VOGONS


Could Glide Have Triumphed?

Topic actions

First post, by kaiser77_1982

User metadata
Rank Newbie
Rank
Newbie

In the history of video games and graphics technology, Glide holds a prominent place as one of the great innovations in the graphics card industry. It was the first major proprietary API specifically designed to leverage the capabilities of 3dfx's Voodoo graphics cards. Glide was characterized by being a simple, fast, and efficient API for 3D graphics during the industry's boom and the onset of what we call the "Great Graphics Purge."

However, like many phenomena that rise rapidly, Glide fell just as quickly—almost like a rock star. Its golden period spanned from 1996 to 1998, followed by a decline at the end of 1999, and its end in 2000. Although some titles continued to use it until 2003, by 1998, it had ceased to be the standard.

Reasons for Its Fall

We can summarize the reasons for its decline into four main points:

  1. 3dfx's self-destruction: A topic we'll explore in another article due to its complexity.
  2. Exclusivity with 3dfx cards: This limited its reach and viability.
  3. The arrival of DirectX: An API that worked with any graphics card, backed by Microsoft—a company with an almost monopolistic hold on computer operating systems.
  4. Competition from Nvidia: While 3dfx offered superior quality, Nvidia began offering similar products at competitive prices, reaching a broader audience.

Glide's success depended directly on 3dfx's ability to dominate the market, ensuring developers worked exclusively with its API. Before 1998, there were no truly decent graphics cards for gaming. Matrox was focused on other sectors, S3 struggled with the limitations of the ViRGE, Rendition failed to take off, ATI was concentrating on other projects, and Nvidia was just starting to gain relevance with the Riva128 in the OEM market. If you wanted to game, you needed a Voodoo—period.

After 1998, everything changed. Nvidia launched the TNT and TNT2, which became direct competitors to 3dfx. ATI improved with the Rage 128, the Intel i740 emerged as a budget-friendly option, and S3 had its brief moment of glory with the Savage. With such fierce competition, maintaining leadership was almost impossible. 3dfx's only hope was to change the game with the Rampage, but failing to do so, Nvidia took the lead with the GeForce 256, supported by Microsoft.

Outside of hardware, the true deathblow for Glide was DirectX, an API that did not rely on specific hardware and was developed by Microsoft—a company with nearly unlimited resources. Understanding why DirectX was inevitably superior is as straightforward as analyzing the Eastern Front in World War II: Stalin always had more resources and men than Hitler. No matter what the latter deployed, Stalin always had more to counter it.

Microsoft invested massively in DirectX, even paying developers to create games compatible with their API, covering salaries and expenses for one or two years. This represented a huge cost saving for companies—something that 3dfx, a much smaller company, could never compete with. Even if the 3dfx merger with STB had been a success (a topic we'll tackle in another article), it would never have had the muscle to stand up to Microsoft.

In fact, even if 3dfx had succeeded in all its goals, it’s likely that Microsoft would have faced it head-on, as they did with Netscape or Sony in later years. This confrontation was hinted at in details like 3dfx's complete lack of support for Microsoft’s Talisman project.

The exclusive reliance on their own graphics cards was a burden for Glide due to its limitations, including:

  • The inability to handle higher resolutions in later years.
  • Lack of support for 32-bit color and advanced effects.
  • The absence of shaders, which were essential for graphical development in the 2000s.

While Glide was revolutionary in its time and deserves recognition for its merits, it had no place in a future dominated by broader and more flexible standards. Glide was a product of its time—brilliant but fleeting. It leaves us a legacy of innovation, great games, and highly collectible artifacts, but also a lesson on how open standards and adaptability are key to surviving in the tech industry.

That’s all for now. Happy New Year, and thanks to read until here.

//-------//

I'm leaving you with a reflection on Glide and 3dfx. I would like to know your opinion.

PS. I have the original text in Spanish in my newsletter, if you see any mistake, let me know and I'll correct it.

Reply 1 of 22, by RetroPCCupboard

User metadata
Rank Member
Rank
Member

3DFX could have triumphed if they innovated and kept pace with the competition. But they would have had to abandon glide and excel at DirectX. From what I have seen 3DFX cards are rather poor in directx mode.

