ICScape: At the Junction of Math & CS, EDA & IP
November 2nd, 2017 by Peggy Aycinena
ICScape is both an EDA company and an analog IP provider. The company’s EDA tools “cover the complete AMS design flow”, its SoC design tools are “point solutions for design closure”, and the ICScape IP catalog includes a variety of analog IPs.
Jason Xing is CEO of ICScape. Prior to co-founding the company in 2005, he was at Sun Microsystems for 7 years. Xing has two PhDs, in Computer Science from the University of Illinois at Urbana-Champaign and in Mathematics from the University of Louisiana. Xing is an accomplished technologist, and a well-informed observer of the semiconductor industry.
We spoke last week about his company and the future of the technology wherein he positions his offerings.
WWJD: How are things going with ICScape?
Jason Xing: Things are going very well for us.
WWJD: Has the company made any recent announcements?
Jason Xing: No formal announcements of late, but we had a merger with a company called Legend Designs Systems in April of this year. The products of the two companies are related to each other, and we shared several mutual customers.
The Legend products are more on the memory characterization side, along with standard cell timing models, where as ICScape offer products for standard cell library qualification. So in that sense, we were not competitors but complementary technologies.
Also, Legend has a special modeling technology that can be used for special designs like flat panel displays – not ICs – so their markets are diverse and interesting.
ICScape is based in Silicon Valley. Our first office was in Cupertino, but then moved several times and are now in San Jose. Legend is also based in the US, but with additional branches in Taiwan.
WWJD: What is the origin of the name ICScape?
Jason Xing: That is an interesting question.
At the time we started the company, this was our thinking: This is an EDA company and we provide tools for design. However, we felt that people would also have a broader view of the IC industry, a more general outlook – something like a landscape or overview, which inspired us to choose the name, ICScape.
Also, you may remember the company Netscape. That also was a factor in our thinking.
WWJD: Who are your customers today?
Jason Xing: We have several top-tier customers including Marvel and HiSilicon, and in Korea, the big guys – Hynex, LG, and Samsung.
WWJD: Do you have to travel a lot to visit customers?
Jason Xing: Actually, I used to travel a lot more, but happily now am traveling less.
WWJD: What has changed to allow you to cut back on your travel?
Jason Xing: The nature of our business has changed.
We now have different people in the different regions who are able to handle things for us there. I can now focus more on product directions, as well as the business direction for the company.
WWJD: Your personal background is quite unusual, with two PhDs. Is there a story behind so much academics?
Jason Xing: Yes. Actually you could say that diploma-wise, I studied in two different fields – computer science and mathematics. From my point of view, however, they are one field.
Math is more computational and related to numerical analysis. When I continued my studies in computer science, my thesis was on placement-and-routing algorithms, design and parallelization. There is a very large computational aspect to these topics, so they are very closely related to mathematics.
WWJD: At this point, ICScape has been in business for over 10 years. Can you look forward to the next 10 years and predict where the company will be at that point?
Jason Xing: Actually, we know we have lots to do going forward. The EDA industry is growing only as long as the semiconductor industry is growing, so our technology must keep advancing.
New products and solutions are always needed by the designers, so we must follow the wave of change in the semiconductor industry by continuing to grow our company and offering even better tools.
WWJD: Does that translate into specific goals over the next 3 to 4 years?
Jason Xing: Yes, it includes developing some selected products.
For example, in the area of EDA solutions: Lately we have found that SPICE simulations – based on technology that is 30-to-40 years old – are not meeting the needs of the design houses who want to develop products using advanced deep-submicron processes.
This means there are new challenges for the SPICE simulator itself, so we have innovated our simulator to provide post-layout simulation for advanced process designs. We have been successful here, and have solutions that have reduced design times from weeks to just a few days, or even to just overnight.
Another area we are addressing: As the designs become huge, just putting everything together for tapeout has become a very painful process. So we have developed technology that will process the layout very fast.
Just as word editing really increased office productivity, our technology is increasing layout productivity, making that operation very, very fast.
WWJD: If Moore’s Law is slowing or dead, where is the motivation for developing the kind of productivity boosts your tools provide?
Jason Xing: As you can easily see in EETimes, and other publications, this or that group continues to say that Moore’s Law is dead. I think it is still too early to say it is dead.
However, the designers really need new tool technology to keep up with the demands of the bigger, more power designs. They need the tool providers to keep Moore’s Law alive.
Manufacturing, of course, also needs to keep up and EUV may be the key to that success.
WWJD: Do we really need the newer nodes, if the emerging IoT market is doing just fine at the older nodes?
Jason Xing: It really depends on the solution you’re looking for.
Some things can be solved by parallel work at the older nodes. But you really have to put a lot of things together successfully there to generate enough power for those devices to function properly.
Also, I believe there should be trade-offs between performance and how many devices you can use when deciding the viability of any particular node.
And no matter how large the IoT market becomes, eventually you will still need innovations in the architecture, new innovations to boost the power of computing.
High-end computing needs going forward will not be met without significant changes in the technology.
WWJD: In other words, a ramp up to non-Von Neumann computing?
Jason Xing: Yes, something that is non-traditional, something that offers at least a different architecture, and maybe even new computational methods.
Lately, we have something that can be modeled by AI and if we do that, it probably will solve a lot of the current problems in a completely different way.
WWJD: These very dynamic and changing times bring up a question: If you were to do your PhD today, where would you put your emphasis?
Jason Xing: That’s a very interesting question.
