San Jose, California (PRWEB) February 22, 2014
Bright Computing, the leading, vendor-independent provider of management solutions for clusters and clouds, announces today that the Swiss National Supercomputing Centre (CSCS) has evaluated Bright Cluster Manager® in order to assess its strengths and weaknesses.
CSCS evaluated Bright Cluster Manager. They found Bright met their needs in providing burden-free support for accelerators and coprocessors. CSCS's evaluation of Bright is available for download.
CSCS built a 20-node cluster to evaluate Bright Cluster Manager. The cluster included both NVIDIA GPU accelerators and Intel Xeon Phi coprocessors so that they could investigate the capabilities of Bright Cluster Manager with hybrid architectures. CSCS confirmed that Bright supports Kepler-architecture GPUs (NVIDIA Tesla K20X GPU accelerators) and the latest version of the CUDA toolkit (CUDA 5.5). Their evaluation also confirmed that Bright significantly eased the provisioning process of systems with Intel Xeon Phi. The evaluation report noted Bright’s integration with the entire Intel software stack and the Slurm workload manager. Bright-integrated Slurm lets users take advantage of multiple execution modes in this CPU-coprocessor architecture. CSCS also validated that both accelerators and coprocessors can be monitored and managed alongside CPUs in their evaluation cluster.
According to the CSCS evaluation, Bright Cluster Manager allows one or two systems administrators to effectively and efficiently manage complex heterogeneous systems or multiple clusters.
“There’s a growing interest in production use of hybrid-architecture systems in our HPC centre,” stated Hussein Harake of CSCS. “This means we need to be able to provision, monitor and manage systems with accelerators and coprocessors with minimal administrative overhead. Our evaluation of Bright Cluster Manager demonstrated that it met our needs in a satisfactory way.”
CSCS’s evaluation is available for download from their Web site.
About Bright Computing
Bright Computing specializes in management software for on-premise HPC, Hadoop, storage, database and workstation clusters, as well as the seamless extension of these clusters into the cloud. Its flagship product — Bright Cluster Manager — with its intuitive graphical user interface and powerful cluster management shell, makes clusters of any size easy to install, use and manage, including systems combining processors with accelerators (e.g., NVIDIA GPUs) or coprocessors (e.g., Intel Xeon Phi). Bright's minimal footprint enables systems to be utilized to their maximum potential, from departmental Hadoop clusters to large-scale supercomputers. Bright Computing partners include Amazon, Cisco, Cray and Dell, while Boeing, ING Bank, NASA, Roche, Saudi Aramco plus Stanford University and Tokyo Institute of Technology are examples of Bright customers. Bright Computing is a Red Herring 2013 Top 100 North America Award winner, the Deloitte Technology Fast50 “Rising Star 2013” Award winner, the Main Software 50 “Highest Growth” Award winner, and Bright Cluster Manager was a "Best of Show Award" winner at Bio-IT World 2013.
Pictures and screenshots of Bright Cluster Manager: http://www.BrightComputing.com/Newsroom.php
For more information
Bright Computing, Inc.
Mr. Lionel Gibbons
2880 Zanker Road, Suite 203
San Jose, CA 95134
Tel: +1 408 337 6076