Imagine if one single desktop computer contained the processing power of an entire data center.
That notion could become reality within the next five years, speculates Jing Li, the Dugald C. Jackson Faculty Scholar and an assistant professor of electrical and computer engineering at the University of Wisconsin-Madison. With support from a National Science Foundation CAREER Award, Li is working to make computers much more powerful by completely reimagining the basic principles and devices underlying the machines.
“We’re rethinking computers from the ground up,” says Li. “It’s not about engineering. It’s not about the components. It’s about a new computing model.”
Computers have relied on the same fundamental mathematical concepts since Alan Turing presented the notion of a universal machine in 1936. Although processing power has increased by leaps and bounds over recent decades, computers still struggle mightily with some types of complicated data—especially the broad class classified mathematically as graphs, which pop up in social networks, maps, or any other enormous collections of information that contain hidden relationships.
Computers have trouble tracing graphs, like the several degrees of separation that can link two individuals within a social network. Every hop from friend to friend (or node to node) represents an unknown, and the organizations of the electronic brains inside computers cannot efficiently handle such unpredictable random-access data patterns.
Graphs are so tricky that they represent one of the grand challenges for modern computing. To solve the graph problem, Li and her students are fundamentally rethinking how computers work.
“Previously we were doing computation in one dimension. From the very original computer with a single core to today’s advanced multicore applications, we have been doing things better in a single dimension,” says Li. “This is trying to explore higher dimensions.”
Higher-dimensional computing needs more than new math. Li envisions machines running on entirely different architectures. Her project is subdivided into three pillars: hardware, software, and algorithms, with the goal of creating a vertically integrated computing ecosystem.
Importantly, however, Li plans to keep the top-level interfaces unchanged so that programmers won’t have problems adapting to the new systems.
“It’s not a very strange, very weird computer that you wouldn’t want to use,” says Li. “From a programmer perspective, it’s a transparent speed-up; you won’t even notice, except your program is running so much faster compared to traditional data centers.”
Although completely overhauling how computers function may seem like a tall order, Li and her students have already generated promising preliminary results. Additionally, through her close collaboration with colleagues at IBM, Li is already working on reducing potential barriers to commercialization and distribution.
“It’s very hard, but it’s very exciting,” says Li. “We are actually trying to connect the science together with engineering and make bigger transformative innovations in the world.”
Li believes that the field is poised for tremendous breakthroughs, and that university researchers are perfectly positioned to help bring about the next big innovations. Yet making tomorrow’s computers a reality will require some big-picture thinking.
“We stay in our comfort zones,” says Li. “Today we are in a time period where we cannot stay in our own little world.”
In her classes, Li is challenging students to step outside their comfort zones by implementing open-ended, interactive projects. She hopes her efforts will start instilling students with creative, big-picture thinking abilities early on.