The relentless march of technological advancement requires corresponding breakthroughs in the systems, structures and frameworks upon which those advancements are achieved. With two new grants totaling $1.2 million, computer science researchers at Florida State University will work to build that crucial infrastructure, creating sophisticated programs that will help facilitate the next generation of computer science discovery.
Weikuan Yu, a professor in the Department of Computer Science, received two substantial grants from the National Science Foundation in quick succession, both of which aim for fundamental, systematic improvements in the complicated machinery of computer science research.
The first, a $700,000 grant Yu received in collaboration with fellow FSU computer science Professors David Whalley and Xin Yuan, will allow for the acquisition and development of an advanced software infrastructure the team has dubbed NoleLand.
NoleLand, which will be housed and operated at Florida State, represents a cutting-edge research infrastructure comprising processors, network equipment and high-bandwidth memory devices that will allow computer scientists to explore the most challenging questions facing high-powered computing and big data research.
NoleLand’s most important feature will be its nimbleness. Its many components are designed for swift and easy reconfigurability.
“They will be able to serve different computer science research groups and then be quickly reconfigured without much labor,” Yu said. “Those different configurations will be able to serve multiple different projects.”
Once it’s up and running, Yu expects that NoleLand’s benefits will expand far beyond the limits of the computer science department. He hopes the system will spur collaborative projects that explore the uncharted frontiers of computer science.
“These days we’re moving into a new era of data science and artificial intelligence, but to perform research in these fields we need big-scale computer systems,” he said. “Similar problems are being faced by many scientific applications sponsored by the NSF and the Department of Energy. We want to play a role in addressing those questions.”
The second grant, a $500,000 award to Yu and Sarp Oral, an adjunct professor in the Department of Computer Science, will provide funding for a research framework called Ephemeral Coherence Cohort (ECC), a project designed to unlock the challenges that hamstring the world’s most powerful supercomputers.
As supercomputers continue to grow at a persistently accelerating rate, there’s an emerging need to make sure the applications supported by those supercomputers are optimized to run at the highest possible performance.
Supercomputers use racks of discrete servers — sometimes numbering in the tens of thousands — to coordinate processes and access memory at unimaginably high speeds. But issues with synchronization and coherence often bedevil these room-sized machines and disrupt their complex operations.
“Put simply, the idea is to avoid dealing with coherence and synchronization needs across the entire system,” Yu said. “Instead, we will partition the system into multiple small cohorts of computer servers. That way, you can address the synchronization or coherence challenges with much smaller granularity, thereby enabling better performance from your applications.”
With these most recent grants secured, Yu and his group have now received nearly $3 million in research funding since his arrival at FSU three years ago. He said that figure represents both the promising growth of computer science research at the university as well as a significant boost for his team, which is growing steadily and expanding its research portfolio.
“It’s helped my group grow to a dozen or so graduate students who are working tirelessly on several cutting-edge research projects,” he said. “I feel very fortunate to have had the chance to work with so many talented students and colleagues. We are determined to embrace more challenges and usher in next-generation computer science research.”