Reply 2 of 22, by DrAnthony

User metadata
Rank Member
Rank
Member

Glide was only ever going to be a bandaid solution. It was super important in the early days before DirextX was mature enough to be useful and OpenGL implementations were lacking. It was too tied to 3Dfx hardware to be made an open standard nor would it have made financial sense to do so. It was the dominant API until just overnight, it wasn't. 3Dfx could have easily made it through the industry purge would better management, but that's been discussed a million times over and I don't want to clog a thread up with my opinions that really only matter to myself.

Reply 3 of 22, by the3dfxdude

User metadata
Rank Oldbie
Rank
Oldbie

In 1997, a 3D accelerator card was not absolutely required for gaming. Many games and engines released that year still supported software rendering. And then too, the over proliferation of 3D game attempts starting that year through the late nineties did kind of put a stigma on 3D gaming across the entire gaming spectrum that 3d games produced rather poor looking games. But you have to couple that 3D technology was not yet standardized, and DirectX wasn't either in 1997, and 3D technology really was limited in what it could do on the PC. Frankly, all this coupled together is why things like S3 Virge was able to get as far as it did with games up through 1997, which was hardly that good for DirectX as an option too, which is why glide still kept getting the attention, since it was ahead on technology.

In 1998, 3dfx was still selling lots and lots, despite its exclusivity. Had they came up with a real plan for what was next, they could have pivoted. This was when they started getting bad direction though.

Into 1999, the STB deal was done. That killed 3dfx period. The entire management knew it. They spent way more money than they were really making. So through 1999, they really started dragging and just simple failed to design a proper next gen chip. When 2000 rolled around, they were done.

The Geforce 256, while a really good design, and barely pre-empted the Radeon release, wasn't necessarily a clear cut 3dfx killer. Had 3dfx made good decisions, I think it's likely anyone of the other top competitors would have also just imploded in 2000 (there were lots of deaths in the 3d market already, as you know). ATI could have been the one to survive, or die. Matrox was already on shaky ground, and maybe wouldn't been able to sell any of their GPU at all. Nvidia had made mistakes too already. So I really just take the situation of 2000 as Nvidia just happened to be the next one in line to be leader at the moment 3dfx imploded.

Frankly everyone was supporting, including 3dfx, the DirectX API by 2000. I don't think Microsoft was rooting for anyone, nor really was trying to corner the graphics API market and were initially were going to let OpenGL or then the Talisman project be the lead. Except they got really good advice by their internal teams to make DX the 3D platform, done right, and what all shown from that is their engineers were right. It didn't happen overnight, showing the iterations of DX versions, and really, the thing about Microsoft, is they never let up and keep trying until something works. Everyone knew that they were going to end up following what Microsoft was doing on PC. So the standard wasn't certain in '97 since the games weren't there, and the manufacturers weren't there, it really was just a matter of putting a team at 3dfx on designing the next GPU to perform well on DX (and keep glide in play if they wanted) in '98 when they were starting to get a little behind on that (and they have an excuse, they did have the lead). Well they didn't make a proper maximally resourced team for that, and instead spent money on manufacturing in-house and killed their entire business on that decision.

Reply 4 of 22, by digger

User metadata
Rank Oldbie
Rank
Oldbie

In the end, vendor-neutral standards always win out. So in the long term, Glide wouldn't have won, simply on account of the fact that it was the proprietary API of a specific graphics accelerator manufacturer, and one that was close to the metal too. And wasn't it to some extent based on OpenGL to begin with?

What's really tragic is that there was a missed opportunity to establish OpenGL as the dominant 3D acceleration standard for games, before Microsoft got the chance to catch up with Direct3D and corner the market. That way, a standard that was both hardware-neutral and OS-neutral would have won out, benefiting everyone, except for maybe Microsoft.

With the realease of GLQuake, followed by many other games based on the same engine, there was a short window during which "MiniGL" became something of a "defacto" standard. Even a number of games that came out at the time that weren't based on the Quake engine appeared to be gravitating towards it. The reason why MiniGL initially started to gain some traction was because the earlier versions of Direct3D were a mess and grossly inadequate. But Microsoft kept watching the market and improving it.

