I first met Rick Munden more than a decade ago when we were both managers attending an Electronic Design Process workshop. I ran into him last December at an SDForum Emerging Technologies SIG meeting and we renewed our acquaintance. I invited him to our Bootstrapper’s Breakfast since he was mulling his new company Epiphyte. This interview grew out of several conversations that we’ve had in the last year. They have been condensed, spell checked, and hyperlinked for your reading pleasure.
Q: You’ve been entrepreneurial since high school. Could you talk about your first company?
My first legal business was a newsstand in Chicago’s Hyde Park neighborhood in 1965 when I was 15. I started it with a friend, Bob Katzman. It was a third kid’s idea but he was not inclined to follow through on it.
I sold my half of the business to Bob after about 16 months (and I was old enough to get a technical job). He grew the newsstand into a chain of bookstores over the following 20 years. Bob has written books about the newsstand and Chicago in that era and blogs at Different Slants.
Q: What were some of the key things you’ve learned from that?
The two things I took away from the experience were a respect for my customers–they are the most important part of any business– and the realization that retailing is not intellectually stimulating for me.
Q: You’ve also been involved in semiconductors, system design, and CAD/CAE for a number of years. What were some of the more interesting problems you had to solve?
I managed design engineering environments from 1987 till 2006, first at TRW in Redondo Beach, CA, then at Acuson/Siemens Ultrasound in Mountain View. During that time, although I had to support everything from chips to systems, including mechanical and software. I was personally more focused on board and system level design and verification.
I found the heart of any CAE system to be the libraries. In a company designing anything but the simplest boards, the libraries must be architected and optimized for efficient data transfer across a variety of tools, often from different vendors. The libraries I designed contained schematic symbols, PCB footprints, electrical information, purchasing information, signal integrity models, functional simulation models, timing information, and traceability information.
Q: What tools or methodologies did you develop that you still use?
The most important thing developed was the simulation modeling methodology. Fortunately, I had some very smart people working with me and we were able to come up with a modeling practice that has needed only a couple of tweaks over the past 12 years. We came up with a coding style based on VHDL/VITAL that allowed us to model a wide range of digital components that we could find no other way to accurately model. VHDL/VITAL was not the first thing we tried but, looking back, I think it was a fortuitous choice.
Q: You also started the Free Model Foundry, can you talk about what led you to do that?
When I was a manager at TRW, one of the engineering problems we had was how to simulate a board in order to reduce or eliminate the number of prototype board spins. Board spins were expensive and consumed way too much schedule. The biggest obstacle to simulation was the lack of models of the parts we wanted to use.
This was in the early ’90s so every tool vendor had their own proprietary simulator and models created for one would not work on any other. I had been writing models for several years but every time we switched EDA vendors I had to start over again.
Then VHDL came out. At first there were compatibility problems and none of the big companies could make simulators that implemented the full language. Eventually, a number of startups succeeded and were soon bought by the major players. In response, Cadence opened up Verilog.
Cadence had Verilog-XL and another product called Veritime that was a static timing verifier that read Verilog models. We thought “wouldn’t it be great if we could write one model that could be used for both dynamic simulation and static timing verification?”
We tried writing some models of small ECL parts in Verilog but could not model all the functionality. We hired some professionals to do the job but they also failed. Then we tried to netlist one of our Cadence schematics to Verilog and found out how difficult that was. We managed to get one design through the process but it was a very bad experience.
About that time, the VHDL/VITAL standard was being tested. One of my colleagues, Russ Vreeland, investigated and suggested we try it. The results were great. We could model our ECL parts easily and Cadence’s VHDL netlister was much better than their Verilog netlister. The next step was to populate our library.
There are a lot of digital parts in the world and people keep designing new ones. TRW did not want to be in the modeling business and at that time, neither did the IC companies. We thought if we documented a successful modeling strategy and published the models we created for our own use, other engineers would join in. Sharing models would be much more efficient than everyone re-writing the same ones. I have been a long time fan of the Free Software Foundation so I suggested we do something along those same lines for simulation models.
In 1995 two other TRW engineers and I incorporated the Free Model Foundation. Because we were trying to solve a problem rather than create a business, we incorporated as a not-for-profit. It took a couple of years to get our tax status set by the IRS and the State of California. In the process, our name was changed to Free Model Foundry.
For a couple of years, we wrote models at TRW and published them. But rather than the ground swell of models we expected to receive from other engineers, we started getting calls from IC companies asking if they could outsource their modeling to us.
It took a while to find the best resources for contract modeling but eventually we did and now model outsourcing has become FMF’s business.
Q: What have you learned about outsourcing? Any guidelines for what kinds of project should be outsourced and what shouldn’t?
I have seen many outsourcing projects go well and a few turn into complete disasters. Differences have been in project scope and the definition of the project deliverables. In general, small, well defined projects are more likely to be successfully outsourced than large poorly defined ones. Communications also plays a roll. The bigger the project, the more important good communications become and the more often it must take place.
I recommend a book titled “Global Software Development” by Dale Walter Karolak and published by the IEEE Computer Society. It covers all the basics in 158 pages.
Q: You are involved in some EDA Open Source efforts. Can you talk about any that you find exciting?
Other than FMF, my involvement with other Open Source EDA efforts is limited to cheerleader and occasionally facilitator. I host a monthly dinner which is attended by people interested in OSEDA.
Q: How would you compare the impact of Open Source vs. Outsourcing on Electronic Design and EDA?
EDA users are a small community. This makes open source less viable for EDA tools than in other areas such operating systems. There are only a few large open source EDA projects going on. I think all of them consist of a single person doing more than 90% of the work and a number of less committed people giving feedback. Smaller projects, such as a Verilog mode for Emacs, work fine.
Projects that are easily outsourced are often also viable as open source projects if they benefit a large enough community. The two examples that come to mind are FMF and OpenCores. These are organized quite differently but they have similar benefits to the engineering community.
Q: For your latest company, Epiphyte, can you talk a little bit about your plans for 2008?
The new company is Epiphyte LLC. It is a platform for exploring various business opportunities. The expectation is that we will try many different things and fail (cheaply) at most of them. The stated purpose of Epiphyte LLC. is for the “rapid exploitation of emerging opportunities”. This roughly translates to “we don’t know what we’re going to do but, we have a lot of ideas”. Among the more likely opportunities are:
- Provide IT support for startups, small businesses and non-profits. We serve organizations that require less than one FTE.
- Provide outsourcing project management for small HW/SW projects. We advise clients on the suitability of the project, help finalize the specifications, find and contract with the performing engineers or organization, manage the communications between the customer and the performer. The trick is to know what can and cannot be successfully outsourced, how to specify the work, and manage the customer expectation. Of course, it also helps to know competent organizations that can do the work.
- Provide contractor management services to companies that desire to keep existing contractors beyond the one year limit HR departments set.
Q: Epiphyte is also supporting “Venture Coding.” What is this and why you are offering it?
We have created a new process to assist start up companies in getting off of the ground that we call “Venture Coding.” Early start up companies often face the dual problems of limited starting funds and the limited engagement (and interest) of short term developers. Venture Coding was conceived to solve both of these problems.
In exchange for equity in a start up, Epiphyte will provide software development resources. This allows a company to preserve precious starting capital and to ensure the continued availability of developer commitment to the success of the start up.s