Over the past 80 years, computers have progressed from room-sized collections of vacuum tubes and mechanical relays to the tiny digital devices we now carry in our pockets. Along with the hardware changes, computer architecture—the organization and design of computer systems and the way components work together—has also evolved, but not always in the most straightforward direction.
Many approaches underpinning the way modern computers operate are based on past constraints, like the types of materials and components available, computing power and operating software. Now, as quantum computing, superconducting materials, and other big changes arrive, it’s worth evaluating whether different architectures might make better sense.
“My research seeks to challenge long-standing assumptions in computer architecture,” says George Tzimpragos, who joined the Department of Electrical and Computer Engineering at the University of Wisconsin-Madison as the James E. Smith Assistant Professor in January 2025. His favorite types of inquiries question the status quo; for example, “Are binary codes and Boolean logic always the most efficient approach to computation?” or “Are conventional memory structures optimal across all technologies?”
Tzimpragos earned a PhD in computer science from the University of California, Santa Barbara, and a master’s degree in electrical and computer engineering from the University of California, Davis. He completed his undergraduate studies in electrical and computer engineering at the National Technical University of Athens in Greece. From 2022 to 2025, Tzimpragos was an assistant professor in electrical engineering and computer science at the University of Michigan.
Among his many ideas, Tzimpragos advocates rethinking the established analog-digital boundary in data representation and processing. He proposes that time should be treated as a computing resource—like its role in the brain—rather than merely a performance metric or functional property. This perspective, he argues, offers transformative possibilities, including new methods for embedding computational capabilities into sensors, scaling up pulse-based and quantum computing, and advancing hardware security and formal verification.
When it comes to computer memory, Tzimpragos suggests revisiting ideas from historical machines like ENIAC, the first general purpose computer developed in 1945, where delay lines served as storage media. While that approach is unreasonable for semiconductors, new superconducting electronics could take advantage of the technique. “This isn’t just a solution for superconductor memory,” he explains, “but a paradigm shift in computer architecture. It’s a departure from current trends of minimizing data movement by adding transistors. Instead, we capitalize on the inexpensive data movement in superconducting interconnects to maximize energy efficiency and minimize hardware complexity, bringing this technology closer to practicality.”
These concepts, Tzimpragos believes, could have wide application. “Many of our current technologies have stood the test of time because they are simple and efficient. Other ideas, though equally elegant, may have simply failed to align with the devices of their era,” he says. “Now is the perfect moment to revisit them, and UW-Madison offers an excellent environment for such explorations.”
The university, he says, is in a unique position, combining a rich history of pioneering work in computer architecture with a recent influx of talented young researchers in electrical and computer engineering, physics, and computer science. In addition, Tzimpragos admits a personal attachment to UW-Madison. Over his academic career, he has both drawn inspiration from and collaborated with Professor Emeritus James E. Smith, a luminary in computer architecture. “It is an honor to hold a professorship named after him. When I think of computer architecture and innovative, out-of-the-box thinking, Jim is the first person who comes to mind,” Tzimpragos says.
Top photo by Joel Hallberg