Unfortunately, id software never released a formal specification on exactly what subset of OpenGL they were targeting with MiniGL, which would have formed a guideline for graphics card manufacturers to support in their drivers. So this lack of clarity inhibited broader adoption of MiniGL. And full-fledged OpenGL drivers were too complex to develop for gaming-oriented 3d accelerators at the time, not to mention too much tailored towards professional use such as CAD and such. So once Direct3D became mature enough, that became the industry standard instead. It was independent of hardware vendor, yes, but unfortunately tied Windows.

Just an RFC or some whitepaper with John Carmack's name under it would probably have been enough for more graphics card manufacturers and other game developers to latch onto it and help it grow as a standard. Alas...

Reply 5 of 22, by Namrok

User metadata
Rank Oldbie
Rank
Oldbie

IMHO, there is room for proprietary, well not an entire api, but api extensions. I mean Nvidia is killing it lately with features like DLSS, Ray Reconstruction and Frame Generation. I think their basic ray tracing API extensions were standardized in DX12 though.

The problem with Glide was that is was lean and mean, and absolutely perfect for when 3d games were heavily CPU constrained. When AMD and Intel went racing for the 1000 Mhz barrier and beyond at a breakneck speed, having highly optimized drivers with the bare minimum of features to get games running with as many FPS as possible just wasn't the selling point it may have been on a PII 300. There was more wiggle room fuller APIs, and drivers with more CPU overhead if it meant more features. This just wasn't in 3DFx's ethos IMHO, and they were never going to keep up. Nvidia was pioneering capabilities with the TNT, and by the time the Geforce 3 came out 3Dfx had an enormous technical debt.

Win95/DOS 7.1 - P233 MMX (@2.5 x 100 FSB), Diamond Viper V330 AGP, SB16 CT2800
Win98 - K6-2+ 500, GF2 MX, SB AWE 64 CT4500, SBLive CT4780
Win98 - Pentium III 1000, GF2 GTS, SBLive CT4760
WinXP - Athlon 64 3200+, GF 7800 GS, Audigy 2 ZS

Reply 8 of 22, by Errius

User metadata
Rank l33t
Rank
l33t

In the mid-1990s only super expensive CAD cards had full OpenGL support.

Glide was a quick and dirty hack of OGL designed to run on much cheaper hardware.

It had a brief moment in the sun in the late 1990s, but that passed as soon as cheap cards offering full OGL compatibility appeared (and DX overcame its initial troubles)

Is this too much voodoo?

Reply 9 of 22, by RetroPCCupboard

User metadata
Rank Member
Rank
Member

I personally missed most of the 3DFX train. I got a Matrox Mystique first, which was just before the Voodoo 1 released. Then I went nvIdia. Riva 128, TNT, TNT2, Geforce, Geforce 2 etc. I think I had a brief period with a Voodoo 2000 between the TNT and TNT2. My first ATI card was the All in Wonder 9700 Pro. Loved that card. Then next went to nVidia 6800 then a laptop with 7800 GTX, then a desktop with 8800 GTS 512 SLI.

I am not sure what my reasonings were for mostly ignoring glide. Perhaps partly because I missed the Voodoo 1 due to having bought a different card just before. Probably the Voodoo 2 didn't appeal because I didn't want to buy a 2D and 3D card? Anyone else do similar?

Reply 10 of 22, by Joseph_Joestar

User metadata
Rank l33t++
Rank
l33t++
RetroPCCupboard wrote on 2025-01-03, 08:19:

Anyone else do similar?

I went from an S3 Trio64V+ (pure 2D card) to a Trident Blade 3D and then to a TNT2. I briefly considered getting a Voodoo 3 instead of the TNT2 but was dissuaded by popular PC magazines of the time. I came to regret this later, as I was mostly playing Unreal Tournament and Diablo 2 back then, which would have worked much better in Glide mode.

Later on, I got a GeForce 2 MX400 and then a GeForce 3 Ti200, but curiosity about Glide made me buy a second hand Voodoo 3 sometime in 2002. It was around 30 EUR back then, and I still have that card to this day.

As for Glide, I like that it was "close to the hardware" compared to DirectX and OpenGL. The S3 Metal API was similar in that regard, but not as popular. I actually prefer lighter APIs like that, compared to the bloat that is the current DirectX.

