feature article
Subscribe Now

A Brief and Personal History of EDA, Part 3: Daisy, Valid, and Mentor Graphics – The CAE Era

By the end of the 1970s, the leading CAD companies, including Calma, Applicon, and Computervision had started to lose interest in the electronics market and turned to mechanical CAD. Quite possibly, this lack of interest reflected the demand by electronics and semiconductor companies for something more than efficient drafting systems. The drawings produced by CAD systems were fully capable of producing photomasks for circuit boards and ICs, but these systems understood polygons. They had no comprehension of the electronics these polygons represented. CAD systems might know the difference between a 14- and a 40-pin DIP, but they did not know what a NAND gate, a microprocessor, or a DRAM were.

Then, in 1979, Professor Carver Mead at the California Institute of Technology (Caltech) and Lynn Conway at Xerox PARC dropped a bombshell on the electronics industry. The bombshell was publishing a textbook titled Introduction to VLSI Systems, commonly known as “Mead and Conway.” The textbook presented a structured IC design methodology that was based on MOS IC design classes that Mead had taught at Caltech starting in 1970, coupled with many practical contributions by Conway.

In 1976, Xerox PARC invited Mead to teach a highly compressed, 3-day version of his course. Lynn Conway, who’d joined PARC in 1973, was in the audience. Before joining PARC, Conway developed computer architectures at IBM and Memorex. She latched onto Mead’s ideas, developed scalable design rules and a formal methodology for developing VLSI ICs, and created a multiproject chip (MPC) methodology for putting multiple designs on one wafer to reduce the cost of IC prototyping.

In early 1976, Xerox PARC and Caltech initiated a collaborative research project to explore easier ways to design systems in silicon. Caltech’s Mead and PARC’s Conway joined the project. Mead taught PARC about MOS design, and Conway taught Mead about computer architecture and design. Other corporate partners in the Caltech Silicon Structures Project included IBM, Intel, Digital Equipment Corp, and Hewlett-Packard. The most important result of this collaboration became known as the Mead-Conway methodology for VLSI design. Doug Fairbairn started working for Conway shortly after Mead’s 1976 presentation at PARC. In his oral history, Fairbairn describes what happened:

“So, during the ’76 to ’78 timeframe, I was mainly working with Lynn [Conway] and helping to refine this methodology and figure out what works and what doesn’t, designing chips, getting them fabricated, figuring out how you’re going to make masks, how you do multi project chips. Another guy, Alan Bell joined at the time. So, Lynn [Conway] started building up a real group around this whole activity. And then right in there, I don’t know exactly when they started, but both Lynn and Carver started collaborating on the book Introduction to VLSI Systems.

“…everybody asked, well, did Carver write it? Did Lynn write it, or whatever? Clearly, the core ideas all started with Carver. Lynn’s background was in computer architecture. But she could see that the– she could see the future of what VLSI promised in the same way that Ivan Sutherland saw in terms of how you could map certain computer architectures into a very high-density, high-performance VLSI circuit – if you thought of them in terms of VLSI architecture as opposed to normal gate-level architecture.”

“…Lynn collaborated with Carver on the book. Lynn physically wrote the book. I think every character in the book was typed by Lynn. But most of the high-level ideas came from Carver. So, Carver was still teaching at Caltech. He’d come up one or two days a week. They would sit and talk, typically in Lynn’s office. And Lynn would type away because she had an Alto [computer] and she was typing. And with the Alto, you could draw diagrams and pictures and all of this stuff. So, she created that book, physically created the book, and contributed, I’m sure, many of the detailed ideas along the way, ways of saying things, and that sort of thing. She’s very detailed oriented. And so, it was very much a collaboration. I mean there’s no way the book would have happened without either of them. Carver would have never written it. Lynn would never have had the idea. So, it’s a perfect example of collaboration.”

Word of the Mead-Conway design methodology spread through university courses and corporate classes in IC design. Conway taught a similar class at MIT based on this material in 1978, and her extensive instructor notes and videotapes were used by dozens of university professors to train an entire generation of IC design engineers. Classes based on this methodology and using manuscript copies of the textbook and Conway’s teaching notes and videotapes were taught by Ivan Sutherland at Caltech, Robert Sproull at Carnegie-Mellon University, and Fred Rosenberger at Washington University in St. Louis.

