Founded in 1994 by Ross Smith, Gary Tarolli and Scott Sellers (all SGI alumni) with backing from Gordie Campbell's TechFarm, 3dfx released its Voodoo Graphics chip in 1996. The company only manufactured the chips and some reference boards, and initially did not sell any product to consumers; rather, it acted as an OEM supplier for graphics card companies, which designed, manufactured, marketed, and sold their own graphics cards including the Voodoo.
3dfx gained fame due to their great success within the arcade market. At the time, arcades were a very visible place to go visit and see the latest in 3D gaming and technology. The first arcade machine that 3Dfx Voodoo Graphics hardware was used in was called ICE Home Run Derby, a game released in 1996. Later that year they were featured in more popular titles, such as Atari's San Francisco Rush and Wayne Gretzky's 3D Hockey. 3Dfx received a lot of focus from the media because of the obvious graphical prowess of these titles, and that new game consoles such as Nintendo 64, Sony PlayStation, and Sega Saturn would be showcases for similar next-generation graphics.
After a fortuitous drop in EDO DRAM prices due to the volatile DRAM market, Voodoo Graphics cards became feasible for the consumer PC market. The Voodoo 1, as the Voodoo Graphics would be later known, was notable for its lack of an onboard VGA controller. As such, a Voodoo-equipped PC still required a separate VGA graphics card, meaning it was very expensive to have both 3D and 2D acceleration. The Voodoo 1 occupied a separate PCI slot and only engaged when the host PC ran a 3D game that had been programmed to use the card. A pass-through VGA cable daisy-chained the VGA card to the Voodoo 1, which was itself connected to the monitor. Although this was a cumbersome arrangement that somewhat hurt the analog signal quality of the separate 2D card, PC gamers were willing to put up with it to gain what was, at the time, the best in 3D graphics.
The Voodoo 1's main competitors were cards from PowerVR and Rendition. PowerVR produced a similar 3D-only add-on card with capable 3D support, although it was not comparable to Voodoo Graphics in either image quality or performance. 3dfx saw intense competition in the market from cards that offered the combination of 2D and 3D acceleration. While these cards, such as Matrox Mystique, S3 ViRGE, and ATI 3D Rage, offered unquestionably inferior 3D acceleration, their low cost and simplicity often appealed to OEM system builders over the addition of another expensive and limited-use card (especially with the then-unproven 3D game market). Rendition's Vérité V1000 was an integrated (3D+VGA) single-chip solution as well that was perhaps Voodoo's closest competitor, but it too did not have comparable 3D performance (equal quality, however) and its 2D was considered merely adequate relative to other 2D cards of the time (slower than ViRGE, Rage, and Mystique).
The Voodoo heralded a new era in 3D graphics. Prior to that, games such as Doom and Quake had compelled video game players to move from their 80386s to 80486s, and then to the Pentium. Intel had developed their MMX instruction set that would increase 3D and multimedia performance, but adding in a Voodoo graphics card would offer a substantial gain in speed, making the graphics card more important than the CPU for 3D games.
Originally developed for use within the specialized resource constrained environment of an arcade game that included non-Intel architectures such as MIPS and PowerPC CPUs, as well as to support DOS-based, full-screen 3D games which were still popular at the time and which neither Microsoft nor Intel desired to support with their own APIs (Intel 3DR and Microsoft 3D-DDI), Glide was created to handle error prone tasks like chip initialization for the programmer, but implemented nothing more than what the Voodoo hardware was directly capable of. This strategy differed from that of other 3D APIs of the era (Direct3D, OpenGL, and QuickDraw 3D), which hid low-level hardware details behind an "abstraction layer," with the goal of providing application developers a standard, hardware-neutral interface.
The advantage of an abstraction layer is that game developers save programming effort and gain flexibility by writing their 3D rendering code once, for a single API, and the abstraction layer allows it to run on hardware from multiple manufacturers. This advantage is still in place today. However, in the early days of the 3D graphics card, Direct3D and OpenGL implementations were either non-existent or, at minimum, substantially less mature than today, and computers were much slower and had less memory. The abstraction layers' overhead crippled performance in practice. 3dfx had therefore created a strong advantage for itself by aggressively promoting Glide, which was designed specifically around the Voodoo hardware, and therefore did not suffer from the performance hit of a higher level abstraction layer.