PC#1: Pentium MMX 166 / Soyo SY-5BT / S3 Trio64V+ / Voodoo1 / YMF719 / AWE64 Gold / SC-155
PC#2: AthlonXP 2100+ / ECS K7VTA3 / Voodoo3 / Audigy2 / Vortex2
PC#3: Core 2 Duo E8600 / Foxconn P35AX-S / X800 / Audigy2 ZS
PC#4: i5-3570K / MSI Z77A-G43 / GTX 970 / X-Fi

Reply 11 of 22, by RetroPCCupboard

User metadata
Rank Member
Rank
Member
Joseph_Joestar wrote on 2025-01-03, 08:30:

I briefly considered getting a Voodoo 3 instead of the TNT2 but was dissuaded by popular PC magazines of the time.

Yes, those were the days. Prior to the Internet taking over, those magazines were great for getting buying advice. And also a never-ending source of great game demos to play. I am sure their opinions, and my limited budget at the time, played a key role in my purchasing decisions.

Never having tried glide I didn't really know what I was missing. Whereas now, retrospectively, I can see it is superior to the others in terms of smoothness and early 3D game support. But, honestly, until the Voodoo 3 came along, the image quality was very poor. By then it was too late.

Reply 12 of 22, by gerry

User metadata
Rank Oldbie
Rank
Oldbie
kaiser77_1982 wrote on 2025-01-02, 21:03:
We can summarize the reasons for its decline into four main points: […]
Show full quote

We can summarize the reasons for its decline into four main points:

  1. 3dfx's self-destruction: A topic we'll explore in another article due to its complexity.
  2. Exclusivity with 3dfx cards: This limited its reach and viability.
  3. The arrival of DirectX: An API that worked with any graphics card, backed by Microsoft—a company with an almost monopolistic hold on computer operating systems.
  4. Competition from Nvidia: While 3dfx offered superior quality, Nvidia began offering similar products at competitive prices, reaching a broader audience.

a good summary and so, no - i dont think it could have triumphed

i am one who went from 2d cards to a tnt2 completely missing the hype. those old 2d cards did all the dos '3d' and all the windows 2d, isometric and so on games that i wanted. i just didn't see what the hype was about. even on youtube videos show casing voodoo i still think its just ok, i mean its better than nothing obviously but really is it *so* great that it deserves all the retro-hype. i clearly just missed it, or i'm a graphics philistine. i'm sure in those short few years it was pretty cool for those who bought it.

Reply 13 of 22, by RetroPCCupboard

User metadata
Rank Member
Rank
Member
gerry wrote on 2025-01-03, 09:50:

but really is it *so* great that it deserves all the retro-hype. i clearly just missed it, or i'm a graphics philistine. i'm sure in those short few years it was pretty cool for those who bought it.

Retrospectively, I'd say, yes. I can understand why it was a gamechanger for those that had a Voodoo 1 or Voodoo 2 back in the day. But, honestly, in some ways I prefer the crisp look of software rendering over the blurry Voodoo look. But the increase in framerate in games that supported glide is undeniable.

Reply 14 of 22, by clb

User metadata
Rank Oldbie
Rank
Oldbie
kaiser77_1982 wrote on 2025-01-02, 21:03:
Reasons for Its Fall […]
Show full quote

Reasons for Its Fall

We can summarize the reasons for its decline into four main points:

  1. 3dfx's self-destruction: A topic we'll explore in another article due to its complexity.
  2. Exclusivity with 3dfx cards: This limited its reach and viability.
  3. The arrival of DirectX: An API that worked with any graphics card, backed by Microsoft—a company with an almost monopolistic hold on computer operating systems.
  4. Competition from Nvidia: While 3dfx offered superior quality, Nvidia began offering similar products at competitive prices, reaching a broader audience.

Glide was a proprietary solution. In a blooming market those never thrive for long.

While I hate Microsoft as much as the next guy, Microsoft's "monopolistic hold" doesn't apply in this case: Microsoft collaborated with hardware vendors to make sure they had DirectX compatible drivers. They actively standardized. In contrast, 3dfx held their API closed, and wasn't open for competitors to also build Glide support. Not hard to see which one is going to do better on the long run.

