The next big leap in computing could be just around the corner, thanks to a project being conducted by UT’s College of Engineering and the State University of New York Polytechnic Institute.
The two are developing a computing system that mimics the nervous system to design a chip that would allow computers to operate at an even faster rate and calculate more complicated processing than simple binary-based computing.
It is known as a memristor-based dynamic neural network array, dubbed mrDANNA.
The Air Force Research Laboratory has provided $2.4 million in funding for the project, to be roughly split between the two institutions.
The mrDANNA is a hybrid neuromorphic computing system—meaning it mimics the nervous system, said Garrett S. Rose, associate professor of electrical engineering and computer science.
It is comprised of both nanoscale memristor devices and more conventional transistor technology integrated on a single silicon chip. This approach allows scientists to take advantage of new technologies that will increase functional density, encourage robust performance, and reduce power consumption.
Memristors—a combination of the words memory and resistor—are devices that allow for increasingly faster, more complicated processing than simple binary-based computing.
That improvement comes because memristors “remember” the history of power that has passed through them, giving them the ability to adjust to more favorable values based on changes in voltage.
“We are excited to play a role in designing the mrDANNA framework which could offer game-changing capabilities to the more advanced computers we may all use,” said Rose. “With research teams from both institutions collaborating, we will be able to develop and test these original designs, setting the stage for an exciting future of greatly expanded computing power and capability.”
The UT team, which includes Rose, John Fisher Distinguished Professor and personal computing icon Mark E. Dean and Professor James S. Plank—all of the Department of Electrical Engineering and Computer Science—will focus on the design of mrDANNA itself, while the SUNY team will turn its attention to the creation of chip prototypes.
Using their world-renowned NanoTech Complex in Albany, New York, SUNY researchers will test the limitations of such chips, confronting known challenges and those that emerge as scaling and production increase.
“We are thrilled to be able to investigate these novel technologies that can usher in a new era of more sophisticated and powerful computing,” said SUNY Associate Professor of Nanobioscience Nathaniel Cady. “We are grateful to the Air Force Research Laboratory for recognizing this trailblazing work, and we look forward to collaborating with the research team at UT.”
Cady stressed that in addition to the scientific gains themselves, the research and lab experience gained by students will also serve both institutions well.
The chance to take part in such important research came on the heels of previous work done by Dean with Professor Emeritus Doug Birdwell—both of whom are members of the National Academy of Inventors—as well as recent graduate Catherine Schuman, all of electrical engineering and computer science.
“We appreciate the opportunity the Air Force Research Laboratory grant provides to explore new and effective ways of building our neuromorphic system structures,” said Dean.
“Through this grant we will be able to develop neuromorphic computing devices and sub-systems that will improve system performance, increase density, and reduce power consumption by at least an order of magnitude compared to our present systems.”
—
CONTACT:
David Goddard (865-974-0683, david.goddard@utk.edu)