NVIDIA at present launches its next generation CUDA GPU structural design, codenamed “Fermi”. The most recent structural design by the graphics leader is the "foundation for the world’s former computational graphics processing units (GPUs), conveying advancement in both graphics as well as GPU computing." So what does Fermi convey? Well, according to NVIDIA, it is a aspect situate that accelerates performance on a wider range of computational applications than yet earlier than.

With more than three billion transistors plus featuring up to 512 CUDA cores, Fermi conveys supercomputing features with performance at 1/10th the cost as well as 1/20th the power of usual CPU-only servers.

Fermi makes GPU as well as CPU co-processing all-encompassing by addressing the full- range of computing programs. Intended for C++ plus obtainable with a Visual Studio development surroundings, it makes parallel programming simpler as well as accelerates performance on a wider range of applications than yet ahead of – with vivid performance acceleration in ray tracing, physics, predetermined component analysis, high-precision scientific computing, sparse linear algebra, sorting, along with search algorithms.

Fermi features several major innovations:

• 512 CUDA cores
• NVIDIA Parallel DataCache technology
• NVIDIA GigaThread engine
• ECC support

Name: 18.jpg Views: 627 Size: 20.6 KB