Comparatif monstre de 35 cartes graphiques

Tom’s hardware publie un impressionnant comparatif de cartes graphiques AGP puisque ce ne sont pas moins de 35 cartes graphiques AGP qui sont testées

  •  GeForce 6800 Ultra
  •  GeForce 6800 GT
  •  GeForce 6800
  •  GeForce FX 5950 Ultra
  •  GeForce FX 5900 Ultra
  •  GeForce FX 5900
  •  GeForce FX 5900 XT
  •  GeForce FX 5800 Ultra
  •  GeForce FX 5800
  •  GeForce FX 5700 Ultra
  •  GeForce FX 5700
  •  GeForce FX 5600 Ultra
  •  GeForce FX 5600
  •  GeForce FX 5200 Ultra
  •  GeForce FX 5200
  •  GeForce4 Ti 4600 (suite pour voir le reste de la liste)
  •  GeForce4 Ti 4200
  •  Radeon X800 XT PE
  •  Radeon X800 Pro
  •  Radeon 9800 XT
  •  Radeon 9800 Pro 256
  •  Radeon 9800 Pro
  •  Radeon 9800
  •  Radeon 9800 SE
  •  Radeon 9700 Pro
  •  Radeon 9700
  •  Radeon 9600 XT
  •  Radeon 9600 Pro
  •  Radeon 9600
  •  Radeon 9600 SE
  •  Radeon 9500 Pro
  •  Radeon 9500 128
  •  XGI V8 Duo V8
  •  XGI V5 Ultra V5
  •  S3 S8 256

After nearly 10 months, we’re back again with an updated version of THG’s VGA Charts. This time around, not quite as many new products were launched in 2004. While an abundance of new products flooded the market in 2003, the two market leaders ATI and NVIDIA held back this year, at least with respect to the number of products launch. At the same time, performance-wise, both graphics processor makers have made great strides with their enthusiast line of products. The performance leap, in fact, was almost desperately needed, as games continue to become more and more demanding with ever-intensifying use of pixel and vertex shaders.

This is what we had to consider when selecting games and benchmarks to use for the test. In addition to well-known, pure benchmark applications such as 3D Mark 2003 and AquaMark 3, we also try to bring in the widest range of game genres, popular titles and games that are technically demanding with regard to 3D. For testing, we either used the benchmark mode built into the game or carried out gaming measurements « manually » with the help of the tool FRAPS. The margin of error for these measurements is very low, less than 1FPS.