While there were many hit games that used Glide, including several coin-op/arcade games from Midway Games and Atari Games (e.g., San Francisco Rush, NFL Blitz, Hydrothunder, etc.) and games like Eidos' Tomb Raider, the killer application for Voodoo Graphics was the MiniGL driver developed specifically to allow hardware acceleration of the game Quake, by id Software, on 3dfx cards. The driver implemented only the subset of OpenGL used by Quake.
By 2000, the improved performance of Direct3D and OpenGL on the average personal computer, coupled with the huge variety of new 3D cards on the market, the widespread support of these standard APIs by the game developer community and the closure of 3dfx, would make Glide obsolete.
In August 1997, 3dfx released the Voodoo Rush chipset, combining a Voodoo chip with a 2D chip that lay on the same circuit board, eliminating the need for a separate VGA card. Most cards were built with an Alliance Semiconductor AT25/AT3D 2D component, but there were some built with a Macronix chip and there were initial plans to partner with Trident but no such boards were ever marketed.
The Rush had the same specifications as Voodoo Graphics but performed worse because the Rush chipset had to share memory bandwidth with the CRTC of the 2D chip. The Rush chipset also was not directly present on the PCI bus but had to be programmed through linked registers of the 2D chip. Like the Voodoo Graphics, there was no interrupt mechanism, so the driver had to poll the Rush in order to determine whether a command had completed or not; the indirection through the 2D component added significant overhead here and tended to back up traffic on the PCI interface. The typical performance hit was around 10% compared to Voodoo Graphics, and even worse in windowed mode. Later Rush boards released by Hercules had 8 MiB VRAM and a 10% higher clock speed to close the performance gap.
A rare third version was produced which featured a Cirrus Logic 2D chip instead of the earlier model. This version fixed the PCI bus collisions and memory interface problems.
In 1998, 3dfx released Voodoo's successor, the popular Voodoo2. The Voodoo2 was architecturally similar, but the basic board configuration added a second texturing unit, allowing two textures to be drawn in a single pass.
A problem with the Voodoo2 was the fact that it required three chips and a separate VGA graphics card, whereas new competing 3D products, such as the ATI Rage Pro, Nvidia RIVA 128, and Rendition Verite 2200, were single-chip products. Despite some shortcomings, such as the card's dithered 16-bit 3D color rendering and 800×600 resolution limitations, no other manufacturers' products could match the smooth framerates that the Voodoo2 produced. It was a landmark (and expensive) achievement in PC 3D-graphics. Its excellent performance, and the mindshare gained from the original Voodoo Graphics, resulted in its success. Many users even preferred Voodoo2's dedicated purpose, because they were free to use the quality 2D card of their choice as a result. Some 2D/3D combined solutions at the time offered quite sub-par 2D quality and speed.
The arrival of the Nvidia RIVA TNT with integrated 2D/3D chipset would offer minor challenge to the Voodoo2's supremacy months later.
SLI capability was not offered in subsequent 3dfx board designs, although the technology would be later used to link the VSA-100 chips on the Voodoo 5.
Having since acquired 3dfx, Nvidia in 2004 reintroduced the SLI brand (now for Scalable Link Interface) in their GeForce 6 Series. ATI Technologies has also since introduced its own multi-chip implementation, dubbed "CrossFire". Although Scalable Link Interface and Crossfire operate on the original SLI principle, the algorithms used are now totally different.
