University to launch updated data analytics system

Leila Ruiz

The University’s Texas Advanced Computing Center will release a new, high performance visualization and data analytics system known as Maverick in February.

Originally slated for deployment in January, Maverick is the result of the center’s partnership with technology companies Hewlett-Packard and NVIDIA. Maverick will serve as the replacement to Longhorn, the collection of software visualization systems currently operating at the center.

According to Kelly Gaither, principal investigator of the Maverick project and the center’s director of visualization, Maverick will be used for scientific research and was designed as an interactive, remote visualization and analytics tool. The system will assist in analyzing mass amounts of scientific data alongside the University‘s supercomputer Stampede, launched to provide an interactive environment for researchers last March.

The research and information collected and analyzed with Maverick will be publicly accessible by the scientific and engineering community. Though initially designed for researchers, Maverick will be available to students through their advisers, according to astrophysics professor Karl Gebhardt.  

Gebhardt said the purpose of systems like Maverick is the ability for users to obtain huge amounts of data, access it quickly and efficiently manipulate that data with software tools. Maverick will be an improvement upon Longhorn in all of these facets.

“I have been using [the center’s] resources to study black holes, including the largest black holes in the universe, and dark matter around galaxies,” Gebhardt said. “Maverick will be essential for our future work with HETDEX, the Hobby-Eberly Telescope Dark Energy Experiment. We will generate many petabytes of data, with the goal of understanding how the universe expands over time. These results will allow us to understand the formation, evolution and long-term fate of the universe.”

Michael Teng, a computer science graduate student, said Maverick’s ability to collect and analyze large amounts of data is useful during scientific investigations.

“[Maverick] utilizes a lot of graphics processing units to accelerate the visualization of large amounts of data,” Teng said. “A lot of scientific problems have to do with the movement of particles, or something that has millions of parts. The best way to demonstrate what happens in the simulation is to play a video of what happens using the system.” 

Maverick contains 132 NVIDIA Telsa K40 graphics processing units, or GPUs, according to Scott Misage, high performance computing engineering director for Hewlett-Packard. Computer sciences senior Craig Yeh said the large amount of GPUs aids in the speed of data analysis. 

“[GPUs] are mostly used for problems that are easily parallelized, allowing for faster calculations,” Yeh said. “Additionally, you can use the GPUs to render the data into videos or stream live visualizations to the researchers and allow for interactivity.”