updated 04:20 pm EST, Wed November 8, 2006
Nvidia's new GPU tech
On top of their GeForce 8800 announcement, NVIDIA has revealed CUDA, a new architecture that underlies the product. By delegating tasks more evenly across hundreds of small-scale processors, such as the shader units of the new GeForce card, programs may run up to 100 times faster than when running solely on a system's main CPUs, according to the company's claims. CUDA requires both development support due to its NVIDIA-specific nature as well as an intermediary driver that communicates with DirectX and OpenGL. However, the company adds that it will release a C-compiler to smooth adoption. While the introduction of CUDA is accompanied only by the GeForce 8800 at present, it will also be used in future versions of the workstation-oriented Quadro card as a way of speeding the calculations of 3D animators, scientists, and other professionals.