It would have put less pressure on the company, but might have taken out some of the 3rd party partners instead. 3DFX would have needed to drastically improved beyond what they offered. NVIDIA was firing on all cylinders, and 3DFX just wasn't keeping up in terms of features.

3DFX dominated early, but then they just only did incremental improvements while NVIDIA was adding new features and doubling their performance every nine months or so. NVIDIA on the other hand had a failure with the NV1, and their deal with Sega on the NV2 ended up not being able to hit its goals. Then they found success with the Riva 128 and kept the pressure up.

Not much would have changed unless 3dfx managed to release the voodoo4 6-12 months earlier and if 3dfx had a change of design philosophy with the rampage chip. Those were the things that doomed the company to follow Trident, S3 and Cirrus. The STB acquisition made the cash flow worse, but didn’t change the general trend.

3dfx would still be around and your goto destination for the most sophisticated AI solutions ....

Nothing much, because the fall of 3dfx wasn't about any single decision the company made. It was gradually losing the edge they enjoyed over competitors in V1 and V2 glory days. Glide was going awat fast and competitors released improved designs and 3dfx couldn't keep up.

That is not to say that STB didn't matter, it surely put the company in financially much worse situation and most likely hastened the collapse considerably.

Aui wrote on 2026-02-23, 01:49:

3dfx would still be around and your goto destination for the most sophisticated AI solutions ....

They would play games on it...

Such a thread definitely already exists on this forum.
In my opinion, a more interesting plot for an alternative history would be: what if VIA hadn't acquired S3, which at the time was the third-largest seller with a significant market share and promising developments.

what if PowerVR SG wasn't kneecapped by the NEC ghosting deal and actually got a PVR2 card out by August 98

If 3DFX prevailed we would now probably have 15th gen Voodoo gpus and Glide 15.x 3D API.

douglar wrote on 2026-02-23, 01:36:

Not much would have changed unless 3dfx managed to release the voodoo4 6-12 months earlier and if 3dfx had a change of design philosophy with the rampage chip. Those were the things that doomed the company to follow Trident, S3 and Cirrus. The STB acquisition made the cash flow worse, but didn’t change the general trend.

I Often Imagine in my head that a Voodoo 3 series line of graphics cards would have been released earlier Maybe I Guess? But I Could Be Wrong Here also I Often Imagine that the 3dfx graphics cards in this timeline probably would have been cheaper since 3dfx does not have to make and sell their own 3dfx their 3rd party board manufacturers would have Intense Partner Rivalry Over Pricing for sure but personally I Think Option would be to just improve on the technology for these cards to kept up with nvidia and ATI line of graphics card rather then lower price

3DFX wasnt serious about engineering, they leisurely progressed in small steps while Nvidia was leaping every 6-9 months:

Riva 128 (April 1997) to TNT (June 15, 1998) 14 months
TNT2 (March 15, 1999) 8 month
GF256 (October 11, 1999) 7 months
GF2 (April 26, 2000) 6 months
3dfx dies here
GF3 (February 27, 2001) 9 months
GF4 (February 6, 2002) 12 months
FX (March 2003) 13 months, etc ...

Afaik Voodoo3 worked exactly like Voodoo2 and Voodoo1 before it. Every single triangle was internally rendered, textured and shaded, no early Z, no Z-compression, not even skipping shading before hitting the Z-buffer. Of course no Texture Compression, no AGP texturing, textured limited to 256x256, framebuffer still locked to 16bit with dithering despite 16MB of ram on the card making arguments about saving space moot.

Sadly 3Dfx deserved to die 🙁

rasz_pl wrote on 2026-03-01, 14:10:

3DFX wasnt serious about engineering, they leisurely progressed in small steps while Nvidia was leaping every 6-9 months: […]
Show full quote

3DFX wasnt serious about engineering, they leisurely progressed in small steps while Nvidia was leaping every 6-9 months:

Riva 128 (April 1997) to TNT (June 15, 1998) 14 months
TNT2 (March 15, 1999) 8 month
GF256 (October 11, 1999) 7 months
GF2 (April 26, 2000) 6 months
3dfx dies here
GF3 (February 27, 2001) 9 months
GF4 (February 6, 2002) 12 months
FX (March 2003) 13 months, etc ...

Afaik Voodoo3 worked exactly like Voodoo2 and Voodoo1 before it. Every single triangle was internally rendered, textured and shaded, no early Z, no Z-compression, not even skipping shading before hitting the Z-buffer. Of course no Texture Compression, no AGP texturing, textured limited to 256x256, framebuffer still locked to 16bit with dithering despite 16MB of ram on the card making arguments about saving space moot.

Sadly 3Dfx deserved to die 🙁

Yeah you do have an good point here

I think they succeeded at first because they saw a consumer need and was able to release a high performance product to satisfy that need before anyone else did. But, once they had competition, the writing was on the wall. They just did incremental changes from the original, whilst the competition was making huge progress every 12 months or so

You could say ATI did that with Rage also, but they also had their skunkworks Radeon to pull out of their ass when the Rage ran out of scalability.

