Arcadia Focuses On Datapath Designs

Electronic News, July 6, 1998 by Ann Steffora

There are a few options in terms of dealing with these problems including a full-custom design flow, which does not use synthesis on the majority of the design, but rather relies on a hand-crafted methodology to optimize the datapath and memories. This methodology has problems, however, in managing the complexity of the design, as well as in meeting the market window for a particular application.

Another approach is based on ASIC methodologies developed over the last 10 years, which is based on synthesis to convert RTL level designs into gates that are placed and routed into standard cells. This method works for logic-intensive designs, but as the amount of datapath and memory per design increases, the effectiveness goes down, forcing many design teams to internally develop tools to maintain the structure of the design and implement the datapath elements as datapath.

This is where Arcadia Design Systems comes into view. The company recognizes that as complexity and performance increases, so does the conservation of power, and new techniques will need to be developed to give system designers more control over the design's performance as well as over the entire design process.

Arcadia is headed up by Wei-Kong Chia, Ph.D., president and CEO, who took some time recently with Electronic News to explain his background and Arcadia's company strategy for its datapath placement tool, Mustang.

Dr. Chia: "I came to the United States in 1979, and I pursued my advanced study at the University of Illinois, Urbana-Champaign. I graduated in '84 with a Ph.D. in electrical engineering. I specialized in computer-aided design field.

"I moved to the Silicon Valley in '84. The first company I (worked at) was Intel, so I worked at Santa Clara, in the corporate CAD group.

"I stayed at Intel for two years and, during (those) two years, I had a very good learning experience, which is still now the number-one chip maker in the world. Besides the technology, I also learned about company culture and infrastructure. I really appreciated that I got the opportunity to pick up something other than technology -- as people know Andy Grove is not only famous for technology, but also for managing.

"As a new college graduate, to work at Intel was a really good experience that enabled me with many fundamental things which, later on in my career, I always recall what I learned at Intel, which is a well-managed company, well-managed system. I still miss Intel a lot!

"Then I moved to LSI Logic, which, by that time, was the number one ASIC company, and I worked at LSI Logic, for a couple years, and moved to Data General.

"I stayed at Data General three years until they shut down the branch office in Sunnyvale. What I did at Data General was the last generation of the 32-bit CPU. We also watched a new era -- moving from a traditional or proprietary CPU company into a more open and standard CPU provider.

"At Data General, we used Hitachi as our foundry. After Data General closed the Sunnyvale Division, I was hired (by) Hitachi, where I stayed five years, until I started (this) company, Arcadia Design Systems.

"When I was working at Hitachi, I was responsible for the overall CAE development and application support, which allowed me to utilize my background in different fields, because I was responsible not only for Hitachi's internal CPU controller memory, but also Hitachi's worldwide ASIC support. So we had to develop ASIC kits and other tools.

"In building my career in these four companies, I got plenty of opportunities to work with in-house developed tools and also, as a user, I had good opportunity to buy the tools from all different EDA vendors.

"As we know, EDA technology has grown out of three major resources. One is universities, another source is the big corporations like IBM, AT&T, Bell Labs, HP. And the third is from small start-up companies. Over the last 10 years, the change in the economics for small mergers and acquisitions, and also because of some other economic considerations, big companies are trimming down in-house development, and big EDA companies feel the opportunities to really grow their technology. Because they cannot afford the lengthy development time, acquiring technology through small start-up companies has become a popular thing.

"So I was lucky, in those 11 years, to have the opportunity to really work from the semiconductor side. And compared to most of my colleagues in college, most of them joined EDA companies, and worked in a really specific, focused field. And I was quite different from most of my EDA colleagues, because I put myself in such a situation that I could expose myself to all different kinds of tools, and talk to users, to get first-hand experience.

"So during those years, I kept thinking, 'What bothers users? What's the most important thing in order to improve the productivity of IC design?'

"I looked at a lot of IC design and many things have seen tremendous changes. Back in the early '80s, or even back to the early '70s, so-called 'IC designs' most of time were hand-cranked. People only needed a pencil and a calculator to calculate the performance of the IC, so the tool was more focused on the simulation, and later on, some schematic editors and some layout editors.