Suddenly many more engineers who wanted to design their own application-specific ICs now had a methodology to do so. Of course, they’d need better tools. Consequently, the Mead-Conway VLSI design methodology triggered a new wave of computer-aided engineering (CAE) tools and trained many of the engineers who developed those tools. The wave hit in the beginning of the 1980s with the creation of three companies that would dominate the decade: Daisy Systems, Mentor Graphics, and Valid Logic.

Daisy Systems

Two Intel engineers, Aryeh Finegold and David Stamm, founded Daisy Systems in 1980 or 1981. (Both years appear in various histories.) David Stamm had joined Intel in January 1974 as the sixth person in the company’s microprocessor group. Initially, he fixed bugs in the company’s 4004 and 4040 microprocessors. He then designed the Intel 4308, a support chip for the Intel 4040 that combined a 1-Kbyte ROM with some I/O expansion ports. He then joined Intel’s 8048 microcontroller project. In an interview with his alma mater, Purdue University, Stamm said:

“At Intel, one of my office mates decided to start his own company. He was having the time of his life. I saw his company progress from the idea stage to shipping a product, and I decided that that’s what I wanted to do. I spent nights and weekends with the other eventual founder of my first company, Daisy Systems, brainstorming through several different ideas.

“We asked a lot of questions: Is there a market for this product? Could we build it? Whom would we sell it to? We came back to one idea that was close to what we knew. Daisy Systems would build software that engineers at Intel could use to do a better job designing microprocessors. Their work would become less labor-intensive and more efficient. Daisy grew from an idea in mid-1980 to a $140 million company by the end of 1985.”

The office mate Stamm mentioned was Aryeh Finegold, who had served as a paratroop commander in the Israel Defense Forces before starting his engineering career. He moved to the US and started working for Intel in 1977, where he headed a team that was developing next-generation microprocessor architectures. While working at Intel, he noted a lack of automation for electronic design and decided he could do something about that. Finegold and Stamm recruited a few other engineers from Intel and started Daisy Systems. They also recruited Calma’s Vice President of Marketing for the company’s Electronics Division, Harvey Jones, and added Vinod Khosla as CFO.

Daisy quickly became a CAE leader, offering a variety of design tools for schematic capture, logic simulation, analog SPICE simulation, timing verification, parameter extraction, printed-circuit board design, and semiconductor chip layout. As was common in those days, Daisy designed its own workstation to run its design tools. Daisy’s Logician and Gatemaster workstations and their successors were based on Intel microprocessors with hardware graphics acceleration added.

 

Daisy Logician workstation circa 1983. Image credit: Steve Elliott
Daisy Logician workstation circa 1983. Image credit: Steve Elliott

 

Daisy’s CAE tools were a hit, until they weren’t. Daisy’s sales skyrocketed from zero to $122 million from 1980 to 1985, but the company’s fortunes reversed in 1986 as its proprietary workstation choice bogged the company down. Upgrading the CAE software and the workstation’s hardware design at the same time turned out to be a poor decision for multiple companies during the CAE era. Daisy’s stock share price dropped from $37 to $5 in mid-1986, and the board ousted Finegold. In 1989, Finegold defended his record by saying in an interview, “Anyone who invested $10,000 in Daisy received $3 million over the years. Daisy’s first two hundred employees are all millionaires today, and that includes the secretary and the janitor.”

In 1988, Daisy decided to try bolstering its sales by acquiring Cadnetix, another CAE company founded in the early 1980s that specialized in printed-circuit board design. Daisy retained Bear Stearns as its financial advisor for the acquisition. However, Cadnetix did not wish to be acquired under the proposed terms, so Daisy initiated a hostile takeover by acquiring Cadnetix stock. Bear Stearns and Daisy then amended the terms of the engagement to say that Bear Stearns would also assist Daisy in obtaining financing for the transaction. Bear Stearns issued two highly confident letters of credit for $50 million and $100 million to help finance the deal.

