldozer. I literally have a boxful of various AMD video cards. That's my story but can't help with the why of the GF3 other than some just want them. My collection goes back to the Oak VGA 640x480 that found its resting place in my 1000SX to my current RTX4090.
 
It's architecturally significant in the history of GPUs. Isn't that why most of us collect these old computer artifacts? It straddled the older fixed functional pipeline and the newer programmable shaders.
 
I've typically only found ATI cards and very early nvidia cards. The later PCI Express products tended to hold their value for many years after they had exited the market. (EG the 7950 GX2 or the 8800GTX)
Even these days I typically only find 9800 series cards in volume, partly because what I work on used them exclusively. I start getting much newer and the cards are either exponentially more expensive or are completely burned out.
 
resman said:
It's architecturally significant in the history of GPUs. Isn't that why most of us collect these old computer artifacts? It straddled the older fixed functional pipeline and the newer programmable shaders.
Click to expand...

THAT is what I was trying to work out, thank you for explaining it :)
 
hunterjwizzard said:
THAT is what I was trying to work out, thank you for explaining it :)
Click to expand...
Glad I was able to satisfy your question. As a little more insight into the way NVIDIA released their products back then, you'd see a new architecture every year or so followed by a tweaked version six months later taking advantage of process advances and some functionality updates. Design teams would leap frog each other for each new architecture. So there were:

NV1 => 2D windows accelerator/brain damaged textured quadratics/digital game controllers/MIDI/audio (let's forget about this one)
RIVA 128 / RIVA 128ZX => 2D window accelerator/3D textured triangle rasterizer
RIVA TNT / RIVA TNT2 + Multi-texture
GeForce / GeForce2 + H/W T&L
GeForce3 / GeForce4 + Programmable shaders
etc.
 
Some people are very interested in doing 'period correct' builds. As in, we'll choose a specific date and then try to use the 'best' components that were available on that date.

I also sometimes see claims that particular games run 'best' on some super old version of video drivers which can't be used with newer video cards.
 
Any game that runs on a GF3 is going to run even better on a GF4 (NV25, not the MX). "Period correct" cases back then didn't have windows, so you will never know the difference. 😛
 
My guess is for nostalgia reasons and bragging rights. If you have to compare whatever card you have versus a Voodoo 3, they're familiar with the Voodoo 3 part of the conversation.
 
Plasma said:
Any game that runs on a GF3 is going to run even better on a GF4 (NV25, not the MX). "Period correct" cases back then didn't have windows, so you will never know the difference. 😛
Click to expand...

You'd be wrong, there were many cases with windows in that era. I still have one, the Antec PlusView 1000 AMG. I even have the silly sling carry strap for it, but I've never used it because it's a bit ridiculous hauling around a 50-70 pound machine.

Antec PlusView 1000 AMG review - DVHARDWARE

www.dvhardware.net www.dvhardware.net

Case was released in 2003.

You have to remember that this was the era of colored CCFL tubes, basically it was the RGB of the era. Plenty of windowed and translucent cases to show off the eye vomit. I've had a few pass through my hands over the years, but I've gotten rid of all of them because they were just eye cancer.
 
GiGaBiTe said:
Case was released in 2003.
Click to expand...

The GeForce3 was released in February 2001 and was discontinued by mid-2002; the entire line was kaput the day their GeForce 4 replacements came out. (Unlike how, say, the GeForce 4MX lingered around for years.) Not to say there weren’t windowed cases in 2001, but my recollection was they were still pretty rare. So… let’s just say that if your deal with wanting a GeForce 3 is to make a perfectly period correct computer, case and all, the “window” where the label on the card is going to matter is pretty small, both literally and temporally.
 
Moondog said:
My guess is for nostalgia reasons and bragging rights. If you have to compare whatever card you have versus a Voodoo 3, they're familiar with the Voodoo 3 part of the conversation.
Click to expand...

Is there any comparison to a Voodoo3 that doesn’t conclude with: “but yeah, all things considered the Voodoo3 really was a bad joke and I have no idea why anyone wants one for anything besides wedging under a table leg these days.”

(I had a Voodoo3 so I’m allowed to say this. Yes, it had Glide for backwards compatibility, that was literally the only thing good about it.)
 
You must log in or register to reply here.
Back
Top