article

3dfx Interactive was a company which specialized in the manufacturing of cutting-edge 3D graphics processing units and, later, graphics cards. After dominating the field for several years in the late 1990s, by the end of 2000 it underwent one of the most high-profile demises in the history of the PC industry. It was headquartered in San Jose, California until, on the verge of bankruptcy, its intellectual assets (and many employees) were acquired by its rival, NVIDIA Corporation.

Early history

Founded in 1994 by Ross Smith, Gary Tarolli and Scott Sellers (all SGI alumni), with backing from Gordie Campbell's TechFarm, 3dfx released its famous 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, who designed, manufactured, marketed, and sold their own graphics cards which included the Voodoo.

3Dfx became popular mainly 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. It generated a lot of very positive mindshare among gamers for 3Dfx. Voodoo Graphics was some of the best arcade hardware at the time, and it was coming to the home PC.

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, alongside Rendition's Vérité V1000 chip and PowerVR PCX1/2, was one of the first 3D accelerators that actually improved graphics performance; most systems of the era performed no better, if not worse, than software rendering engines (i.e. ViRGE).

The Voodoo 1, as the Voodoo Graphics would be later known, was notable for its lack of an onboard VGA controller. This meant a Voodoo equipped PC still required a separate VGA graphics card, meaning it was very expensive to have both 3D and 2D acceleration, and the user lost an extra slot. The Voodoo 1 occupied a separate PCI slot and only kicked in when the host PC ran a 3D game that had been programmed to use the Voodoo. A passthrough VGA cable daisy-chained the VGA card to the Voodoo 1, and then out to the monitor. Although this was a cumbersome arrangement that hurt the analog signal quality of the separate 2D card, hardcore PC gamers were willing to put up with it to gain what was (then) the ultimate in 3D graphics.

The Voodoo 1's main competition was PowerVR and Rendition. PowerVR produced a 3D add-on card, whereas Rendition offered an integrated (3D+VGA) single-chip solution. Neither competitor achieved the Voodoo 1's popularity among gamers and developers, or equivalent performance.

Glide API

In order to ensure better performance, 3dfx developed the proprietary Glide API for game developers to use while writing their 3D games. Glide exposed the Voodoo's internal hardware to application programmers directly; it was essentially a small subset of OpenGL that could be implemented in hardware. This strategy differed from that of other 3D APIs of the era (Direct3D, OpenGL, and QuickDraw 3D), which all hid low-level hardware behind an "abstraction layer," with the goal of providing application developers a standard, hardware-neutral interface.

The ostensible advantage of an abstraction layer is that game developers save programming effort by writing their 3D rendering code one time, for a single API, and the abstraction layer takes care of managing all the differences in all the hardware cards available to consumers. This advantage is still in place today. However, in the early days of the 3D graphics card, Direct3D and OpenGL were 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 actually implemented in hardware, and therefore effectively eliminated the speed and memory problems of an abstraction layer. Although a full OpenGL library was available for the Voodoo, most developers were glad to spend extra programmer time to support Glide instead, so their games would run faster and look better.

The killer application for the Voodoo 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 1999, the improved performance of Direct3D and OpenGL on the average personal computer would make Glide obsolete.

Voodoo Rush

In August 1997, 3dfx released the Voodoo Rush chipset, combining a Voodoo chip with a 2D chip from Alliance Semiconductor that lay on the same circuit board, eliminating the need for a separate VGA card. Unfortunately it performed worse than the Voodoo 1, primarily because the 2D and 3D cores shared the same memory interface and could not master the PCI bus correctly, incurring a 10% performance hit. Later versions released by Hercules had 8 MB VRAM and a 10% higher clock speed to close the performance gap, but in the marketplace the damage had already been done. 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, but there was little interest from graphics board manufacturers and only a few hundred units (if even that) were produced before the Rush was finally discontinued in early 1998.

Voodoo2

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. The Voodoo2 also had a faster clock rate (90MHz), a wider memory bus (192-bit, compared to Voodoo's 128-bit), and support for larger amounts of memory (up to 8 MB texture / 4 MB frame buffer compared to the Voodoo's 4 MB texture / 2 MB frame buffer.) A single Voodoo2 board could display a maximum resolution of 800×600 with higher quality textures.

