Microway Inc. (Plymouth, MA), a developer of of HPC (high-performance computing) workstations, clusters, servers, InfiniBand-based switches, and data storage solutions, has announced SimCluster, an off-the-shelf integrated cluster solution designed for life science researchers. SimCluster, which includes NumberSmasher 1U NVIDIA Tesla GPU (graphic processing unit) servers for scientific applications, is said to be fully optimized to simulate large size models and achieve higher accuracy. This, in turn, is said to allow research groups to move from GPU-accelerated workstations to the SimCluster configuration. In synch with the SimCluster announcement, Microway has launched a "Test Drive" program that enables interested users to benchmark their applications on a SimCluster.
The Microway SimCluster with four NumberSmasher 1U Tesla GPU nodes.
"Molecular dynamics, quantum chemistry, and bioinformatics applications are particularly well suited for massive parallelism,” said Stephen Fried, Microway's president and CTO, in a press statement. “Many of these applications have been ported and run efficiently on the 512-core NVIDIA Tesla GPUs in SimCluster. With speedups exceeding 10x on many applications, scientists can leverage Tesla GPUs to achieve smaller footprints with increased computational power. Microway's Tesla-based platforms offer the ideal foundation for this accelerated research."
SimCluster features NVIDIA Tesla M2090 GPUs and Intel Xeon 5600 series CPUs connected by InfiniBand. According the company, the system, like all its Microway GPU-based platforms, is delivered fully tested and integrated with Linux or Windows, NVIDIA Tesla Bio Workbench applications, and the NVIDIA CUDA parallel programming environment. The company also offers Bright Cluster Manager software as an option to the SimCluster. Bright Cluster Manager is said to provide IT administrators easy-to-use cluster management and GPU monitoring that enables them to install, manage, and use multiple clusters simultaneously, without the need for expert knowledge of Linux or high-performance computing.
Microway's SimCluster "Test Drive" program provides interested users access to a benchmarking cluster so that they can test their GPU-enabled applications and benchmark their speedups. After registration, researchers simply log on the cluster, upload, and run their AMBER, MATLAB, or NAMD models. Benchmarking of custom CUDA code is available with advanced notice.
The Microway NumberSmasher 1U Tesla GPU Node with two Tesla M2090 GPUs.
Additionally, life researchers can purchase the exact SimCluster configuration used to benchmark their applications: a four node, eight GPU configuration, said to be available for shipment within 48 hours after test. Larger scale, custom configurations based on the tested cluster are deliverable in three to four weeks, depending on specific user requirements, according to the company.
The company also offers WhisperStation workstations for life science applications. These workstations, which are equipped with four NVIDIA Tesla C2075 GPUs, are said to be suitable for smaller platforms and budgets.
Microway's Tesla solutions are delivered fully integrated with CUDA SDK and Linux or Windows. For more information on Microway's SimCluster with NVIDIA Tesla M2090 GPUs, go here.
Learn more about the SimCluster test drive program.
Download the Bright Cluster Manager software brochure.
Visit Microway's library of technical papers.
Learn more about Tesla GPUs.
Go the Microway WhisperStation workstation landing page.
See why DE's editors selected Microway's SimCluster with NVIDIA Tesla M2090 GPUs as their Pick of the Week.
Sources: Press materials received from the company and additional information gleaned from the company's website.