The Death of Superman, er, 3Dfx
Dec 19 '00 (Updated Dec 31 '00)
Well, I guess we all know that nVidia won this one. I have nothing against 3dfx product, I think generally they make servicable graphics cards.
This shows what happens when a company chooses to procrastinate rather than innovate. No one in their right mind would run a company trying to tell the market what the market wants, rather than listening to the market's opinion. This is what 3dfx did, and this is why they failed. Years from now, when we're all in the holodeck, us old farts will be saying "Hey, remember, 3dfx? they're the ones that started it all". Because of that, it is a shame to see them go. How could a company go from releasing a product that was probably 3 times as fast as anything on the market to being an utter failure? Also, where does this leave nVidia and the graphics market?
Flashback to Wolfenstien 3D and Doom. Commercial graphics chip maker S3 is the king of the hill, Diamond is one of the best video card brands, 4MB of video ram is alot, SVGA hasnt been around very long in unified form, and games are rarely played in any resolution higher than 640x480.
At this point in time, S3 and brands like Matrox have just begun to incorporate the most basic of 3D features in their cards, and due to the lack of RAM on these boards, what little 3D acceleration there is, is generally useless. These were the days when graphics cards were having a hard time with the Windows openGL screensavers. 16-bit game consoles like the Super NES have better graphics.
Enter the Voodoo, a 3D only add-on card, purchased only by the most insane of gamers due to its price. I remember thinking it would be nice to have one, but at the time, the CPU was king, and spending 1/5 of the value of your computer on a graphics card was a ridiculous idea. There were those with vision and money, and they bought it. And they played and they played, Doom, Quake... Until the Voodoo2 came out and destroyed everything in its path. An accelerator so powerful it can still be used in low resolution on the newest games out today. And here it was, accelerating DOOM. It sparked the 3D revolution in computer gaming, and for the first time since the early 80s, computer game graphics were definitely better than their console counterparts. The Voodoo2, 3dfx's hallmark card, if you will, set the standard, and stayed atop its throne as king of the video cards for a good 2 years.
There were rumblings from beneath. Other companies flirted with higher power 3D acceleration. nVidia had put together its first commercial product for the forgettable Sega Saturn, and had released the Riva to the computer market. 2D-3D combination boards were becoming more popular, and 3dfx was stuck with a 3D only solution. They jumped into the 2D-3D market without looking. Their product was basically a 2D board with a crippled Voodoo2. It was their chance to innovate, to crush the opposition, and instead they whimpered. It was the first chink in the mighty armor, and many people stung by the Voodoo Banshee never forgot. I personally, at the time, had to choose between the Banshee and the nVidia Riva TnT. I bought the TnT, and so did alot of other people.
nVidia was making a move toward 32-bit color in 3D games. Almost no games supported it, and the cards at the time hardly had the power to accelerate it, but the option was there none-the-less, and would be there in future product that actually could handle 32-bit. 3dfx insisted there was hardly any difference between 16-bit and 32-bit, and claimed that somehow their 16-bit was really 22-bit anyway. Very few people bought it. Although, there really wasnt much difference at the time because there simply werent any games out that took advantage of 32-bit color.
So 3dfx rested on their throne, releasing a weak series, the Voodoo3. Basically a good 2D card with a glorified Voodoo2 on-board. Still no 32-bit support, and something that is coming back to haunt Voodoo3 owners, it only supported textures up to 256x256 in resolution. It's main competition was the nVidia Riva TnT2, and this is where the whole 3dfx vs. nVidia arguement began. The specs for both were similar, and so was the performance. 3dfx's ace was Glide, nVidia's was 32-bit color. Glide was beginning to fade, 32-bit color beginning to rise, and that probably was the writing on the wall. The crusher was Quake 3. Maybe the first game designed with 32-bit color in mind, here was a game where the difference was evident between 16-bit and 32-bit color. After playing Quake3 in 32-bit, 16-bit actually became an annoyance, a distraction, and it was at that point that the world knew, 16-bit color had to go.