 

Cadnetix CAE and CAD workstations circa 1985. Image credit: Steve Leibson
Cadnetix CAE and CAD workstations circa 1985. Image credit: Steve Leibson

 

After learning that Bear Stearns intended to finance the acquisition, hostile or not, Cadnetix reversed its position and began to negotiate a friendly takeover or merger. Daisy and Cadnetix agreed on a two-phase merger where Daisy would first purchase 50.1% of the outstanding Cadnetix shares for cash. Then, six months later, Daisy would purchase the remaining shares using cash and convertible debentures. After completing the first phase, Daisy began to arrange financing for the second phase, and things didn’t go well. Daisy was forced to request a bridge loan from Bear Stearns. The request was denied, but Bear Stearns told Daisy that Heller Financial was willing to finance the second merger phase, at disadvantageous terms for Daisy. Ultimately, the entire reverse-acquisition/merger failed. Daisy’s financial condition deteriorated after the merger to the point where creditors forced the company into involuntary Chapter 11 bankruptcy in 1990.

(Note: I designed workstations based on Motorola’s 68000 microprocessor family at Cadnetix from 1982 to 1985. I left before this merger business started.)

Valid Logic

In 1980, Dr. Jared “Jerry” Anderson wanted to found a new technology company. He’d founded multiple startups, including Decision Inc, a supplier of disk and tape controllers for Data General minicomputers, and Two Pi Corp, which made minicomputers that executed the IBM 370 instruction set. The Two Pi V32 minicomputer was based on a microcoded processor built from 4-bit AMD 2901 processor slices. Ball Corp bought Decision, Inc, and N. V. Philips acquired Two Pi. Anderson was looking to continue his serial entrepreneurism, and his quest took him to Lawrence Livermore National Lab (LLNL), where two PhD students named Tom McWilliams and Curt Widdoes had spent the last five years working on the S-1 supercomputer project.

Anderson’s presence at LLNL wasn’t accidental. These were familiar haunts. Anderson had gotten his own PhD at LLNL back when it was named the University of California Radiation Laboratory, Livermore Branch. He’d worked on the team that developed the hydrogen bubble chamber, which was used to investigate particle interactions and resonant states. Professor Luis W Alverez directed that team and won the Nobel Prize for physics in 1968 for that work. When federal funding for work at LLNL started to dry up, Anderson left LLNL and started developing high-tech businesses.

Anderson had read the published papers about the SCALD design tools developed for the S-1 computer project, which was the brainchild of Dr. Lowell Wood, a physicist at LLNL and protege of Edward Teller. Wood envisioned a 5-phase project for the S-1, but only the first two phases were completed. Phase 1, completed in 1978, was called the Mark 1. It was a single-node, 10-MIPS computer built from 5300 ECL-10K ICs clocked at 10MHz. Widdoes and McWilliams designed the Mark I hardware. Mike Farmwald wrote the microcode. Jack Rubin joined the team and helped to debug the machine. The Mark IIA computer, a 15-MIPS machine based on ECL-100K ICs, would not be completed until after Valid Logic Systems was founded.

The most significant part of the S-1 project wasn’t hardware, however, it was the graphical, hierarchical design tools developed to create the hardware. Collectively, these tools were called SCALD – structured computer-aided logic design. The first generation of SCALD tools was based on the Stanford University Drawing System (SUDS), a graphical schematic-capture system. SUDS resembled the first-generation 2D CAD tools from Calma, Applicon, and Computervision. McWilliams wrote a macro expander, and Widdoes wrote a wire router and netlist program to round out the first-generation SCALD tool set. The second generation of SCALD tools added a packager, a timing verifier, and an automatic placer, written respectively by Widdoes, McWilliam, and Rubin.

These three people agreed to found Valid Logic Systems with Anderson, and the company was incorporated in January 1981. Initially, Widdoes joined Anderson to start the company. McWilliam and Rubin stayed at LLNL to complete the Mark IIA computer but concurrently served as consultants to Valid Logic. The SCALD software, which had been put into the public domain, served as the prototype for the tools developed at Valid Logic Systems. In fact, the company’s name was originally SCALD Corp, but the financial backers didn’t care for the name. The SCALD tools were university-grade. They’d not been productized and had been written in Pascal, so they were completely rewritten in C during development of Valid’s tools, which initially focused on gate-array and circuit-board design. Like Daisy, Valid elected to design and build its own workstation hardware. That was an obvious choice because the company was founded by a bunch of computer designers.