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 this shortcoming (and the 16-bit 3D color rendering limitation), no other manufacturers' products could match the smooth framerates that the Voodoo2 produced. It was a landmark (and expensive) achievement in PC's 3D-graphics. The arrival of the NVIDIA RIVA TNT with integrated 2D/3D chipset would eventually challenge the Voodoo2's supremacy months later.

SLI

The Voodoo2 introduced Scan-Line Interleave (SLI) to the gaming market. In SLI mode, two Voodoo2 boards were connected together, each drawing half the scanlines of the screen. For the price of a second Voodoo2 board, users could essentially double their 3D throughput. A welcome result of SLI mode was an increase in the maximum resolution supported, up to 1024x768. Despite the high cost and inconvenience of using three separate graphics cards, the Voodoo2 SLI scheme was clearly the pinnacle of gaming performance at the time.

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.

Voodoo Banshee

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 an overclocked 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 undisputably fixed part of the weakness of the older Voodoo Rush with its slow 3rd 2D chip. [http://www.3dfxzone.it/dir/3dfx/prodotti/vbanshee.htm

Sega Dreamcast

In 1997, 3dfx was working with Sega to develop Sega's next video game console. The process involved two competing designs: a unit called "Katana" being developed in Japan using NEC and VideoLogic technology vs. the "Blackbelt", a system designed in America using a GPU from 3dfx. This deal had the potential to get 3dfx's foot in the home console door, provided the Blackbelt became the console that would become the Sega Dreamcast. Unfortunately for 3dfx, Sega chose the NEC solution. 3dfx sued Sega for breach of contract when the Katana was chosen, accusing Sega of starting the deal in bad faith to take 3dfx technology, and eventually the case was settled out of court; but the failure of the Blackbelt was 3dfx's own doing.

When 3dfx declared its Initial Public Offering (IPO) in April 1997, it made the mistake of revealing every detail of the contract with Sega. By law, when a company files an IPO in the United States, it has to make public all details of its business and financial situation, but sensitive information can be kept secret, so long as it does not materially affect the company's statement of its financial position and outlook. Sega had been keeping the development of its next-generation console secret during this competition, and was outraged when 3dfx publicly laid out its deal with Sega over the new system in the IPO; Sega quickly quashed the Blackbelt and used the Katana as the model of the Dreamcast.

Decline

In early 1998, 3dfx embarked on its "Rampage" development project, which was to be a new graphics card that would take two years to develop, and would supposedly be several years ahead of the competition once it debuted. The company hired hardware and software teams in Austin, Texas to develop 2D and 3D Windows device drivers for Rampage in the summer of 1998. The hardware team in Austin initially focused on Rampage, but then worked on Transformation and Lighting (T&L) engines and on MPEG decoder technology. (Later, these technologies were part of the NVIDIA asset purchase in December 2000.)

Voodoo3 and strategy shift

In mid-1999 the single chip Voodoo3 was released, which was at heart a full Voodoo2 pipeline (unlike Banshee) with Banshee's 2D core.* Codenamed "Avenger", it was originally supposed to be named "Banshee 2" but Voodoo3 was adopted instead since the Voodoo brand had superior recognition. It was a compelling solution, since it met or exceeded the performance of a Voodoo2 SLI setup because of a far higher clock speed, while providing strong 2D and occupying only one slot. This made it the fastest graphics card on the market in many situations. However, due to the Voodoo3's design legacy, it lacked support for several key technologies that its competitors, ATI Technologies, Matrox and NVIDIA, had since integrated. Most notable were 32-bit color, textures greater than 256*256 pixels in size (up to 2048x2048, or "2k*2k"), and AGP texturing.

