Nvidia Plans Assault On Supercomputers With Fermi GPU

Nvidia has released the details of its new graphics processor architecture (GPU) called Fermi at its inaugural GPU Technology Conference on Wednesday.

The company believes that Fermi, which shares its name with legendary Italian Physicist Enrico Fermi, will be the basis on which its next generation of GeForce, Quadro and Tesla processors will be designed

It is expected to offer the unique ability of a graphics processor to take up general processing tasks. The GPU architecture is essentially designed for high end general purpose parallel computing.

Fermi can be even scaled up to supercomputing levels and may herald Nvidia’s move into mainstream computing offerings. It is also highly suited for rendering complex images and running multithreaded applications and Fermi also offers extensive software support which ranges from OpenCL, C++ to DirectX.

Nvidia is recognised for its high end graphic processors that typically are used for gaming systems and the firm's latest move points to its continuous effort to improve its offerings and slide in high performance computing (HPC) arena.

It is interesting to note that Nvidia has changed it focus from desktop graphics to low end devices like mobile handsets and netbook computers, while it has also made a focused attempt to reach the high end server market with its Tesla processor range.

Our Comments

It always surprised me that Nvidia, which is essentially a one-horse company with its GPU, is more than twice the size of its arch-rival, ATI/AMD, which have a whole range of processors and GPUs. Nvidia first introduced the GPU more than a decade now and the Geforce range has set the standards for visual excellence.

Related Links

Nvidia plans new GPUs based on Fermi


Nvidia Puts On Graphic Power Display With Fermi


Nvidia fakes Fermi boards at GPU Technology Conference


NVIDIA details next-generation 'Fermi' GPU architecture