Invent A New Way To Do Your Job

My friends own a farm in the southwest of France, and though I spent most of my recent decades around big cities, my village-raised roots are sending me working in the fields every time I have the time. I don’t really care what I’m assigned to, as long as soil, the nearby forest trees, and the sky (preferably blue) will take part. If the job consists of repeating actions, I like to come up with an efficient way of doing it.

It’s not because I’d like to finish it fast, I just like doing my job in a new way every time. Finding a new way means you have an old one. You start a task for the first time and you execute it as you were shown or in the way that looks natural. Then, after a while, you realize you can change the order of things, share the tasks within the team differently, and the machine is shifting gears. Say you are building a new fence around a newly planted vegetable field, and suddenly a two-day job ends in a few hours. This is a great feeling.

The power of methodologies

Some 150 km from my friend’s farm resides the famous Petrus winery. Now, go and try telling these guys about inventing new ways of doing things. They make one of the best wines in the world. For hundreds of years, they have done exactly the opposite. They are able to maintain their methodologies and repeat an excellent process year after year.

Debugging methodology

Working as a verification engineer, manager, tools developers, I’ve been observing engineers doing work they are familiar with for over 20 years. It always amazes me how little innovation our verification job demands from us. I could apply today’s verification techniques that I learned some 20 years back at National Semiconductor and I’d still be okay. I’d imagine that in software, as an example, my knowledge from back then would be mostly irrelevant. The fundamentals of good engineering remain the same, but there is always a better way to do something. All you have to do is discover it.

When it comes to debugging, this story takes a turn. Debugging is solving a problem. It’s a detective challenge. You have the facts, you have the clues, and everything you think you know — your thought process, good and bad conclusions, the whole package.

Nobody teaches you how to debug, there is no proven way or methodology. Some people do it better than others, and surely experience comes into play because you develop some kind of a hunch.

But what if there could be a way to set the fundamentals in place, to provide a proven, applicable scheme that engineers can simply follow and get at least a decent productivity doing so? Later on, new ways will be invented that diverge from the basic one, providing the evolution that is a must if we are to continue to innovate in the semiconductor industry.

Our new Cogita machine learning-supported features are aiming exactly at that point. We’ve designed basic recipes for various debugging problems, from which each engineer can develop new ways in dealing with their specific challenges. Stay tuned, more will be coming soon.