Altair introduced PBS Analytics 12.0, an advanced version of its data analysis and visualization tool for high-performance computing. According to the company, the solution has been rebuilt to meet a growing demand for increasingly comprehensive analysis of HPC usage with sophisticated charts and dashboards, tracking jobs and resources, implementing chargeback and showback accounting, performing root-cause analyses and visualizing power use.
"Our latest PBS Analytics software is brand new, both inside and out," said Bill Nitzberg, CTO for PBS Works at Altair. "We've incorporated a new, high-performance database, re-engineered our data collectors to be more robust and completely redesigned the user experience. PBS Analytics 12.0 provides the charts users need out-of-the-box, delivers snappy performance, and scales up to support millions of job records."
A new utilization chart shows the percentage of utilization adjusted for configuration of the system, accounting for the addition or deletion of nodes over time. This capability can improve overall IT efficiency and justify IT investments by illuminating actual usage. It also enables discovery of usage hot spots, pinpointing where adding additional resources will have the greatest impact and usage "deserts," where resources can be replaced.
In addition to the out-of-the-box charts, PBS Analytics 12.0 is more customizable. A new chart designer makes creating custom charts simple, and the solution has added support for site-defined PBS Professional resources, such as GPUs, licenses, and scratch space. Finally, administrators can define totally customized metrics, adding their own enterprise-specific measures directly into the data stream.
The company also unveiled Display Manager, a new portal for remote visualization of extremely large models developed from Big Data resources. Display Manager, which will become generally available in the near future, enables users to run graphics-intensive applications on their cluster while visualizing those applications on their desktop.
Because applications run directly from the data center, users can take advantage of shared, GPU-enabled hardware in the data center, and use their Web browser to view, edit, interact with and control the graphical apps remotely. As a result, the enterprise can buy fewer high-end graphics systems, and avoid purchasing a high-end graphics desktop for each engineer. Display Manager enables sharing of HPC-class systems housed in the datacenter (or in an HPC cluster) among all the engineers in the enterprise.
In addition, multiple users can collaborate directly from within Display Manager by sharing sessions. Authorized users can all view and control a single session without downloading any files. It can be used as a standalone technology or can be seamlessly integrated with Compute Manager for enterprises with large HPC clusters.
For more information, visit Altair.
Sources: Press materials received from the company and additional information gleaned from the company's website.