Silicon Mechanics now offers NVIDIA Tesla K20 GPU Accelerator

Designed for high throughput of both dual- and single-precision workloads

Silicon Mechanics, Inc., a leading manufacturer of rackmount servers, storage, and high-performance computing hardware, announces that it now offers the new NVIDIA® Tesla® K20 GPU Computing Accelerator in its hybrid architecture server product line. The new ultra-efficient Kepler-based GPU is designed for high throughput of both dual- and single-precision workloads. It is ideal for climate and weather modeling, computational fluid dynamics, computer-aided engineering, computational physics, biochemistry simulations, and computational finance.

Developers and researchers can enjoy up to three times more performance per watt with the NVIDIA® Kepler™ compute architecture, powered by CUDA®, the world's fastest and most efficient high-performance computing (HPC) architecture. With innovative computing technology and features, it is applicable to a broader range of scientific computing applications and makes hybrid computing more accessible for application developers and researchers.

With teraflops of single­ and double-precision performance, NVIDIA® Kepler GPU Computing Accelerators are companion processors designed for double-precision applications and the supercomputing market. The Tesla K20 delivers three times the double-precision performance compared to the previous generation Fermi­based Tesla M2090, in the same power envelope. The K20 GPU Accelerator features a single GK110 Kepler GPU that includes the innovative Dynamic Parallelism and Hyper­Q features, which boost performance and power efficiency, and deliver record application speeds.

"With more than one teraflop peak double-precision performance, this GPU Accelerator is ideal for the most aggressive high-performance computing workloads," said Sumit Gupta, senior director of Tesla business at NVIDIA. "Since it is not limited to single-precision applications, it is also a well-rounded offering that can be used by more people for more tasks."

With a Silicon Mechanics hybrid architecture server equipped with the NVIDIA® Tesla® K20 GPU Accelerator, applications run significantly faster, as users can process large datasets with a massively multi-threaded computing architecture.