3dfx missed their product cycle that year and nVidia was developing at a torrid pace. nVidia released the GeForce, 3dfx released nothing. While the original SDRAM GeForce was not a revolution in performance, it was slightly faster than the TnT2 and the Voodoo3, and 3dfx didnt have an answer. For the first time ever, 3dfx was argueably in 2nd place. The DDR GeForce was next, and the mess was horrible. 3dfx body parts lay strewn across the landscape. Here was a card that was definitely faster than everything else on the market. The tables had turned, where 3dfx's Voodoo2 once laughed at nVidia's Riva128, nVidia was now doing the laughing, all the way to the bank.
It had been a time to scramble. 3dfx was struggling to get their Voodoo5 ready, and placed their hopes on the ill-timed and conceived anti-aliasing scheme rather than following nVidia's lead into hardware Transform & Lighting. Once again, 3dfx was dissing innovation, claiming there was no use for hardware T&L, the same way they did with 32-bit color. If released in a timely fashion, the Voodoo5 would have made the SDRAM GeForce and hardware T&L look bad, and would have at least been competitive with the DDR GeForce, but that didnt happen. By now nVidia was on the ball, and they broke 3dfx's back by rushing out the not-quite-ready GeForce2 ahead of the release of the Voodoo5. Driver and compatibility problems persisted for nVidia, and because of the rush, their anti-aliasing scheme was broken upon release. But the raw power of the GeForce2 was a factor the Voodoo5 couldn't escape. In almost every benchmark, the Voodoo5 cant compete, and to make matters worse, the GeForce2 is actually cheaper.
3dfx fans await the calvary, the return to glory, the horribly overbearing Voodoo5 6000 with its massive pcb, 4 processors, 128mb of ram, and expected $600 retail pricetag. They wait, and wait. nVidia releases their own ridiculously priced card, the GeForce2 Ultra, capable of accelerating Quake3 in 32-bit color, 1600x1200 resolution, at nearly 60fps. Unheard of power, a giant killer, a Voodoo5 6000 competitor. 3dfx brags that their video cards are the best selling on the market. What they don't mention is that there are a dozen or so companies making GeForce2 boards, and the top 2 alone are equaling Voodoo5 sales. Just as a nail in the coffin, ATi releases the respectable RadeOn, which manages to handily surpass the Voodoo5 in pretty much every 32-bit benchmark. The 3dfx products arent budget items, and they are slower than the leaders, there is now no real reason to buy one.
The company, bleeding money, decides to fold, and all of it's graphics card assets will probably be purchased by nVidia. nVidia's big problem at the moment is bandwidth. While hidden surface removal is one way around that, it cant be the total solution in the future. With this acquisition comes GigaPixel, which holds the key to nVidia's entry into tile-based rendering. Once tile-based rendering is intergrated into nVidia's scheme, the bandwidth problem should melt away. As you can see from the GeForce2, it is such a powerful processor there is no practical memory system that can tap it's full potential. nVidia's next product line is looking to be twice as powerful. Combine nVidia's great team of engineers with it's current product, and GigaPixel technology, and you have some very exciting possibilities in the future. Let us all hope that some other company, like ATi or STMicro can manage to keep up and provide at least some competition. If that can happen, the future of computer gaming looks intense.
Forward 6 months. Heavily T&L titles like Giants and Black&White are smash hits, T&L is finally vindicated, Wolfenstein and Doom are games shown to children as a display of how primitive games were when they were born. Guillemont, with its brilliant acquisition of the Hercules name has risen to the top of the video card market along with Creative; Diamond doesnt exist. 64mb of RAM on board is still trying to gain popular acceptance, 4mb is now a texture size. Games are routinely played in a resolution as high as the monitor can handle, 640x480 is the realm of consoles. At this point in time, nVidia and ATi both have full feature 3d accelerators with hardware TC&L and hidden surface removal, S3 is gone, and Matrox concentrates on the business market. Computers make the once touted Playstation 2 look like a joke, the upcoming Xbox, featuring an nVidia chipset, is already obsolete. Almost every game on the market is accelerated at 60fps in resolutions as high as monitors will go. The sky is the limit.
game, point, match, nVidia
 |
|
|
|
|
|
|
Epinions.com ID:
clburdette
|
|
Member: Curtis Burdette
Location: california
Reviews written: 46
Trusted by: 19 members
About Me:
question everything
|
|
|
© 1999-2009 Shopping.com, Inc.