Computing royalty, including Ivan Sutherland, the father of computer graphics, and Wesley A. Clark, the designer of the world’s first personal computer, will gather for a symposium from 1-5:30 p.m. March 26 in Whitaker Hall Auditorium.
As part of the University’s 150th anniversary, participants will honor time by contemplating how computing can evade time as the industry prepares to go clockless.
The Department of Computer Science and Engineering will present “Clockless Computing: Coordinating Billions of Transistors,” to honor both the University’s sesquicentennial and the 30th anniversary of the completion of the seminal project on macromodule computer design — work that anticipated current endeavors to go clockless, or asynchronous.
This sort of computing marks an important change from present systems, which are based on a regularly ticking clock, said symposium organizer Jerome R. Cox, Sc.D., senior professor in computer science and engineering.
“Clocked technology is inadequate to deal with very large integrated circuits,” Cox said. “Systems of the future will certainly have clockless technology or a blend of clocked and clockless types.”
The key reason that clockless computing is essential to computing’s future is that engineers now are placing literally billions of transistors on computer chips that are roughly the same size as those that contained only thousands of transistors decades ago.
To comprehend clockless computing, consider the analogy of a system of traffic lights programmed to go green on a regular, clocked schedule. This would entail many hundreds of lights in synch, say in Manhattan
Imagine, now, another system of billions of lights (similar to billions of transistors), some of them far apart, scattered all over the world. There is no reason to have them all synchronized.
“In most computer chips today, everything marches to the beat of the same drummer,” Cox said. “The cost in design time, chip power and circuit area devoted to clock distribution gets larger and larger as the number of transistors gets larger. Another way must be found instead of lockstep throughout billions of transistors.”
Clark was a University faculty member from 1964-1972 and has been a full-time consultant since then.
“I expect that the symposium will assess the early macromodular work in the much broader and more difficult context of today’s clockless-system developments,” Clark said. “The taming of unplanned events in enormous ‘state-transition spaces’ still remains the key challenge in clockless system design.”
Designers are developing chips with diverse clocked domains, breaking tasks up into multiple domains. Clockless takes that concept a step further.
Consider the traffic light example again: Imagine sensors for traffic lights that change the colors according to local conditions, enabling freedom from the central clock. A clocked system must wait until the tardiest signal in the whole bunch makes its transition; a clockless system allows for signals to switch without unnecessary waiting for others.
Clockless computing provides numerous advantages. It facilitates easier power supply design, reduces noise that a clocked system creates and allows parts of a system to become idle, reducing power requirements.
“Theoretically, it can lead to faster systems, and we’re on the threshold of being able to realize that theoretical goal,” Cox said.
Robert Reuss, program manager with the Defense Advanced Research Projects Agency, said there is great interest in clockless computing from both industry and the Department of Defense perspectives.
“The appeals are lower operating power, faster performance and reduced electromagnetic interference on the chips,” Reuss said. “A challenge is the complexity of designing very large chips that are approaching 1 billion transistors. Clockless logic has the potential to impact these issues.
“From the Department of Defense perspective, we are all the more interested because we do not have the resources to devote to a thorough and long chip-design cycle. So, the DOD is interested in how clockless logic might help us in regards to economy of scale.”
In 1962, future University computer science engineers Clark and the late Charles E. Molnar and others in Massachusetts Institute of Technology’s Lincoln Laboratory Group designed the Laboratory Instrument Computer (LINC). With its digital logic and stored programs, the LINC has been recognized by the IEEE Computer Society as the world’s first interactive personal computer.
In 1964, Cox founded the Biomedical Computer Laboratory at the Washington University School of Medicine. That same year, a team of engineers headed by Clark and Molnar formed the Computer Systems Laboratory at Washington University.
Together, Biomedical Computer Laboratory and Computer Systems Laboratory engineers brought about profound changes in the nature of laboratory and clinical computing worldwide.
Sutherland, vice president and fellow of Sun Microsystems, will provide the keynote address. In 1988, he received the A.M. Turing Award, the highest honor of the prestigious Association for Computing Machinery. His acceptance talk was titled “Micropipelines” in which he described how computer-system designers are constrained by the clocked-logic framework.
The time required to design systems grows annually, but Sutherland’s vision sees micropipelines and clockless computing removing the barriers to the design of ever-larger and more-capable systems.
Other pioneers in clockless computing will speak at the symposium, providing a glimpse of future trends in computer engineering.
• Clark is a principal of Clark, Rockoff and Associates.
• Uri Cummings is co-founder and vice president of product development of Fulcrum Microsystems, which has developed the industry’s first high-performance clockless crossbar switch.
• Al Davis, professor and associate director of the computer science department at the University of Utah, has an interest in advanced computer architectures.
• Steve Furber is the ICL Professor of Computer Engineering in the Department of Computer Science at the University of Manchester (England). His research focuses on asynchronous logic design.
• Steve Nowick is associate professor of computer science and electrical engineering at Columbia University. One of his research interests is computer-aided design of low-power and high-performance asynchronous digital circuits.
The symposium is open to the public. Those interested in trends in microelectronic systems are encouraged to attend.
To register, call 935-6132 or go online to cse.seas.wustl.edu/clockless/registration.asp.