ENGINEERING SCHOOL
PARTNERS WITH MICRON ON
GROUNDBREAKING PROCESSOR
A WORLDWIDE CENTER FOR
ADVANCED PROCESSING RESEARCH
M
icron Technology finds itself in an unusual position for a company known for innovations in computer memory and
storage. It has entered new territory by developing a processing chip capable of opening the door to advances in fields
like bioinformatics, video/image analytics and network security. It turned last year to computer scientists at the University
of Virginia with expertise in novel processor architectures and relevant applications to build a worldwide community
devoted to realizing the Automata Processor’s potential. The Center for Automata Processing, cofounded by Micron and the
University, is a virtual collaboration of universities, companies and government agencies.
“Our goal is to foster collaborations among industry and academic researchers to advance the field of automata computing,”
says Professor Kevin Skadron, chair of the Department of Computer Science and the center’s director. “We hope to facilitate
teaming on proposals and provide industry with a source of academic expertise in solving big-data problems.” He points out
that the center will also generate research opportunities for University graduate and undergraduate students and could enrich
existing courses in areas such as processor design and data mining.
The Automata Processor addresses a critical drawback of conventional processors, one that impedes the progress of
research in scores of fields. Computers are very good at producing exact matches, combing through millions of records
in milliseconds, for example, to find the precise match to DNA lifted from a crime scene. But when it comes to imprecise
matches — identifying commonalities in strings of genetic code from two different persons — they slow to a crawl, undone
by the limited bandwidth between processor and memory.
Micron used insights gleaned from its many years as one of the world’s leading providers of computer memory to
overcome this problem. Conventional processor designs require that instructions and data be fetched from and stored in
memory, while the Automata Processor does not, thus avoiding memory-access bottlenecks.
“The Automata Processor preconfigures each of its 1.5 million pattern-matching elements in advance, so that a huge number
of parallel matching operations can occur simultaneously and instantaneously for each input item,” Skadron says. “This sets the
stage for high-speed, comprehensive search and analysis of complex, unstructured data streams.” Thanks to this architecture,
the Automata Processor can not only help researchers find matches among seemingly dissimilar items, but it could also help
them deduce the factors that produce those patterns.
Skadron credits Electrical and Computer Engineering Professor Mircea Stan and Materials Science and Engineering
Professor Stuart Wolf — who are associate directors — for being instrumental in securing the center for the University.
Thanks to their expertise in memory technologies, both faculty members have longstanding research collaborations with
Micron. This positioned them to introduce Micron to Engineering School expertise in processors and in applications, such
as image analytics, where the Automata Processor would be highly effective.
“This is an extraordinary opportunity,” says Terry Leslie, Micron’s director of business
development for Automata processing. “The University of Virginia’s partnership is a
critically important part of building an ecosystem for this exciting new architecture.”
The center has already attracted interest across the University. In addition to faculty
from computer science as well as electrical and computer engineering, there are a number
of interdisciplinary collaborations underway with faculty from the departments of
biomedical engineering, biochemistry and molecular genetics, systems and information
engineering, and public health sciences. “The University’s Institute for Advanced
Technology in the Humanities is also participating in the center,” Skadron says. “These
diverse collaborations are an indication of just how powerful and broadly applicable the
Micron’s new Automata Processor can
Automata Processor is.”
perform massively parallel operations
For more information on the center, visit www.cap.virginia.edu.
much faster than other chips.
U.Va. ENGINEERING UNBOUND
7