Initially, Valid Logic Systems was very successful. The company had a successful IPO in 1983, just two years after it was founded. The company’s success continued, and Valid acquired the electronics portion of Calma, one of the original big-three CAD companies, in January 1988. However, as the years rolled along, the use of proprietary workstations was overshadowed by the increasing performance of standard engineering workstations from companies like Apollo, Hewlett-Packard, and Sun. Anderson insisted that Valid stay with its proprietary workstation design, and the company started to fall behind. Cadence Design Systems acquired Valid in 1991.

Mentor Graphics

Unlike Daisy and Valid, which started in and around Silicon Valley, the Mentor Graphics story starts farther north, in Beaverton Oregon. Tom Bruggere joined Tektronix in early 1977 after spending a few years at Burroughs, one of the major mainframe computer manufacturers. He left Burroughs to escape the crush of people in southern California. Bruggere’s specialty was software. At Burroughs, he developed operating systems. At Tektronix, he managed software application development for the company’s desktop computers. At the time, Tektronix offered a desktop computer based on the company’s storage-tube display technology. The display technology had been adopted by almost all the CAD companies in the 1970s, but Tektronix was not a computer company, and its 4051 desktop computer with its 8-bit Motorola 6800 microprocessor and its large, slow DC300 integrated tape drive was not competitive with machines like the 9825 and 9845 from arch-rival Hewlett-Packard. (Trust me, I was working for HP’s Desktop Computer Division at the time.)

Tektronix started to develop a next-generation desktop computer in 1979. The company planned on developing its own unique microprocessor IC, writing an operating system from scratch for that proprietary processor, creating a graphics subsystem for its unique storage-tube display, and developing an application software suite for this completely proprietary system. Bruggere could see that the project was heading for trouble, and, when it imploded in 1980, he started thinking seriously about starting his own company.

Bruggere got together with some of his friends and co-workers and, after hours, they discussed the kind of company they might create. In an article titled “The Vision Trap,” Gerard H. Langeler, who was one of those co-workers, wrote:

“On the basis of our talents and backgrounds, we knew the product should have something to do with computer graphics, and it didn’t take us long to determine that the most promising application area was computer-aided engineering or CAE—the automation of schematic capture and simulation for engineers designing complex integrated circuits and printed circuit boards.”

Of course, that was precisely the same perspective that drove the creation of Daisy and Valid during that same period. Bruggere left Tektronix first and was joined a month later by Gerry Langeler, and Dave Moffenbeier, also from Tektronix. The three started Mentor Graphics in April 1981 and started hiring software engineers to develop the programs needed to create a CAE workstation. Unlike Daisy and Valid, however, Mentor did not develop a proprietary workstation. Because of his prior experiences with hardware development at Burroughs and Tektronix, Bruggere was reluctant to have Mentor relearn those lessons. Fortunately, one of Mentor’s initial backers had also invested in Apollo Computers, one of the earliest workstation companies, so Mentor initially adopted Apollo workstations and sold them through an OEM agreement with Apollo. The DMV companies in the CAE Era all earned significant income from hardware sales.

With the company founded and the workstation hardware selected, the core team started recruiting engineers, mostly from Tektronix, to develop the needed software. Their target completion date was the Design Automation Conference to be held in Las Vegas in June 1982. It was a last-minute rush to complete the IDEA 1000 CAE workstation, but it was ready for demos in time. The IDEA 1000 CAE workstation was a hit, and the Mentor team made several appointments for future sales calls. The company managed to ship about $1.6 million worth of its CAE systems by the end of 1982. From there, Mentor’s revenues climbed steadily, until 1990 when trouble arose.