So that's where Rampage SHOULD have been, a late 99, early 00 release, leave the V5 to duke it out with the Rage Fury Maxx and take on the new gen with Rampage.

It did not seem they were as plugged in to the DX dev pipeline as other makers, which those had to do as a lifeline early on with Glide dominating, whereas less commitment to it from 3DFX bit them in the ass later.

BitWrangler wrote on 2026-03-01, 15:19:

You could say ATI did that with Rage also, but they also had their skunkworks Radeon to pull out of their ass when the Rage ran out of scalability.

Initial R100 Radeon was good and R200 better, but it became great with R300 designed by former Silicon Graphics engineers from ArtX. ATI and AMD both have a history of having asses saved by buying their way out of trouble. For AMD it was NexGen becoming K6. Sadly Management learned nothing from this and kept underfunding R&D hoping to keep repeating this miracle, results were failed investments and multiple write-offs like Geode, Alchemy, SeaMicro and Nitero. Also sold Imageon for mere $65mil just as phone market was getting hot.

to answer the specific question : How Would PC Gaming Be Different

it wouldn't really, game developers would have adapted to whatever was out there and likely we would still have pretty much the same gaming experiences.

I don't really share the enthusiasm for 3dfx / voodoo stuff though, when fans enthuse about it and about games i must assume it looked like breaking new ground to them for that sliver of time it was ahead of others and i must have been playing doom, duke3d and 2d games too much to appreciate it 😀

gerry wrote on 2026-03-02, 08:53:

to answer the specific question : How Would PC Gaming Be Different

it wouldn't really, game developers would have adapted to whatever was out there and likely we would still have pretty much the same gaming experiences.

I don't really share the enthusiasm for 3dfx / voodoo stuff though, when fans enthuse about it and about games i must assume it looked like breaking new ground to them for that sliver of time it was ahead of others and i must have been playing doom, duke3d and 2d games too much to appreciate it 😀

GLiDE compared to the alternatives at the time was a major leap. It wasn't just the interface or how it was programmed. Watch some videos going over titles and there was just no comparison. The original Voodoo was unchallenged for almost a year until the Riva 128 was released, but then the Voodoo 2 came out. The TnT followed, but didn't quite top the Voodoo 2 in performance. And the Voodoo 3 and TnT 2 traded blows nicely.

The early versions of Direct3D were nothing like we had today. And different hardware could have rendering issues. But if you had a Voodoo, you just activated GLiDE instead and had a perfect experience which at times had more complex effects than what was offered while running a game in D3D instead.

They had a good run from fall 1996 until fall of 1999. While only three years when products were coming out every few months that could double performance, it felt a lot longer.

IMO GLQuake releasing just over three months after Quake first came out is what really caught gamers' attention. PCs that struggle at 320x200 were running the game at 640x480 with enhanced graphics. I had a similar experience with Tomb Raider. I couldn't run it filling the screen, switch over to the GLiDE version, and it's smoother and rendered at a higher resolution.

To continue its dominance, 3dfx had to release Voodoo 2 in 1997. This reflects the increased performance between the two models. The first chip already supported up to 8 MB of memory and SLI. There was too little new in the second chip to delay release for so long. Voodoo 3 should have been released in 1998, and Voodoo 4-5 in 1999. In this case, 3dfx could compete for leadership.
Without these conditions, they somehow got involved in a fight with Nvidia, ATI, S3, Matrox, Intel with dubious chances of winning after 2000 without a decisive advantage. It wasn't a matter of buying STB.

Before Nvidia going Super Saiyan no one was taping out big ASICs in under two years.

Console money...

3Dfx got stiffed by Sega over the Dreamcast graphics, not only did they not make any money because their chip didn't go into the Dreamcast, Sega seems to have used the deal to gain all console rights to Voodoo Graphics architecture, leaving 3Dfx unable to market it elsewhere. Seems no coincidence that Rampage project launched right after this debacle, they needed ALL NEW tech, or they would be buying yachts for Sega's lawyers.

ATi on the other hand got a wad from Nintendo for the Flipper going into the gamecube, and "flipped" that tech and team into 9x00 series, so benefitted by money and tech and team experience.

Likewise, nVidia got an advance of 200M in 1999 for the xbox graphics, basically MS paid for their R&D on the Geforce 3. Or you could say, they put that money in their back pocket for the GPU they were developing anyway, and grabbed it out again when they got the chance to buy 3Dfx.

Anyway, what was a huge net benefit to two competitors was a loss to 3Dfx, which had the first "go around", so did they watch that, learn from it, and make better deals? Was 3Dfx the price for not having all GPU tech locked up in the console market and unavailable for desktop?