Near the end of 1998, 3dfx released the Voodoo Banshee, which used a lower price to aim at a more mainstream consumer market. A single-chip solution, the Banshee was essentially a legacy VGA core combined with a higher clocked but incomplete (only one Texture Mapping Unit) Voodoo2. The Banshee's single-chip form factor dictated a 128-bit memory bus, like the first Voodoo. Performance wise, the Banshee was a mixed bag. In scenes which used multiple textures per polygon, the Voodoo2 was substantially faster, due to the second TMU. In scenes dominated by single-textured polygons, though, the Banshee would match (or even slightly exceed) the Voodoo2 due to its higher clock speed. While it was not a hit on the scale of the Voodoo 1 or 2, the Banshee sold a respectable number of units, although 3dfx started losing some market share to Nvidia's RIVA TNT.
While the 3D performance was somewhat of a disappointment, Banshee's 2D core was anything but. It rivaled the fastest 2D cores from Matrox, Nvidia, and ATI. The chip was equipped with a 128-bit 2D GUI engine and a 128-bit VESA VBE 3.0 VGA core. It was perhaps the fastest MS-DOS performer released, certainly at the time. Windows performance was equally impressive. The graphics chip capably accelerated DirectDraw and supported all of the Windows Graphics Device Interface (GDI) in hardware, with all 256 raster operations and tertiary functions, and had hardware polygon acceleration. All of this helped the 2D core be able to boast near-theoretical performance with a null driver test in Windows NT. 3dfx had indisputably fixed part of the weakness of the older Voodoo Rush with its slow 3rd-party 2D chip.
Many OEM's at the time refused to use Voodoo Banshee as it did not support true 32bit color. This feature has by now become a checkbox item when NVIDIA shipped TNT. 3dfx had a small window of opportunity to address this limitation or have OEM customers commit themselves to NVIDIA's developing roadmap.
When 3dfx declared its Initial Public Offering (IPO) in April 1997, it mistakenly revealed confidential details of the Sega contract. By law, an IPO filing in the United States must include the complete details of the company's business and financial situation for public review. However, information that does not materially affect the company's statement of its financial position and outlook may be withheld. Sega had been keeping the development of its next-generation console secret during this competition, and was outraged when 3dfx's IPO included extensive details of the console development contract.
Sega quickly quashed "Blackbelt" and used "Katana" as the model for the product would be marketed and sold as the Sega Dreamcast. 3dfx sued Sega for breach of contract, accusing Sega of starting the deal in bad faith to take 3dfx technology. The case was settled out of court.
3dfx executed a major strategy change just prior to the launch of Voodoo3 by purchasing STB Technologies, which was one of the larger graphics card manufacturers at the time; the intent was for 3dfx to start manufacturing, marketing, and selling its own graphics cards, rather than functioning only as an OEM supplier. STB was obviously intended to give 3dfx access to that company's considerable OEM resources and sales channels, but the intended benefits of the acquisition never materialized. The two corporations were vastly different entities, with different cultures and structures, and they never integrated smoothly.
STB prior to the 3dfx acquisition also approached NVIDIA as a potential partner to acquire the company. At the time STB was NVIDIA's largest customer and was only minimally engaged with 3dfx. 3dfx management mistakenly believed that acquiring STB would ensure OEM design wins with their products and that product limitations would be overcome with STB's knowledge in supporting the OEM sales/design win cycles. NVIDIA decided not to acquire STB and to continue to support any branded graphics board manufacturers. After STB was acquired by 3dfx, NVIDIA focused on being a virtual graphics card manufacturer for the OEMs and strengthened their position in selling finished reference designs ready for market to the OEMs. STB's manufacturing facility in Juarez, Mexico was not able to compete from either a cost or quality point of view when compared to the burgeoning ODM's and CEM's that were delivering solutions in Asia for NVIDIA. Some of 3dfx's OEMs warned the company prior to the STB merger finalizing that any product from Juarez will not be deemed fit to ship with their systems, however 3dfx management believed these problems could be addressed over time. Those customers generally moved to NVIDIA solutions and no longer chose to ship 3dfx products.
