SCIENCE

SGI announced availability of the latest NVIDIA Tesla 20-series GPUs in all of its server product lines. As a leader in acceleration and design-to-order high performance computing (HPC) solutions, SGI is uniquely positioned to bring customers hybrid solutions of industry-standard microprocessors and GPU accelerators that solve critical technical computing problems.

Last year, SGI deployed a GPU-equipped Altix UV system at the University of Tennessee’s Center for Remote Data Analysis and Visualization (RDAV), bringing a machine with 4 TB of available shared memory to GPU computing. SGI has also designed custom GPU solutions for customers through its Custom Engineering group using SGI Rackable products as building blocks, providing customers choices among the rich motherboard, processor, and memory options available today. SGI also offers the SGI Prism XL system, a custom accelerator solution, that supports NVIDIA GPUs and has the capability to scale to 10s of petaflops. SGI added support for NVIDIA Tesla GPUs in its Altix UV 10 as well as Rackable servers C2112-4TY14, C3108-TY11, C1103-TY12, C1104-2TY9, and H2106-G7.

“Our new Altix UV 1000 with 1,024 Intel Westmere-EX cores, 2 TB of memory and NVIDIA Tesla S2050 GPUs all under a single operating system provides a unique development platform for our researchers,” said Professor Sylvio Quezado, vice director of the International Institute of Physics at Federal University of Rio Grande do Norte, Brazil. “We have visiting researchers from around the world doing multidisciplinary research with this system.”

“Interest in GPU computing is growing fast,” said Bill Mannel, vice president of product marketing at SGI. “The latest NVIDIA Tesla M2090 GPU coupled with their CUDA Toolkit 4.0 software release brings superior capability into the hands of developers. At SGI, we offer among the most extensive options available for GPU programmers, from access to large shared memory in our Altix UV line to the extreme flexibility offered with our Rackable rackmount line, and everything in between.”

“Adding GPUs to a server delivers application acceleration up to 10 times – in some cases even more, and this is having a profound effect on many important industry segments,” said Andy Walsh, director of marketing, Tesla business at NVIDIA. “SGI’s new Tesla-enabled servers give developers and researchers a powerful and flexible platform upon which to innovate.”

New NVIDIA products, including the Tesla M2090 and the NextIO vCORE Express 2090, along with the CUDA Toolkit 4.0, will be available in the Altix UV, the Altix ICE 8400, Rackable, and SGI Prism XL server systems. In addition, SGI has upgraded its Accelerator Execution Environment (AEE) available for SGI Prism XL to the CUDA Toolkit 4.0.

Availability

All configurations are available now.