There aren't many currently used EDA tools that are celebrating 40th anniversaries this year, but the SPICE circuit simulator, first released by the University of California in 1971, is. SPICE creators and innovators spun compelling and amusing tales of the development of the venerable circuit simulator at a 40th Anniversary panel at the Computer History Museum in Mt. View, California, Feb. 23.
The event was co-sponsored by Cadence, Mentor Graphics, Synopsys, and the IEEE Solid State Circuits Society. The following SPICE veterans participated in a panel discussion.
Ron Rohrer is currently professor emeritus of electrical and computer engineering at Carnegie Mellon University. While on leave from Fairchild Semiconductor, he taught the graduate course in 1968 that resulted in the creation of SPICE.
Larry Nagel is currently an independent consultant. He was a student in Rohrer's class and was the "ad hoc team leader" and primary author of the first SPICE program, which formed the basis of his PhD dissertation.
Kim Hailey, now an entrepreneur, was co-founder of Meta-Software, one of the first providers of a commercial version of SPICE. Originally a microprocessor designer, he was a developer of HSPICE.
Ken Kundert, now a consultant at Designer's Guide, developed the Cadence Spectre simulator and derivations including SpectreHDL and SpectreRF. He has contributed to both the Verilog-A and Verilog-AMS languages.
David Hodges, professor emeritus at U.C. Berkeley, moderated the panel. From 1970 to 1998 he was involved in teaching and research on IC technology and design, including mixed-signal circuits.
Here are some interesting facts you may not have known about the history of SPICE.
Changing the agenda. When Rohrer returned to U.C. Berkeley in 1968 to teach a circuit synthesis class, he didn't have time to prepare to teach it, and didn't have much faith in circuit synthesis. So instead of teaching that topic, he announced on the first day of class that the students would write a circuit simulator similar to the FairCirc program he used at Fairchild. Success would mean "As." Failure would mean oral exams.
Don't try this again. Following the success of SPICE, Rohrer tried similar kinds of classes, and they all failed. The Berkeley class succeeded, he said, because of Larry Nagel.
Finding a better name. What we call SPICE (Simulation Program with integrated Circuit Emphasis) was originally CANCER (Computer Analysis of Nonlinear Circuits, Excluding Radiation).
10-transistor simulators. There were other circuit simulators at the time, Nagel noted, but they were very limited in capacity and generally provided only one kind of analysis, such as dc bias points or ac poles and zeroes. "Most early simulators were good to 10 transistors. That was the end of your computing budget or the end of your patience."
Early days of computing. The compute environment used to develop SPICE was a CDC 6400 with 200K octal words of memory (one word = 60 bits), half of which was taken up by the operating system. Students ran punched cards at night, hoping that a single mistake wouldn't prevent compilation. This $6 million computer had a 10 MHz clock.
Open source pioneer. Open source programs existed before SPICE, but they didn't have much commercial value, Hodges noted. SPICE did, and open-source Berkeley SPICE later co-existed with for-profit commercial versions, showing that this could be done.
Restricted distribution. Despite the fact that Berkeley was providing SPICE to anyone for $20 per copy, it was illegal during the Cold War to ship SPICE to Communist countries.
Ease of use matters. In the 1960s IBM had a public-domain circuit simulator called ECAP. It was so hard to use that Nagel, the author of SPICE, couldn't figure out how to run it. "I crashed and burned," he said. How many people have heard of ECAP today?
A "sparse" breakthrough. What gave SPICE the edge over other simulators at the time? The sparse matrix technology, Nagel said, is what made it possible "to actually start simulating real circuits." Rohrer asked one of Nagel's classmates, a mechanical engineering student, to investigate sparse matrices after the student complained that he knew nothing about circuits and that Rohrer wasn't teaching about them.
The payoff. Some may have wondered why Berkeley was giving SPICE away, but there was a payoff. SPICE helped DEC sell a lot of VAX computers, and DEC contributed $18 million to help the university build a fifth floor on Cory Hall. Berkeley probably wouldn't have gotten $18 million in licensing revenues, Hodges noted.
Commercial success. One of the first commercial SPICE simulators was HSPICE from Meta-Software. By providing support on multiple platforms, offering support, and fixing problems in the source code, Meta-Software established a successful business and sold 11,000 copies before the company was acquired by Avanti, Hailey said. In-house modeling was also key to their success.
New approaches take time. When Kundert developed Spectre at Cadence, he was looking for some unique feature that would set it aside from HSPICE. He came up with SpectreHDL, which pioneered analog behavioral modeling and was a precursor to Verilog-A. But getting people to learn behavioral modeling was not easy, and it took a long time for the methodology to become established. (Today Cadence provides behavioral modeling in Virtuoso-AMS).
Back to the future. Will SPICE be with us forever, Hodges asked? Kundert replied: "If you look at SPICE today it is basically the same as it was back then. People have been trying to change it for many years, but I think SPICE will stay the way it is until something major changes - and I don't see what that would be."
A Plug for the Museum
If you haven't been to the Computer History Museum, site of the SPICE 40th anniversary panel, go see it. From the abacus to Charles Babbage's Difference Engine (right) to the Internet, it has what is claimed to be the largest collection of computing artifacts in the world, including hardware, software, documentation, photographs, and videos. You'll find punched card readers, minicomputers, mainframes, early PCs, and probably any type of computing platform you remember using.
If you're not in the area, the museum has on-line exhibits as well. Further information is available at the museum web site.