Issue link: https://iconnect007.uberflip.com/i/1295812
OCTOBER 2020 I DESIGN007 MAGAZINE 27 going to be interacting with the components. On the other side, you have placement and routing algorithms that have nothing to do with physics, but rely more on math principles such as statistics and graph theory. How do you opti- mize the placement? All of those things have to work together, which is the challenge. In the end, you have your package that you place on an EDA tool. You put it on a tool like the major vendors. They get that. As of now, they have some offerings, but we're working on the roadmap. We're trying to figure out what it is that's going to be needed in the next 10–20 years. What they're doing now is more to satisfy the current needs of their customers. Johnson: That's one of the places where the roadmap makes sense. That may not be the best way to frame the question. But one of the reasons for the roadmap is to give a long-term vision to part of the industry, such as the CAD tool sector, to help them understand what it is they need to put in with staff for development. What has the response been so far? Schutt-Ainé: My experience is limited as far as my interactions with the EDA people, but they are very customer-driven. If you come up with a new idea, such as, "Here is this thing that you could use to design a quantum computer," they probably would not be excited by some- thing like that because they don't have any customers right now that have that demand. They are always more excited by what custom- ers are asking for. I don't know how you fix that, but most of what they put out is based on customer demand. For instance, machine learning is a very exciting discipline right now in academia, especially for design. But it's a difficult sell to the EDA vendors because they're not get- ting that request from the consumers, from the customers. They tend to want more traditional things. It has always been my experience. Usu- ally, that comes from the demand that they get from the customers. Johnson: The readership that should create the demand for your audience is probably reading Chapters 5, 6, and 7 around the application. This is where things start to dovetail. How do you coordinate those messages in the roadmap with your messages? Schutt-Ainé: What would trigger the EDA ven- dors to get more on board is if you had the actual manufacturers and designers of these futuristic systems putting things together, say- ing, "We need a tool to do that." Let's say you want to do an optical SerDes working 56 gigabits per second or even higher. They're talk- ing about 128 giga- hertz per second. I don't think that you have a commercial tool that can do that, especially if it combines optical with electrical components. But if you have an active program or designer looking into doing it and there's a budget for it, that would get the EDA vendors on board. Shaughnessy: One thing you pointed out was that everything, including simple things like the data formats and types for packages and boards, is completely different than those used at the IC level. Schutt-Ainé: PCB designers also have to learn about electromagnetics. The clock rates are getting faster, and the end result is that the signal that goes to your PCB is more like a microwave signal. Component dimensions are comparable to the wavelength, which means that the wire is not just a simple wire; it's a transmission line. The physics are completely different because it behaves like a wave. In the mid-'90s, I traveled a lot, giving basic elec- tromagnetic courses to PCB designers. They didn't worry about that in the '70 or '80s because the clock rates then were relatively slow. Things started changing in the late 1980s. At the same time, they were working on things where the clock rates were approach- José Schutt-Ainé