NVIDIA GeForce 7 Phase 2 Review :: A Short History of NVIDIA Video Cards

NVIDIA was founded in 1993. The first video card they released was the NV1. This card was rather unique in the history of video cards in that it had onboard audio, had the ability to use Sega Saturn controllers, and more. The rendering on that card used quadratic equations instead of the polygons. When Microsoft introduced Direct3D with their Windows Operating System, they chose to use polygons, making the NV1 obsolete. The second chip, NV2 never made it to the market due to the advent of Direct3D.

The next chip was the NV3, or Riva 128. This was the first NVIDIA Direct3D supporting video card. The Riva 128 has up to 4MB of framebuffer, a clock speed of 100MHz, memory speed of 100MHz, 230MHz RAMDAC and more. The Riva128 competed relatively well with the competition, the 3dfx Voodoo and the ATI Rage Pro chip, but the 2D was less than spectacular compared to the competition.

6 months after the release of the Riva 128 NVIDIA released the first "refresh" part, the Riva 128ZX. This was the first video cad with support for 8MB of frame buffer memory. The Riva 128ZX was their first foray into the OEM market, with many systems sold with versions of this card. This set the stage for a new video card every 6 months or so. NVIDIA has released a huge amount of chips since that time.

The Fall of 1998 brought NVIDIA into the high-end gaming arena with the introduction of the TNT. The TNT was the first NVIDIA video card to support dual textures in a single pass. It was the first NVIDIA card to support 32-bit color. The competition of the time, the Matrox G400 the 3DFX Voodoo 3 and the ATI Rage 128 chips were either late (ATI), didn't support 32-bit color (3DFX) or had trouble with no OpenGL ICD (Matrox)

The next card in NVIDIA's lineup was the TNT2. This was a die shrink of the original TNT moving from the 0.35 nanometer process to the 0.25 nanometer process. The TNT2 was the card the TNT was supposed to be with fillrates of 250 Megatexels a second (125MHz clock speed), faster performance in 32-bit color to compete with the ATI card of the time, and was generally considered a big success.

I would consider the introduction of the GeForce in the fall of 1999 to be the first card to support DirectX 7.0 features including Hardware Transformation and Lighting, bump mapping and more. NVIDIA has kept the name GeForce for each of the subsequent cards after the first one. The GeForce came in two flavors initially, the GeForce SDR and the GeForce DDR. This was the first card that really differentiated NVIDIA cards from their competition.

The spring of 2000 saw the introduction of the GeForce 2 GTS card. This increased the clock speed of the video card from 120MHz on the GeForce to 200MHz, and the memory clock speed from 150MHz to 166MHz. The GTS was the first commercially available video card with over 1 Gigatexel fill rates. The GTS was on a 0.18 micron process, NVIDIA's first card on that process.

The fall of 2000 release was interesting to say the least. NVIDIA decided to release a higher clock speed version of the GeForce 2 GTS card. The clock speed of the video card was 250MHz and 200MHz for the memory. This card brought the price of the gaming video cards to an unheard of $499 MSRP. This was the first video card to break the $399 price barrier and was the realm of only the craziest gamers.

NVIDIA won the Xbox contract in 2000. The graphics chip on the Xbox was a modified version of their next generation architecture, the NV2x. The PC version of the NV2x was called the GeForce 3 released in the spring of 2001. The GeForce 3 was the first video card to support Programmable Pixel Shader 1.1 and Vertex Shader 1.0 standard introduced with the DirectX 8.0 API from Microsoft.

The fall 2001 product lineup consisted of the GeForce Ti500 for the high-end, and GeForce 3 Ti200 for the performance mainstream. The Ti500 was clocked at 240MHz for the core and 250MHz for the memory, compared to 200MHz for the core and 230MHz for the memory. The Ti500 was a small jump from the original Geforce3, meant to compete with ATI's RADEON 8500 series of cards.

NVIDIA's plans for 2002 really centered on the launch of Microsoft's DirectX 9.0 API, Unfortunately, everything did not go according to plan. The spring of 2002 saw the introduction of the Ti4xxx series, in the form of the Ti4200, the Ti4400 and the Ti4600, the performance mainstream, the middle range and the high-end. The 4xxx series nearly doubled the performance from the GeForce 3.

The first DirectX 9.0 card from NVIDIA was called the GeForce FX 5800 Ultra. The original plan was to launch this card in 2002, but there were problems with the transition to the 130 nanometer process and there were serious issues with the design which was basically a DirectX 8.0 core with DirectX 9.0 feature support. Several of the DirectX 9.0 games released in the last couple of years had the 5800 series default to DirectX 8.0 due to performance issues. The mainstream version of the 5xxx series the 5600 was basically 1/2 a 5800, and substandard for most DirectX 9.0 games like HL2. The value segment was served by the 5200, a decent competitor to the ATI 9200, but not a card I'd recommend even then.

Due to the late delivery of the 5xxx series, NVIDIA rapidly released the 5900 Ultra just 3 months later than the 5800 Ultra. The 5900 Ultra used a 256-bit memory bus and increased performance in every instance over the 5800 Ultra. The 5900 Ultra was rapidly replaced by the 5950 Ultra in the fall of 2003, slightly increasing the core clocks and memory speed. The 5700 Ultra was the replacement for the 5600 in the same timeframe. The 5700 Ultra still had many of the shortcomings of the 5xxx series but performed better and was fairly competitive with the 9600XT cards.

The debacle of the 5800 launch forced NVIDIA to. NVIDIA had a plan to take over feature and performance leadership in the video card market take their competition seriously and this culminated in the release of their next generation architecture, the NV40 or GeForce 6 series. Announced in April of 2004, this was the first video card to support Pixel Shader 3.0 and Vertex Shader 3.0, being released a full year and a half before ATI released their R520 chip. The mainstream chip for the GeForce 6 series was called the 6600, a very successful video card. The value part was called the 6200.

The GeForce 6 family was to remain NVIDIA's high-end card until the GeForce 7 series was announced in June of last year. This was the first time in nearly 6 years that NVIDIA didn't refresh their high-end for a year. The 7800 GTX increased the number of pixel pipelines to 24 from 16 on the 6800 Ultra, increased performance in MADD (multiply add divide) operations and had higher performance throughout the spectrum. 7800GTX was built on the 110 nanometer process, a budget process upgrade from the 130 nanometer process found on earlier NVIDIA cards. When ATI moved to 90 nanometer with their X1K series, NVIDIA decided to wait for this Spring with the release of the refresh which I'm reviewing here, the XFX 7900GTX XXX Edition, the EVGA 7900GT CO Superclocked, and the XFX 7600GT cards.