The acquisition of STB was one of the main contributors to 3dfx's downfall; the company did not sell any Voodoo 4 or 5 chips to third party manufacturers which were a significant source of revenue for the company. These now cut off third party manufacturers turned into competitors and began sourcing graphics chips from NVIDIA who was now turning into a beneficiary of the STB/3dfx merger. This also further alienated 3dfx's remaining OEM customers, as they had a single source for 3dfx products and could not choose a CEM to provide cost flexibility. The OEMs in fact saw 3dfx as a direct competitor in retail. With the purchase of STB 3dfx created a line of Velocity boards (a STB brand) that used crippled Voodoo3 chips, as a product to target the low-end market. The chip came with only a single functional TMU, making it similar to a Voodoo Banshee.
As 3dfx focused more on the retail graphics card space further inroads into the OEM space were limited. A significant requirement of the OEM business was the ability to consistently produce new products on the six month product refresh cycle the computer manufacturers required; 3dfx did not have the methodology nor the mindset to focus on this business model. In the end, 3dfx opted to be a retail distribution company manufacturing their own branded products.
The Voodoo 3 was heavily hyped as the graphics card that would make 3dfx the undisputed leader but the actual product was below expectations. Though it was still the fastest by a small margin, for 3dfx API(Glide) lacked 32-bit color and large texture support. While at the time, few games supported large textures and 32-bit color, and those that did generally were too demanding to be run at playable framerates, the features "32-bit color support" and "2048x2048 textures" were much more impressive on paper than 16-bit color and 256x256 texture support. The Voodoo3 sold relatively well, but was disappointing compared to the first two models and 3dfx gave up the market leadership to Nvidia. Ironically, 3dfx's actions in the STB buyout had turned a lot of its former board partners into Nvidia-exclusive suppliers (ATI did not start outsourcing its products until 2002), giving Nvidia a major advantage in the marketplace.
As 3dfx attempted to counter the TNT2 threat, they were caught off guard by Nvidia's GeForce 256. With its high invests in T&L hardware capacities and increased pixel pipelines (unaffordable for 3dfx at the time), Nvidia took Geforce 256 to a new degree of 3D quality, making it superior to the Voodoo 3 in all respects (except price).
The company's next (and as it would turn out, final) product was code-named Napalm. Originally, this was just a Voodoo3 modified to support newer technologies and higher clock speeds, with performance estimated to be around the level of the RIVA TNT2. However, Napalm was delayed, and in the meantime Nvidia brought out their landmark GeForce 256 chip, which shifted even more of the computational work from the CPU to the graphics chip. Napalm would have been unable to compete with the GeForce, so it was redesigned to support multiple chip configurations, like the Voodoo2 had. The end-product was named VSA-100, with VSA standing for Voodoo Scalable Architecture. 3dfx was finally able to have a product that could defeat the GeForce.
However, by the time the VSA-100 based cards made it to the market, the GeForce 2 and ATI Radeon cards had arrived and were offering higher performance at that price point. The only real advantage the Voodoo 5 5500 had over the GeForce 2 GTS or Radeon was its superior anti-aliasing implementation, and the fact that it didn't take such a large performance hit (relative to its peers) when anti-aliasing was enabled. 3dfx was fully aware of the Voodoo 5's speed deficiency, so they touted it as quality over speed, which was a reversal of the Voodoo 3 marketing which emphasized raw performance over features. 5500 sales were respectable but volumes were not at a level to keep 3dfx afloat.
The Voodoo 5 5000, which had 32 MB of VRAM to the 5500's 64 MB, was never launched, as the smaller frame buffer didn't significantly reduce cost over the Voodoo 5 5500.
The only other member of the Voodoo 5 line, the Voodoo 4 4500, was as much of a disaster as Voodoo Rush, because it offered performance well short of its value-oriented peers combined with a late launch. Voodoo 4 was beaten in almost all areas by the GeForce 2 MX — a low-cost board sold mostly as an OEM part for computer manufacturers — and the Radeon VE.
One unusual trait of the Voodoo 4 and 5 was that the Macintosh versions of these cards had both VGA and DVI output jacks, whereas the PC versions only had the VGA connector, and lacked DVI. Also, the Mac versions of the Voodoo4 and 5 had an Achilles' heel in that they did not support hardware-based MPEG2 acceleration, which hindered the playback of DVDs on a Mac equipped with a Voodoo graphics card.
