Archive notice: This article was originally published on September 28, 2009. Links and embedded videos are preserved as part of the historical record.
In close cooperation with Microsoft, NVIDIA is working to optimize the use of NVIDIA Tesla GPUs for high-performance computing in Windows HPC 2008 operating-system environments.
NVIDIA has already developed several GPU-based applications on the Windows HPC Server 2008 platform: for example, ray-tracing applications for photorealistic modeling in the automotive industry. In doing so, NVIDIA worked with Microsoft to build a large Tesla GPU computing cluster.
“The combination of GPUs and CPUs demonstrates the enormous power and possibilities of multicore processing,” emphasizes Dan Reed, Corporate Vice President Extreme Computing at Microsoft. “NVIDIA’s collaboration with Microsoft around the Windows HPC Server platform will help researchers and developers use supercomputer performance for a wide variety of applications.”
Numerous enterprise applications already benefit from the massively parallel CUDA architecture on which NVIDIA’s GPUs are based: for example, in data analysis, machine learning or business intelligence, and in scientific applications involving molecular dynamics, finance or seismic research.
The CUDA architecture enables developers to use the computing power of the CPU and GPU together. Compute-intensive parts of an application use the GPU’s parallel computing architecture, while sequential parts of the application code run on the CPU.
“Science and research were the first fields to recognize the potential of the GPU, whose computing power enables a 20- to 200-fold acceleration of complex applications,” emphasizes Andy Keane, General Manager NVIDIA Tesla Business.
NVIDIA Tesla GPUs support Windows XP and Windows Vista on workstations and Windows Server 2003 and Windows Server 2008 in data centers. Systems with Tesla C1060 and Tesla S1070 GPUs are available from Cray, Dell, HP and Lenovo.