Compare to Commodore Amiga vs IBM PC. One was a closed ecosystem, and the other generally expandable with common off the shelf components. IBM PC was rather crap and expensive when it started compared to Amiga, but it didn't matter, since IBM (maybe somewhat accidentally) enabled an ecosystem to thrive. Commodore didn't, and neither did 3dfx.

Reply 15 of 22, by BinaryDemon

User metadata
Rank Oldbie
Rank
Oldbie

Maybe - If 3dfx had a crystal ball and immediately cheaply licensed GLIDE API to any and all videocard manufacturers. I went with "cheaply licensed" instead of "open-sourced" because 3dfx clearly still needed some additional funding to remain competitive with hardware development. Ultimately unless Microsoft licensed Glide for bundle with Windows 9.x, it was probably still going to be a dead end.

Reply 16 of 22, by zb10948

User metadata
Rank Member
Rank
Member
gerry wrote on 2025-01-03, 09:50:

i am one who went from 2d cards to a tnt2 completely missing the hype. those old 2d cards did all the dos '3d' and all the windows 2d, isometric and so on games that i wanted. i just didn't see what the hype was about. even on youtube videos show casing voodoo i still think its just ok, i mean its better than nothing obviously but really is it *so* great that it deserves all the retro-hype. i clearly just missed it, or i'm a graphics philistine. i'm sure in those short few years it was pretty cool for those who bought it.

Hey I'm the 2D guy too.I did not follow early 3D nor know/heard about Voodoo until Voodoo 2 age.

I did follow local mags - not the gamer ones though but general PC related. People weren't that much into Voodoo, I think it was too expensive for our market. Stuff was expensive back when there were import tarrifs and such, people did not upgrade PCs easily in early 90s.

Nowadays if you judge per YT seems like everyone had or wanted Voodoo back in the day, and played Tomb Raider. A lot of people I know did not play that kind of games - 3rd person 3D was new, not everyone was impressed. And people I've seen play Tomb Raider IRL just played it normally without Voodoo.

Maybe I'd be impressed if I'd seen one in action at that impressionable age.

Reply 17 of 22, by RandomStranger

User metadata
Rank Oldbie
Rank
Oldbie

As said, no chance. Proprietary APIs have a really hard time succeeding long term. And it's got more and more difficult as time went. Nvidia tried that approach a lot in the past 20 years (PhysX, Hairworks, G-Sync, etc). The best they could achieve is a head start, then the industry moved over to more open options. Unlike 3DFX, Nvidia was always smart enough not to be over-reliant on their own proprietary solution and keep up with the competition on the open standards.

sreq.png retrogamer-s.png

Reply 18 of 22, by darry

User metadata
Rank l33t++
Rank
l33t++

Mostly echoing what other have said, for Glide AND 3dfx to triumph over other 3D APIs while still remaining proprietary to 3dfx AND not even licensed to other hardware vendors would have required that

a) Glide be the only option
OR
b) Glide provide functionality that others options did not (features, image quality)
OR
c) 3dFx kept dominating in performance over other vendors to an extreme extent or at least very significant extent

Scenario a) was never really the case (OpenGL was already there and Direct3D was becoming useful as it improved). While Glide was arguably the only high performance consumer solution at the beginning, that did not last.

Scenario b) did not apply as OpenGL and Direct3D features and rendering quality on other hardware eventually matched and even exceeded Glide on 3dfx hardware

Scenario c) not longer applied when other hardware vendors caught up on the performance front.

So, to answer your question, IMHO, there's nothing that 3dfx could have done to hang on to a) . Scenario b) would have been tough to pull off (and 3dfx did try). Scenario c) would have been tough as well.

If 3dfx had opened up Glide at the right time, the API might have triumphed, at least for a while, but it would not have helped 3dfx, which would have been giving up a competitive advantage (at the time).

Arguably with the benefit of hindsight, by the Voodoo 2 era, it was clear that while Glide might survive it was not going to keep market share. And 3dfx might have survived and thrived IF it had remained competitive on performance, features and quality past the Voodoo 3 era.