3dfx executed a major strategy change just prior to the launch of Voodoo3 by purchasing STB Systems, 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. This alienated 3dfx's OEM customers, all of whom chose to switch, and source their 3D chips from other manufacturers, rather than do business with a company who was their direct competitor at 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 Banshee chip. Interestingly, it was possible to re-enable the disabled 2nd TMU to turn the board into a full Voodoo3. *

This strategy change 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. The company was also presumably distracted by the need to focus both on the retail market as well as the OEM market, selling cards to computer manufacturers. The latter was hard-won business, but provided a steady income to fund subsequent development. 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 the focus on this business model. In the end, 3dfx opted to focus on the retail business using its own manufactured and branded products. The Voodoo3 sold relatively well, but disappointingly compared to the first two models and 3dfx gave up the market leadership to NVIDIA's RIVA TNT2 and then the GeForce 256.

Voodoo 4 and 5

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 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. It was built upon an enhanced 0.25µ, 6-layer metal process. The "enhanced" 0.25µ process refers to the fact that it takes advantage of shorter gate lengths, which allow for faster switching, which allow higher frequencies at the same manufacturing process and greater yields at those frequencies.

The two initial products were the Voodoo 4 4500 (single chip) and the Voodoo 5 5500 (dual chip), with a further two parts, the Voodoo 5 5000 (dual chip, but with a smaller frame buffer) and the Voodoo 5 6000 (quad chip) due to be launched later. 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 substantially higher performance. 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 as large a performance hit when anti-aliasing was enabled. Voodoo 4 4500 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 Radeon VE. The Voodoo 5 5000 was never launched, as the smaller frame buffer didn't significantly reduce cost over the Voodoo 5 5500.

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 while the Voodoo 5 6000 would have outperformed the GeForce 2 GTS, it would have been slower than the GeForce 2 Ultra and the GeForce 3, even if it had been equipped with a T&L engine.

Voodoo 4 was as much of a disaster as Voodoo Rush, and while Voodoo 5's sales were respectable, they were nowhere near as good as 3dfx needed. In late 2000, 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.

One unusual trait of the Voodoo4 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.

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 DVD's on a Mac that was equipped with a Voodoo graphics card.

Cause for Decline

3dfx's decline is a matter of debate. Some attribute it to 3dfx lavishly spending on its employees - it reported spending $30,000 just on company lunches and other perks a month, even up to the last two weeks before it went under. Another, earlier contribution to its decline was the Voodoo3's lack of 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. Had those technologies been implemented in the Voodoo3, 3dfx might have sold far more of their product, and perhaps been able to stay above water.

3dfx's fall is most often attributed to managerial prioritizing of research and development. When Greg Ballard became CEO of 3dfx in 1997, analysts marked it as a turning point since Ballard was a marketing guru, but he failed to understand R&D in the graphics industry. His attempt to develop a single-card 2D/3D solution in the forms of the Voodoo Banshee and the Voodoo3, even though that was 3dfx's weak point, ended up costing the company millions in sales and lost market share as well as diverting vital resources from the Rampage project. Voodoo cards were typically highly expensive, and left the mid and low end of the market to ATI and NVIDIA. NVIDIA chose short development cycles, whereas 3dfx pursued lengthy, ambitious development cycles, and NVIDIA and ATI cards eventually ended up with better overall performance, with Matrox holding the edge in image quality.

The "Rampage" project, which 3dfx put much effort into but never was able to bring to market, debuted in 3dfx's labs in December of 2000, within weeks of the sale of 3dfx's assets to NVIDIA. The Rampage design team was using a pioneering synthesis tool set which was still under development as the design proceeded.

In addition, the company continued to vacillate on its commitment to the delayed Rampage project versus the need for short-term retail products, such as the Napalm/VSA-100. Because Rampage was oft-delayed — it had been scheduled to show at the 1998 Comdex — 2D and 3D driver software was up and running when it hit the lab.

However, the impending release of Rampage was too little, too late. The deal to "wind down" the company was less than 2 weeks from closure at that point. 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 of those arrangements (with respect to 3dfx's creditors and its bankruptcy proceedings) was still being worked through the courts as of September 2005.