The Voodoo 5 6000 never made it to market, due to a severe bug resulting in data corruption on the AGP bus on certain boards, and was limited to AGP 2x. It was thus incompatible with the then-new Pentium 4 motherboards. Later tests proved that the Voodoo 5 6000 outperformed not only the GeForce 2 GTS and ATI Radeon 7200, but also the faster GeForce 2 Ultra and Radeon 7500. In some cases it was shown to compete well with the GeForce 3, trading performance places with the card on various tests. However, the production cost of the card, particularly the 4 chip setup, external power supply and 128 MB of VRAM, would have likely hampered its competitiveness.
In late 2000, not long after Voodoo 4's launch, several of 3dfx's creditors decided to initiate bankruptcy proceedings. 3dfx would have had virtually no chance of successfully contesting these proceedings, and instead opted to be bought by Nvidia, ceasing to exist as a company. Most of the design team working on "Rampage" (the successor to the VSA-100 line) was transferred to the team working on what has since become the GeForce FX series.
After Nvidia acquired 3dfx, mainly for its intellectual property, they announced that they would not provide technical support for 3dfx products. Drivers and support have since been available at Voodoo Files and other various websites. Nvidia offered a limited time program where 3dfx owners could trade in their cards for Nvidia cards. On December 15 2000 3dfx apologized to the customers with a final press release.
Second, the company prioritized its next-generation designs and resulting high-end products, frequently ignoring the mid and low end sectors. Nvidia chose comparatively short development cycles, typically releasing a new product line each cycle and a high-end refresh midway through, and generally had a product priced to suit each market segment as older products and technologies moved progressively down-market. ATi quickly adopted a similar schedule. 3dfx however pursued lengthy, ambitious development cycles on higher end products and then attempted to fill in the down-market sectors with detuned versions of the newest release. This strategy eventually faltered as Nvidia and ATI cards built successively upon earlier gains, eventually exceeding 3dfx in overall performance. Nvidia's flagship GeForce 256 and GeForce 2 GTS are often given credit for the demise of the competing Voodoo 3 and Voodoo 5, respectively, but it is also important to note that the GeForce 2 MX midrange derivative was what successfully targeted the mass-market. The MX and acquired substantial market share for Nvidia, and 3dfx's reaction, the Voodoo 4 4500, was uncompetitive in both price and performance when finally released.
Third, Greg Ballard became CEO of 3dfx in 1997, and analysts marked it as a turning point since Ballard was a marketing guru. However Ballard failed to understand R&D in the graphics industry. Single-card 2D/3D solutions were taking over the market, and although Ballard saw the need and attempted to direct the company there with the Voodoo Banshee and the Voodoo3, both of these cost the company millions in sales and lost market share while diverting vital resources from the Rampage project. Then 3dfx released word in early 1999 that the still-competitive Voodoo2 would only support OpenGL and Glide under Microsoft's Windows 2000 operating system, and not DirectX. Many games were transitioning to DirectX at this point, and the announcement caused many PC gamers – the core demographic of 3dfx's market – to switch to Nvidia or ATI offerings for their new machines.
Fourth, the merger with STB alienated 3dfx from much of its previous market, and never developed into the right kind of arrangement to broaden 3dfx's distribution base from what the company had when still acting as a chipset supplier to third parties. The merger consequently diverted vital resources from 3dfx's chip development operations without ever producing a corresponding payback in sales.
Fifth, the next-generation "Rampage" project was exceedingly ambitious and delayed repeatedly as the company continued to vacillate on its commitment to Rampage development versus short-term retail products, such as the Voodoo 3 and Napalm/VSA-100. The Rampage design team was using a pioneering synthesis tool set which was still under development as the design proceeded. Originally scheduled for demonstration at the 1998 Comdex event, the first functional samples of rage only debuted in 3dfx's labs in December of 2000. The 2D and 3D driver software was already up and running to support the test hardware, but the impending release of Rampage was too little, too late. The deal to "wind down" the company and transfer core assets to Nvidia was less than 2 weeks from closure at that point.
