Interview with Ray Salemi

Featured Engineer

Ray Salemi

Ray Salemi - Verification Consultant

How did you get into electronics/ engineering and when did you start?

I got my first job in engineering right out of college when I joined the CPU Group at Prime Computer in 1985. This was a big deal for me because Tracy Kidder had just written THE SOUL OF THE NEW MACHINE, a book that told the story of a CPU development team. My first job was to write microcode for a machine that was intended to compete with the MicroVax.

I designed my first hardware using a precursor to today’s FPGAs—the Altera EP900 EPLD. This was a programmable logic device that could be erased by putting it under UV light. The 900, in EP900 stood for 900 equivalent gates!

I remember that I simulated even that small design and I found a bug in the design software that caused my state machine to head into the weeds. That was the positive reinforcement I needed to simulate every design from then on.

You have a Bachelor of Science in Computers Systems Engineering with a minor in English. That’s an odd combination. How did that happen?

Truth be told, I got a higher score in my English SATs than my Math SATs, so I had chosen a degree that didn’t really match my aptitude. I started taking English courses at UMass that taught me to analyze science fiction novels. I kept focusing on English as a way to balance my heavy math, science, and programming load.

How did you become an engineer?

When I was in the seventh grade in the 1970’s I took an aptitude test that said I could become a computer engineer. I didn’t know exactly what a computer engineer was, but I did go to the Museum of Science in Boston once a month and I loved playing with the computer exhibit. So I looked at the results of the test and decided that I would get a job “In Computers.”

What are your favorite hardware tools that you use?

I’ve become what I call a “Workstation Specialist.” This is an engineer who works entirely with a workstation and an editor. So I don’t really have a favorite hardware tool. I do have a favorite editor—EMACS.

What are your favorite software tools that you use?

I just love simulators. I think that my love of simulators goes beyond digital simulation. At Prime Computer I started a Strat-o-Matic baseball league that simulated baseball games. I also played war strategy games and, of course, role-playing games. (Yes, that’s right!) All these games had the common theme that they simulated real life.

So it made sense that my first job in EDA would be running customer support for the company that invented the Verilog language and simulator.

What is the hardest/trickiest bug you have ever fixed?

Recently I had my first experience with deadlock in a software program. This was a stumper. The simulation would just hang for no reason. I was distracted by looking for bugs in the simulator, the Makefiles, and the Verification IP. But the problem was in my own code. I had written code that caused two threads to wait for each other to give up different resources. It was classic deadlock.

What is on your bookshelf?

I write detective fiction and have two shelves of books on writing. (I recommend WRITING AND SELLING YOUR MYSTERY NOVEL by Hallie Ephron for anyone who would like to start.)

I got my MBA in 2002 and I wrote the management book LEADING AFTER A LAYOFF, so I have another two shelves of business books. (My favorite is THE INNOVATOR’S DILEMMA by Clayton Christensen).

Then I have another shelf of science, physics, and general nonfiction. I recommend anything written by Michael Lewis, the author of MONEYBALL and THE BLIND SIDE.

What inspired you to write your book FPGA Simulation?

It was clear to me that all the books on verification were focused on delivering cutting edge verification techniques to ASIC Verification engineers. I knew many FPGA designers who didn’t need that kind of power, yet nobody was writing a book to address their needs. I knew that FPGA designers who needed to simulate would be a growing market, and I figured that I could write the book in six months. It actually took two years. I had to rewrite the book twice to make it as clear and straightforward as possible.

Do you have any tricks up your sleeve?

I have a simple rule not to write more code than I can debug. So I write all my code incrementally. I write some, simulate it, write more and simulate that. I can always break down complex problems into a series of simple programs this way.

What has been your favorite project?

I designed a video game once that could run on a VT100 terminal using an 8085. That was the most fun.

Do you have any note-worthy engineering experiences?

Sadly, the most note-worth thing I’ve done in the lab is to put an old DIP package into a board upside down, and then wonder why the numbers on my circuit were changing without a clock. I got my answer when the package blew up, hitting the ceiling with a piece of plastic.

What are you currently working on?

My current project is a simulation test bench that can be controlled by the software running on a simulated SOC. It’s very cool.

What direction do you see your business heading in the next few years?

I see that FPGA Engineers are going to need to simulate their designs more completely in the future. However, they won’t have the time to devote weeks or months to test bench development. I see that the industry will need to supply more off-the-shelf components for simulation.

What challenges do you foresee in our industry?

I have an unpublished mystery that told the story of a murder at the Design Automation Conference. The book described a booth at DAC that had no sales people or demos. It just had a parrot that said, “Chips are getting bigger. Chips are getting faster.” The parrot had been coming to DAC for 15 years.

And so it goes. Chips are getting bigger and chips are getting faster. I don’t see that slowing down anytime soon, and I don’t see our schedules getting longer to accomodate more complex designs. So I see that our industry will create larger and larger pieces of IP to connect, and it will need more complete verification tools to make sure that those complex chips work.