I would probably stay in the same line, but I would want to do just one PhD, and that would be in Computer Science. However, the first half of my studies in CS would focus on mathematically related issues like algorithms and architecture.
Then in the second half, I would pick a particular application field like AI or next-generation user interfaces.
[laughing] When I was in school in the 1990’s, nobody picked AI as a subject for their studies. Only after graduation did I see this huge increase in interest in AI, which has brought us to today’s focus.
WWJD: It seems ‘AI’ covers a host of different topics, and is really too vast to be described in one simple way.
Jason Xing: Yes, artificial intelligence goes way beyond computing. It includes the whole of humanity and a lot of things related to that discussion.
WWJD: Why do you belong to the ESD Alliance?
Jason Xing: The ESD Alliance provides us a platform where we can go and exchange ideas, and where we can see market directions, especially those focuses on software solutions and the IP side of design.
We also learn there how to interact with the industry, with customers, with the design houses and the foundries. Overall it is a very good platform for all of that for us.
WWJD: Could the ESD Alliance do more to assist you?
Jason Xing: So far, I feel they are already doing what we need to have done.
For companies in the ESD Alliance, we need major events and conferences to have a chance to really interact and talk with the customers. Things like DAC, or foundry events like TSMC. These really help, especially for smaller or medium-sized companies who need help showing what they have and interacting with customers.
Also, ESD Alliance hosting seminars is very helpful, getting people to come in and give case studies and tell about their success stories. These events are very informative.
WWJD: How will the semiconductor industry change in the future?
Jason Xing: Certainly semiconductor is the key enabler in all markets, and it is really booming.
And the overall economic outlook says that we will all need so many additional applications – more powerful chips for AI and automotive, for instance.
All of these applications will need very powerful chips to run on, and at the system level over the next 3-to-5 years, there will be even more developments.
I expect to see auto-piloted cars that you won’t need to drive. And in AI, expert systems built on computers or even mobile phones, so for any given outcome, you won’t be able to tell if it’s a human or a device driving the system.
That will definitely be some kind of dream come true for computer scientists.
WWJD: To some, it sounds more like a nightmare.
Jason Xing: It is only a nightmare because people are worried about losing their jobs, and perhaps computer intelligence overpowering human intelligence.
But we should not worry too much about those things right now. When the time comes, humans can invent ways to guarantee that we are always in control.
WWJD: So would you travel in an autonomous-driving car today?
Jason Xing: [laughing] Well, perhaps right now I wouldn’t do it. The safety guarantees are not there yet.
WWJD: How do you hire bright new employees?
Jason Xing: Through the years, because of our graduate studies, we have made very good contacts with the universities, and with the research groups in those universities.
In the US, for example, we have good contacts with the University of Illinois at Urbana-Champaign, and communicate often with both the head of Computer Science and the head of Engineering. Also, in China we have joint research projects with Tsinghua University.
These are the places where we typically find the best talent, although it may depend. For someone to be a successful AE, for instance, they don’t really require an advanced degree, although they sometimes do better if they have a masters degrees.
WWJD: A masters in EE or CS?
Jason Xing: It’s really a mix. For most of our R&D teams, our engineers are either from EE or CS, although Mathematics is now very popular as well.
If we are working on some really new algorithms to speed up simulation, for instance, the bottom line is that mathematical computation is much more relevant than electrical engineering.
And certainly if you plan to put machine learning into your EDA products, mathematics is even more important.
WWJD: Do you think professors should encourage their CS students to go into EDA?
Jason Xing: Most certainly, for the reason that throughout my education and my professional career, EDA has been the most challenging field. It basically requires knowledge of mathematics, computer science and electrical engineering.
To make things computationally efficient requires a background in math and CS, and to understand devices requires a background in EE.
This is a very challenging field. If you feel that you are smart, this is the field where you will really find a lot of room to grow.
The semiconductor industry is growing, and really needs people like those of us in EDA to take those challenges and stay in the field.
If you measure success by how much money you make, this may not be the top field, but it certainly is if you want the best challenges.
WWJD: If I want to make a lot of money, what field should I go into?
Jason Xing: [laughing] If you want to make a lot of money, go to Wall Street. Or start your own business and make it explode into the markets.
About ICScape …
“ICScape Inc. is an advanced EDA software and Analog IP provider. ICScape EDA tools covers complete custom AMS design flow: schematic entry, layout editing, circuit simulation, LVS/DRC, and RC extraction; ICScape SoC design tools are point solutions for design closure, including timing ECO, clock analysis/verification, IP/Library QA, layout review. ICScape IP catalog includes analog IPs such as PLL, ADC, DAC, PMU, and Serdes.”
Jason Xing is co-founder, President and CEO at ICScape. Previously he served as CTO and VP of Engineering of ICScape, where he architected and lead the development for ICScape timing closure, clock analysis, and leakage power reduction products.
Xing has 20+ years’ EDA research and development experience. In 1997, he joined Sun Labs after receiving his PhD in Computer Science from the University of Illinois at Urbana-Champaign. At Sun Labs, Xing did research on physical and logical concurrent design methodologies and shape-based routing technologies. In 2001, he joined the Sun Microsystems internal CAD development team before he co-founded ICScape in 2005. Xing also holds a PhD in Mathematics from University of Louisiana.
Tags: HiSilicon, Hynex, ICScape, Jason Xing, Legend Designs Systems, LG, Marvel, Samsung, Tsinghua University, University of Illinois at Urbana-Champaign, University of Louisiana