While some have speculated that shipping the "Rampage" might have saved 3dfx, the fact remains that the company never mastered the new concept of relatively cheap, high-performance dies with integrated 2D acceleration, which was rapidly becoming the de facto standard of PC graphics. The success of "Rampage" would not have simply depended upon raw performance, but also the cost of manufacturing. It remains unknown whether "Rampage" would have been a practical product, let alone enough to keep the company alive in the card industry. The history of and participants in the 3dfx/Nvidia deal making can be read in the respective companies financial filings from that time period. The resolution and legality of those arrangements (with respect to the purchase, 3dfx's creditors and its bankruptcy proceedings) was still being worked through the courts as of August 2006, nearly 6 years after the sale.
|Chipset||Release Date|| Components|
| Core |
|Voodoo Graphics||October 1996||SST-1 chipset (1x1). No VGA.||50||50||50/50||64-bit×2||4/6 EDO||PCI|
|Voodoo Rush||April 1997||SST-96 chipset (1x1). 2D Chip.||75||75||75/75||64-bit×2||6/8 EDO||PCI|
|Voodoo2||January 1998||SST-2 chipset (1x2). No VGA.||90||90||90/180||64-bit×3||8/12 EDO||PCI|
|Voodoo Banshee||October 1998||Single-Chip (1x1) (2D/3D)||100||110||100/100||128-bit||8/16 SD/SG||AGP 1x/PCI|
|Velocity 100||July 1999||Avenger core (1x1). 1 Disabled TMU.||143||143||143/143||128-bit||8 SG||AGP 2x|
|Velocity 200||Not released||Avenger core (1x1). 1 Disabled TMU.||143||143||143/143||128-bit||16 SG||AGP 2x|
|Voodoo3 1000||March 1999||Avenger core (1x2)||125||125||125/250||128-bit||8 SG||AGP 2x|
|Voodoo3 2000||April 1999||Avenger core (1x2)||143||143||143/286||128-bit||16 SD/SG||AGP 2x/PCI|
|Voodoo3 3000||April 1999||Avenger core (1x2)||166||166||166/333||128-bit||16 SD/SG||AGP 2x/PCI|
|Voodoo3 3500||July 1999||Avenger core (1x2), A/V processor||183||183||183/366||128-bit||16 SD||AGP 2x|
|Voodoo 4-2 4200||Not released||1 x VSA-1018 (2x1)||183||183||366/366||64-bit||32 DDR||AGP 4x/PCI|
|Voodoo 4 45005||October 2000||1 x VSA-100 (2x1)||166||166||333/333||128-bit||32 SD||AGP 4x/PCI|
|Voodoo 4 4800||Not released||1 x VSA-100 (2x1)||166||166||333/333||128-bit||64 SD||AGP 4x/PCI|
|Voodoo 5 5000||Not released||2 x VSA-100 (2x1)||166||166||667/667||128-bit×2||321 SG||AGP 2x/PCI|
|Voodoo 5 55005||June 2000||2 x VSA-100 (2x1)||166||166||667/667||128-bit×2||642 SD||AGP 2x/PCI|
|Voodoo 5 6000||Not released||4 x VSA-100 (2x1)||166/1834||166/1834||1333/1333||128-bit×4||1283 SD||AGP 2x|
|Spectre 1000||Not released||1 x Rampage (4x1)||200-250 (planned)||200-250 (planned)||800-1000||128-bit||64 DDR||AGP 4x|
|Spectre 2000||Not released||1 x Rampage (4x1) + 1 x Sage T&L-Chip||200-250 (planned)||200-250 (planned)||800-1000||128-bit||64 DDR||AGP 4x|
|Spectre 3000||Not released||2 x Rampage (4x1) + 1 x Sage T&L-Chip||200-250 (planned)||200-250 (planned)||1600-2000||128-bit||128 DDR||AGP 4x|