August 9, 2016 - By Lorena Anderson, University Communications - Faculty members from the Electrical Engineering and Computer Science, Chemistry and Chemical Biology and Mechanical Engineering groups are forming the newest research center at UC Merced, seeking the best ways to conduct high-performance computing.

With a hardware donation from the NVIDIA Corporation and through the company’s GPU Centers Reward Program, professors Dong Li, Christine Isborn and Yanbao Ma are co-leading the effort to optimize the use of data and explore new computer programming methods.

Graphics processing units — most commonly used in video games for faster and more realistic play — are becoming the standard for high-powered computing, as well. More companies are looking to hire people who have experience with the hardware, and Li said that’s one reason the center is important to the campus.

“We’re going to be preparing graduate and undergraduate students for future careers,” he said.

NVIDIA, based in Santa Clara, has 230 research centers at universities around the world through its GPU Research Centers Program. The company designs hardware and software for the gaming market, professional visualization, data centers and automotive applications, including self-driving cars.

“Being a GPU center helps UC Merced gain more recognition as a place where exciting research in high-performance computing is happening,” Isborn said.

GPUs use large numbers of processing cores to solve computational problems faster and more efficiently. GPU-accelerated computing leverages parallelism to deliver dramatic increases in performance for scientific, artificial intelligence, deep learning, graphics, engineering and other demanding applications.

The new center aims to harness the power of GPUs to solve the increasing demands for efficient data processing.

But as modern memory systems on GPUs become increasingly complex and sophisticated, it will be critical to have manageable memory systems that programmers are able to work with efficiently. The new center will focus on a software-manageable system, as well as study the implications of some emerging memory techniques.

The professors and their students plan to work on projects that fit within their areas of expertise — such as Isborn’s large-scale quantum chemistry calculations, which focus on modeling how molecules interact with light and absorb energy — to answer questions related to GPU computing overall.

They’ll look at performance modeling techniques; data placement in the memory systems; and using GPUs for realistic scientific and engineering applications using new techniques developed through the research.

The new center is highly interdisciplinary, Li said, but will be housed within the School of Engineering.

NVIDIA’s GPU Research Centers are institutions that embrace and utilize GPU accelerated computing across multiple research fields, and are at the forefront of some of the world’s most innovative scientific research, the company said.

Li said students are already working on hardware programming and efficiency studies and techniques, and he plans to take students to major GPU conferences each year, hopefully to present as well as network.

“NVIDIA includes us in the seeding program for each major new GPU — it may not be even for sale yet, and we get to use it before everyone else,” Li said. “We’re very pleased to be able to start this center, because it puts us at the forefront of GPU research.”
Source: UC Merced