Gill Wins National Science Foundation CAREER Award

A University of Kansas professor has received one of the National Science Foundation‘s most prestigious awards for junior faculty members.

Andy Gill, an assistant professor in electrical engineering and computer science, earned an Early Career Development (CAREER) Award.  National Science Foundation CAREER Awards support new faculty who have shown exceptional promise in teaching and research. Dr. Gill receives a five-year, $521,000 award to develop software that explores tradeoffs in the design of high performance computing (HPC) systems.

“I am delighted to receive this award,” Dr. Gill says. “The CAREER funding will allow KU to investigate new ways of writing high performance systems, building bridges between existing tools that will allow a wider range of programmers to develop high performance solutions. We want high performance computing to be accessible to non-specialists, not just computer scientists and computer engineers.”

Designers are reaching the limits of miniaturization. It is much harder to continue making smaller, faster processors for computers, smartphones, and other electronic devices.  Instead, researchers are linking computers together to create powerful research platforms. If used efficiently, the new approach offers enormous computing power. These multiple cores, or computers, crunch data in a fraction of the time it would take using individual desktop computers.

High performance computing systems.allows complex problems to be solved in hours or days rather than years because its processes different parts of a problem simultaneously. It does not solve a problem in a step-by-step fashion, but rather works in parallel. Using HPC resources, researchers are attempting to solve complex problems, such as the causes of autism and climate change.  

Dr. Gill says an efficient HPC platform can run many, many times faster than one slowed down by disorganized processes. His project, “CAREER: Filling the Gaps in Domain-Specific Functional-Based Solutions for High-Performance Execution,” will create software to make efficiency more accessible in high performance computing.  Just as power steering makes it easier for drivers, the KU software will allow designers to easily assess whether they are “squeezing the maximum performance out of the hardware.”

To boost performance, researchers must consider a series of tradeoffs in power, memory, and storage.  For example, Dr. Gill says it is sometimes more efficient to use only a few cores rather than the thousands available, because for some problems, communication among cores costs more that the computations themselves. Dr. Gill compares the proposed tools to using Google Maps, where users easily compare the costs of walking, taking the bus, or driving a car to reach their destination. In much the same way, the KU tools will allow programmers to compare many different ways of using computing resources to complete complex tasks. By quickly comparing solutions, users can spend more time focusing on what a program does, rather than how it is executed, Dr. Gill says.

In addition to using surfaces like GPUs and FPGAs on established problems, KU graduate and undergraduate students will use the software to test their hardware/software designs for international competition. For the first time, KU students will enter the Formal Methods and Models for System Design (MEMOCODE) contest.  Previous student competitions include locating millions of sequences in the human genome and extracting 3D information from digital images, with teams having a month to create the best possible high performance solution.

“The problems posed by MEMOCODE are an ideal test bench for our ideas. This is a great opportunity to compete against others, and gain objective feedback about how our tools perform in practice,” Dr. Gill says.

In 2013, Dr. Gill received a Distinguished Visiting Fellowship from the Scottish Informatics and Computer Science Alliance (SICSA). He gave a series of lectures on his research at leading Scottish universities.

He has earned multiple teaching awards at KU. Graduating EECS seniors selected him for the Harry Talley Excellence in Teaching Award in 2009.  The following year, he was an honoree at the Celebration of Teaching Reception hosted by KU’s Center of Teaching Excellence.