NVIDIA

NVIDIA are the world's leading producer of graphics processors. Although best known for their chipsets for PC and Mac graphics cards, NVIDIA's product range also includes motherboard chipsets as well as digital video hardware. NVIDIA don't manufacture cards or fabricate processors, instead contracting a partner company to fabricate their chips and then selling them to a variety of board manufacturers.

Company History

1993: nVidia (as it was then) is founded by Jen-Hsun Huang (President and CEO), Curtis Priem (Chief Technical Officer) and Chris Malachowsky (VP of Hardware Engineering).

1996: Their first chip, codenamed the NV1, is largely unsuccessful and brings the company close to bankrupcy. Around this time 3D accelerators begin to emerge as desirable consumer products, with 3Dfx's first generation Voodoo chipset rapidly grabbing the spotlight over slower competitors.

1997: The NV3, or in marketing speak the Riva128, is more successful, and begins to secure a steady stream of OEM supply deals with PC builders. At this stage nVidia also find themselves fighting off the occasional 'nuisance' lawsuit accusing them of patent infringment.

March 23, 1998: Riva TNT announced."The first single-chip 128-bit 3D processor that can process two pixels per clock cycle enabling true single-pass multi-texturing." (Buzzwords hadn't been adopted by NVIDIA yet.) OEM deals and board partners continue to roll in.

January 22, 1999: NVIDIA's IPO.

March 15, 1999: Riva TNT2 announced. NVIDIA finally deliver a product that outperforms its 3dfx rivals- the Riva TNT 2, which offers a higher fillrate, generous texture sizes and 32-bit colour, although its high price means that it is strictly for enthusiasts for a long time. This is the beginning of the end for 3dfx, who firstly make the ill-fated decision to manufacture and market their own consumer products, and then run into eventually insurmountable obstacles in trying to develop a new generation of chips. (It having become an industry joke that instead of developing new chips, they were content to simply add more running in parallel.) NVIDIA eventually buy the remaining part of their rival (neatly ending a lawsuit that 3dfx had filed against NVIDIA).

August 31, 1999: This is the turning point, as NVIDIA deliver a chip that decisively blows 3dfx's Voodoo series out of the water: the GeForce256. NVIDIA trumpet this chipset as the world's first GPU (Graphics Processor Unit). It is the first consumer chip to feature hardware transform and lighting (T&L), stages of the rendering pipeline that had previously been carried out on the CPU. Because all 3D calculations are now handled by the graphics chip, the GeForce is capable of maintaining high performance (relatively) irrespectively of the CPU speed and load.

March 10, 2000: Microsoft entered into a deal with NVIDIA to provide the GPU and MCP (a sound and networking chip) for their Xbox console. The announcement of the Xbox deal sent the now publicly traded NVIDIA stock price skyward, certainly bestowing a significant advantage to the company's subsequent efforts to remain at the forefront of the market. (Unfortunately the relationship did not remain totally rosy, with NVIDIA taking Microsoft to court over a dispute in payment for the chips it delivered. Microsoft were subsequently sniffing around for a new GPU partner, but as of February 2002 NVIDIA and Microsoft have kissed and made up.)

Cannily, NVIDIA have also used the technology they developed (with Microsoft's money) for the Xbox in their PC component lines, with the XBOX GPU (sans one shader unit) emerging as the GeForce 3 series, and the MCP sound and I/O processor appearing on the first generation nForce motherboard.

April 26, 2000: GeForce 2 announced. The second generation of the GeForce family began NVIDIA's trend of paired generations- odd numbers being the first chips to introduce substantially new architecture, with their even-numbered followups offering performance enhancements. GF2 included a more user-programmable iteration of hardware T&L, a dedicated pixel shader, and improved FSAA (although still fairly crude). John Carmack has mentioned that he doesn't think they should have called it 'GeForce2', as the chip was not a vast improvement on the GeForce256. (It was basically a revised, modestly faster and cheaper to produce replacement for the GeForce256, which was rapidly phased out.)

Around this time, NVIDIA also branched off a product line for professional 3D users (the Quadro and Quadro2), and introduced a reduced price cut-down chip (the wretched Geforce 2 MX), which successfully exploits the growing number of budget-conscious consumers who want to get into 3D gaming (and allows them to continue to supply to OEMs, who are not keen on the expensive standard GeForces). From this point on, NVIDIA make more effort to offer a gradiated scale of products to address competition across the price spectrum. They also introduce the GeForce 2 Go chip for laptops.

December 15, 2000: 3dfx buyout announced.

February 22, 2001: GeForce 3 unveiled at Tokyo Macworld Expo. A significant enhancement over what had gone before, the GeForce 3 featured programmable pixel and vertex shaders (dubbed the nfiniteFX engine), and a new, faster 'HRAA' anti-aliasing method called Quincunx (which makes too great a sacrifice of image quality, in my opinion). This is made possible by a new memory system rather grandiosely named the Lightspeed Memory Architecture.

February 5, 2002: NVIDIA ships 100 millionth GPU.

February 6, 2002: GeForce 4 announced. The current generation as of this writing, available in three 'flavours' - 4200, 4400, 4600 - as well as a pointless low-end version (MX 440). Very, very fast cards with glorious image quality, that visually put the PC out in front again compared to the latest generation consoles.

June 13, 2002: Cg shader language announced.

September 9, 2002: NVIDIA ranked fastest growing company in America by Fortune magazine.

November 18, 2002: GeForce FX (the fifth generation GeForce) is announced. It is the first GPU to use a 0.13 micron semiconductor process. It is promoted with the new buzzword 'cinematic computing'.

The Future?

Although NVIDIA currently have a strong (bordering at times on the monopolistic) position in the marketplace, they still face credible challenges from their competitors. Recently the ATi Radeon 9700 has nosed ahead of their GeForce 4 generation in terms of performance, and Creative Labs, board manufacturer and long-time partner of NVIDIA, have ended their relationship with the company, planning to develop their own product through recent acquisition 3D Labs. The arms race continues.

Currently NVIDIA have their headquarters in Santa Clara, California, offices around the US and subsidiaries in the UK, France, Germany, Japan, Singapore and Taiwan.

http://www.nvidia.com/