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 to become the de facto standard of PC graphics cards very soon. The success of "Rampage" would not have simply depended upon raw performance, but also the cost of manufacturing, very much reflected in retail prices. According to documents from late in 3dfx's life, the "Rampage" core was evidently not too much more than a more powerful version of the VSA-100, with an entirely separate chip code-named "Sage" required for T&L and hardware shader operation; though to their credit, support for DDR-RAM was implemented. It remains unknown whether "Rampage" would have been a practical product, let alone enough to keep the company alive in the card industry.

Chip table

Chip Release Date Components (PPxTMU)6 Core
(MHz) 		Memory
(MHz) 		Fillrate7
Pix/Tex 		Memory
Bus Width9 		Memory
(megabytes) 		Bus
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-1 chipset (1x1). 2D Chip. 75 75 75/75 64-bit×2 6/8 EDO PCI
Voodoo2 January 1998 SST-96 chipset (1x2). No VGA. 90 90 90/180 64-bit×3 8/12 EDO PCI
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 ? 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) 166 166 166/332 128-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 45005 October 2000 1 x VSA-100 (2x1) 166 166 332/332 128-bit 32 SD AGP 4x/PCI
Voodoo 4 4800 Not released 1 x VSA-1018 (2x1) 183 183 366/366 128-bit 32 DDR AGP 4x/PCI
Voodoo 5 5000 Not released 2 x VSA-100 (2x1) 166 166 664/664 128-bit×2 321 SG AGP 2x/PCI
Voodoo 5 55005 June 2000 2 x VSA-100 (2x1) 166 166 664/664 128-bit×2 642 SD AGP 2x/PCI
Voodoo 5 6000 Not released 4 x VSA-100 (2x1) 166/1834 166/1834 1328/1328 128-bit×4 1283 SD AGP 2x

  • 2D/3D - products before the Banshee contain only a 3D core (no legacy/VGA)
  • 1 Shared by two processors; effectively 16 MB VRAM.
  • 2 Shared by two processors; effectively 32 MB VRAM.
  • 3 Shared by four processors; effectively 32 MB VRAM.
  • 4 The Voodoo 5 6000 was originally intended to have a core and memory clock of 183 MHz, but all of the prototypes running at 183 MHz stopped working after a short while. The only still-working Voodoo 5 6000s all run at 166 MHz, and 3dfx had decided to drop the 183/183 MHz idea anyway.
  • 5 Macintosh versions were released later with the same specs, but only in PCI.
  • 6 Pixel pipeline/s x Texture management unit/s (TMU).
  • 7 Fillrate is measured in Megapixels/sec for pixel fillrate and Megatexels/sec for texture fillrate. Taking advantage of double texture fillrate per clock cycle requires the game engine to make use of multitexturing.
  • 8 Daytona core
  • 9 Voodoo boards with multiple chips often have independent memory busses for each chip's memory bank. This resulted in boards with large amounts of memory, but the memory was split between the completely separate chips. So while a Voodoo2 has 8 or 12 MB of RAM, this RAM is split up between the texture and framebuffer.
In the case of Voodoo2 12 MB, there is 4 MB texture RAM for each texture processor and 4 MB for the framebuffer. When a game is not multitexturing, a total of 8 MB texture memory is available. With multitexturing, this is halved to 4 MB. SLI, however, does not double the available RAM because the boards separately calculate their part of the scene.

This split-RAM architecture caused the resolution limits on the older boards. For Voodoo Graphics' 4 MB RAM (2 MB framebuffer) framebuffer size only allowed for a 640 × 480 resolution when z-buffering was used. Some games managed 800 × 600 without z-buffer, but they were few and far between. MechWarrior 2 3Dfx Edition * was one such game.

See also

External links

Defunct computer companies of the United States | Graphics cards | 1994 establishments | 2000 disestablishments

3dfx | 3dfx | 3dfx | 3dfx | 3dfx | 3dfx | 3dfx | 3dfx Interactive | 3dfx

 

This article is licensed under the GNU Free Documentation License. It uses material from the "3dfx".

Home Page • arts • business • computers • games • health • hospitals • home • kids & teens • news • physicians • recreation• reference • regional • science • shopping • society • sports • world