Univa Expands Scheduling to Support new Intel® Xeon Phi™ Coprocessors

Native Support for Univa Grid Engine 8.1.3 Improves Application Performance

Hoffman Estates, IL, November 13, 2012 – Univa, the data center automation company, today announced the release of Univa Grid Engine 8.1.3 with native integration support of Intel® Xeon Phi™ coprocessors. By adding native integration support for Univa Grid Engine 8.1.3, it enables the management and scheduling of applications onto Intel® Xeon Phi™ coprocessors. This move further widens the gap between open source versions of Grid Engine and Univa’s quality controlled, commercial, and fully supported software.

"Univa's early support for HPC environments incorporating Intel Xeon Phi coprocessors, provide Univa Grid Engine users with an established, feature-rich, and commercially supported resource management solution," said Bill Magro, director of Technical Computing Software Solutions at Intel.

Through this integration, organizations can confidently utilize the full power of hybrid clusters designed with Intel Xeon Phi coprocessors with simple, easy-to-use functionality for scheduling jobs. The integration allows organizations to increase utilization on Intel Xeon Phi coprocessors by ensuring more workloads are processed, delivering faster time to solution. In addition, the integration provides insight into processor cores, memory and power usage on Intel Xeon Phi coprocessors allowing Univa Grid Engine customers to manage the usage and understand costs of running workload on Intel Xeon Phi.

The new Intel Xeon Phi coprocessors support provides:

  • New Resource Maps allow customers to specify exactly which Intel Xeon Phi coprocessors card a job will run on avoiding accidental scheduling of multiple jobs onto a single Intel Xeon Phi coprocessor.
  • New Resource Topology selection matches a requested Intel Xeon Phi coprocessor with the correct machine processor ensuring maximum performance by placing processor workload as close as possible to the Intel Xeon Phi coprocessors card on the PCI Express bus.
  • New Load and Metric collection tool allows for simple configuration of available Intel Xeon Phi coprocessor metrics ensuring Intel Xeon Phi coprocessors jobs run on the best (least loaded) Xeon Phi coprocessor.

"Hybrid architectures such as those built with Intel Xeon Phi coprocessors are clearly at the forefront of creating High Performance Computing (HPC) systems," said Fritz Ferstl, CTO of Univa Corporation. "They allow for responding to challenges such as scalability, energy consumption and cost effectiveness. Customers will, however, best be able to take full advantage of the benefits offered by Intel Xeon Phi coprocessors if access to these critical resources is well managed and seamlessly integrated with the workload and resource management software orchestrating the workflow through the HPC system. This is exactly what we are able to provide with our integration of Intel Xeon Phi coprocessors with Univa Grid Engine and why it will be a key differentiator for Intel Xeon Phi coprocessors users."

Univa Grid Engine is engineered, tested and fully supported by the Univa Grid Engine Team. Univa Grid Engine has been proven to perform in many of the largest production environments around the world.

More details on Univa Grid Engine can be found at http://www.univa.com/products/grid-engine

About Univa Corporation

Univa, the Data Center Automation Company, is the leading provider of automation and management software for computational and big data infrastructures. Our products and global enterprise support give our customers the power to manage their entire computational space, no matter how big or where it is deployed. Many of the leading brands in the world depend on Univa’s unsurpassed expertise, and premier services and support. Univa is headquartered in Hoffman Estates, Illinois, USA, with offices in Markham, ON, Canada; Austin, Texas, USA and Munich, Germany.

Media Contact: William McCormick Grayling Connecting Point (415) 442-4023 william.mccormick@graylingcp.com