Dr CHI-FOON CHAN,
Dr Chi-Foon Chan, Chief Operating Officer of Synopses, is
assigned to look after the company’s internal operations and customer satisfaction.
Taiwan-born Dr Chan joined Synopses in 1990 after holding important positions in NEC and
Intel. An MS and PhD in computer engineering from Case Western Reserve University, Dr Chan
had maximum realization of his academics on Electronic Design Automation (EDA) industry.
An industry that enables automation of the semiconductor design. Let’s find out what Dr
Chan has to share with DATAQUEST on his experiences in the EDA market.
What sort of solutions do you
I think there are two challenges. One main challenge is productivity. As the complexity of
electronic systems gets bigger and bigger, you need equally more and more complex chips.
And yet the time cycle is not getting longer. For example, if you consider digital
cameras, they operate on a life-cycle of a few months. So productivity is actually the key
The second issue is of complexity. There is
no way to do it manually. We are packing more and more function into the chip when the
design is getting lesser than the tenth of a micron. The physics of electron tunneling
also gets into play. So you have to wrestle with physics while you worry about wireless
phones etc. So the design tools allow you to work at the level of complexity that ensures
that your design can get down to the details.
As an analogy, you can look at an architect
designing a 100-storey building without worrying about every brick that goes on the
DA is said to be an ‘enabling
technology’. Why is this so?
I was just telling you about the digital camera. The amount of technology that goes into
products like a digital camera or a wireless telephone is tremendous. The number of
variations are many. Without proper automation tools, if everybody had to recreate the
design again and again, you see, it’s like building a 100 storey skyscraper to re-draw all
the blue prints for every small change. You would never be able to come up with a final
blue print. Also, you will have to verify that this 100 storey building better stand
before you build it. Also, after you draw the blueprints, you have to be sure that if you
had some major or minor change you can react fast. That’s why EDA tools are the enabling
technology, because, without these tools, first you won’t even have the productivity, and
secondly you won’t be able to work on your design. Thirdly, you don’t have the flexibility
to react to customer needs. We live in an age where the first person to get to the market
has a major advantage; but that advantage is an advantage only if you can react to the
customer’s needs. If you get to the market first and the market says "this is great,
but this is what I want", then the second guy actually wins because you take so long
to adapt to the reality, and you have gained nothing. So as the consumer world gets more
competitive, the ability to respond rapidly to the customer becomes more important. And
that’s why I think that EDA is a major enabling technology.
Tell us about some breakthroughs in
the EDA industry?
I can give you a one-minute summary of what happened in the EDA industry during the last
30 years. In the seventies, engineers were designing in polygons because designing a chip
is really a photographic process. So engineers really had to draw or photo-build the
‘negative’. In the eighties, they used gates as the building blocks for designing chips.
In the nineties engineers use languages to design chips.
One of the new breakthroughs is the
presence of a whole new verification paradigm, which involves much higher performance than
our traditional simulation tools. We have recently announced formal verification tools. We
are talking about a different methodology of verification of many levels of chip design. I
think that design technology follows a path of evolution as opposed to a revolutionary
one. So you won’t see people changing radically tomorrow. Rather, tomorrow you will see
very creative tools coming out, and they will get slowly absorbed, and the methodology
will slowly change.
Can you tell us which direction EDA
is heading in?
To us, this is fairly clear. Where we are today is a short run from managing the paradox
of a system on a chip. When I say system on a chip, you are putting many building blocks
on a single chip. [The paradox is that on one side you have to go to a higher level of
design because there are many complexities and in the other because of the problem of the
complexities you cannot go to a higher level].
When there are many complex problems in
front of you, there are two ways to solve the problem. One is to go to a higher level of
abstraction-if you are going to build a 100 storey building, you can’t think of every
brick that you are going to use but every storey, or every room. In other words, you go to
a higher level of abstraction. The second thing is you do reuse. So you know which part of
the building you are going to reuse and you just reproduce it. That is one way to handle
complexity, a very common engineering way of doing it. The paradox is, we are doing a
system on a chip and we are putting millions of gates on a chip, so there is also physics
to contend with. Issues like electron tunnelling, reliability etc. are important here. So
you have to manage the details, and managing complexity and managing details are almost
opposite ends of a paradox. We want to ensure that from the high level you can safely
guarantee reliability. That’s one area we are working on.
Buggy chips are not uncommon these
days. What sort of testing procedures and practices are used in the indus-try today?
There are many new and innovative tools coming on board, like simulation tools and
verification tools. It’s really a matter of coverage, how many times you can go through
the design, and how many angles of testing you can cover? If the tools are faster, you can
do more testing. So that is one direction. At some point, the design is so complex that
appropriate coverage cannot be given, because one cannot spend months simulating the
design. It is here that formal testing methods are useful, where one isolates timing and
functionality. I believe that this will be a major area.
Your corporate history is rich in
the area of mergers and acquisitions. Is this typical of the EDA industry?
Mergers and acquisitions are typical of our industry basically for two reasons. First, in
our industry, to make any kind of sustainable progress, you have to be in one of the top
three positions. The number one company makes a good amount of money, the number two makes
some money, and number three breaks even. It’s not an industry where you can find 500
people making money at the same time. Basically, here, synergies are the issue.
With so much automation in design,
what is left for the human mind to do?
That’s a good question. You know, to answer that question you have to separate the human
beings into different segments. The bulk of human beings are actually in the high level
design. But you still need human beings at every stage of the process, it is only that you
need less of them. It’s almost like division of labor. For example, before you may have
needed 500 people to build a house; today, you may only need 20, but you still need one
guy to see the lumber, and do stuff like that. But the bulk of the engineering team will
think at a higher level, and at their level they are worrying about the architectural
constraints, verification, functionality, power consumption, speed etc.