Reply 19 of 22, by Jo22

User metadata
Rank l33t++
Rank
l33t++
zb10948 wrote on 2025-01-03, 12:11:
Hey I'm the 2D guy too.I did not follow early 3D nor know/heard about Voodoo until Voodoo 2 age. […]
Show full quote
gerry wrote on 2025-01-03, 09:50:

i am one who went from 2d cards to a tnt2 completely missing the hype. those old 2d cards did all the dos '3d' and all the windows 2d, isometric and so on games that i wanted. i just didn't see what the hype was about. even on youtube videos show casing voodoo i still think its just ok, i mean its better than nothing obviously but really is it *so* great that it deserves all the retro-hype. i clearly just missed it, or i'm a graphics philistine. i'm sure in those short few years it was pretty cool for those who bought it.

Hey I'm the 2D guy too.I did not follow early 3D nor know/heard about Voodoo until Voodoo 2 age.

I did follow local mags - not the gamer ones though but general PC related. People weren't that much into Voodoo, I think it was too expensive for our market. Stuff was expensive back when there were import tarrifs and such, people did not upgrade PCs easily in early 90s.

Nowadays if you judge per YT seems like everyone had or wanted Voodoo back in the day, and played Tomb Raider. A lot of people I know did not play that kind of games - 3rd person 3D was new, not everyone was impressed. And people I've seen play Tomb Raider IRL just played it normally without Voodoo.

Maybe I'd be impressed if I'd seen one in action at that impressionable age.

Hi, I feel similar. I had been a 2D person, too, and I think I've missed out on these things.

The magazines I read wrote about PC hardware, software and online services, telecommunications.
Back then I wasn't even really aware of DirectX, but I knew of WinG, GDI and DCI.

Probably due to multimedia hype and also because I had Windows 3.1 and my dad had Windows 95 RTM, so DirectX wasn't preinstalled.
And DirectX runtime (fat, how many floppies?) had been bundled with commercial games only. And shipped on cover disks.

At the time, a friend had an N64 with Mario Kart 64. But the graphics were ugly compared to what I had seen before.
Sonic 2 on Mega Drive, Super Mario Kart/Star Wing/F-Zero/Pilot Wings etc on Super NES. They all looked more advanced to me.

Speaking of the latter, I remember how we played Donkey Kong Country on it.
That game was a very impressive "3D" game at the time and was prettier than anything I've seen on N64.

Or on PC (in the news of computer shows), for the matter.
That's why I did prefer to stick to 2D desktop games on Windows,
which had been made by bed room programmers and small companies.
They ran nicely in 640x480 resolution, at least.

That being said, I think memory limitations were some reason why early 3D was so ugly.
The N64 needed the Expansion Pak (4MB) for a total of 8 MB of RAM.
Without it, the games ran below 640x400 resolution, which was ugly.
Especially if textures had been applied.
Then games looked like as if tattered magazine pages from trash can had been glued to them.

Same as with DOS games in 320x200. Ugly as an Xenomorph.
Unless being blurred by a CRT (N64 shipped with RF adapter for good reason).

The late, unreleased game "Dinosaur Planet" on N64 convinced me that a fully maxed out 3D console can indeed provide acceptable visuals.
Lylat Wars on same console was the ugliest thing I've ever seen, by contrast.
No wonder Star Fox 2 hadn't been released. The Super NES version was much prettier.

Sorry if that's a bit off-topic, but I didn't have an 3dfx card at the time so I can't share any experience.
I've rather played games like Need for Speed II SE via software rendering on Windows 98 by late 90s/early 2000s.

I was so happy and proud that my on-board graphics card had been a PCI-based GUI accelerator with DirectDraw support! Yay! 🥳

At the time, I was still being used to dumb ISA graphics cards and this was quite an improvement.
The 2D chip had the ability to scale in hardware, provide video overlay, do colour space conversion and so on.
Things from the multimedia area.

To my defense, though, the press did tell us that 3D accelerators are just a fad.
My father and me believed that SIMDs like MMX are the future, since they would allow for DSP applications (digital signal processing).
Such as implementation of modems in software (Win Modems, soft modems).
Which was much more about then just graphics.

And strictly speaking, that thinking at the time was not wrong I think.
Graphics cards would be superflous if x86 processors had corresponding DSP or SIMD capabilities built-in.

But modern world is doing other way round now.
We have graphics cards, GPUs, that become more and more CPU-like.

de-typoed.

"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel

//My video channel//