Mentor had decided to rewrite its entire software suite to fit into a unified framework with a unified database. The new product was dubbed version 8.0 (eight dot oh). However, the company essentially tried to swallow a whale in tackling version 8.0. The product was delayed so much that “eight dot oh” became known widely among Mentor salespeople and customers as “late dot slow.” Worse, because Mentor’s customers knew the new version of the software was coming, sales of the current products fell in 1991 and continued to fall in 1992 and 1993, thanks to the Osborne effect. When version 8.0 finally arrived, its performance validated the “slow” part of its nickname. Mentor was now in serious existential danger. Langeler and Moffenbeier left the company, and a new CEO, Wally Rhines, was brought in. Mentor then underwent a painful metamorphosis to survive. It ceased to be a CAE company and became an EDA company, as I’ll discuss in the next article in this series.

References

Lynn Conway, “Reminiscences of the VLSI Revolution: How a series of failures triggered a paradigm shift in digital design*,” IEEE Solid State Circuits Magazine, Vol 4, No 4, Fall 2012

Oral History of Douglas Fairbairn, Computer History Museum, October 6, 2016

Oral History of Lynn Conway, Computer History Museum, February 24, 2014

Steve Leibson, “A History of Early Microcontrollers, Part 4: The Intel 8048 and 8748,” EEjournal, December 5, 2022

David Stamm Interview, Purdue University, 1999 Distinguished Engineering Alumni

Adi Mendelsohn, “Finegold Already Considering Next Project,” Globes, February 14, 1999

Daisy: A Warning for Investment Bankers, Findlaw, March 26, 2008

Bruce Upbin, “Higher Ground,” Forbes, October 27, 2003

New Products, IEEE Computer, June 1978

SCALD Oral History: #1 of 3 (Tom McWilliams and Curt Widdoes Together), Computer History Museum, February 12, 2008

SCALD Oral History: #2 of 3 (Curt Widdoes Alone), Computer History Museum, February 12, 2008

SCALD Oral History: #3 of 3 (Tom McWilliams Alone), Computer History Museum, February 12, 2008

Mentor Graphics Oral History Panel Session: David Moffenbeier, Gerry Langeler, and Tom Bruggere, Computer History Museum, March 22, 2013

Gerard H. Langeler, “The Vision Trap,” Harvard Business Review, March-April 1992

Dr. Walden C. Rhines, “From Wild West to Modern Life,” A SemiWiki Project, 2015

2 thoughts on “A Brief and Personal History of EDA, Part 3: Daisy, Valid, and Mentor Graphics – The CAE Era”

    1. Thank you for the compliment, metasequoia. I hope you find the series useful. There are five more articles to go in this EDA history, with a few things that surprised me, and I was there!

Leave a Reply

featured blogs
May 8, 2024
Learn how artificial intelligence of things (AIoT) applications at the edge rely on TSMC's N12e manufacturing processes and specialized semiconductor IP.The post How Synopsys IP and TSMC’s N12e Process are Driving AIoT appeared first on Chip Design....
May 2, 2024
I'm envisioning what one of these pieces would look like on the wall of my office. It would look awesome!...

featured video

Introducing Altera® Agilex 5 FPGAs and SoCs

Sponsored by Intel

Learn about the Altera Agilex 5 FPGA Family for tomorrow’s edge intelligent applications.

To learn more about Agilex 5 visit: Agilex™ 5 FPGA and SoC FPGA Product Overview

featured paper

Altera® FPGAs and SoCs with FPGA AI Suite and OpenVINO™ Toolkit Drive Embedded/Edge AI/Machine Learning Applications

Sponsored by Intel

Describes the emerging use cases of FPGA-based AI inference in edge and custom AI applications, and software and hardware solutions for edge FPGA AI.

Click here to read more

featured chalk talk

Designing for Functional Safety with Infineon Memory
Sponsored by Mouser Electronics and Infineon
In this episode of Chalk Talk, Amelia Dalton and Alex Bahm from Infineon investigate the benefits of Infineon’s SEMPER NOR Flash and how the reliability, long-term data retention, and functional safety compliance make this memory solution a great choice for a variety of mission critical applications. They also examine how SEMPER NOR Flash has been architected and designed for functional safety and how Infineon’s Solutions Hub can help you get started using SEMPER NOR Flash in your next design.
Apr 22, 2024
3,431 views