Janakiraman (Jani) has been an engineer
since before the dawn of the electronic design automation era. When he started,
he was handcrafting designs and building systems with a then-whopping 4Kbytes
of DRAM. The explosion in tools, design methodologies, and productivity—not to
mention his front-row seat to the Indian electronics revolution—makes him an
excellent interview for our Cadence 25th Anniversary blog series. Today,
Jani is president and chief technology officer for Mindtree, the global IT, IP,
and design-services company (a longer biography is listed at the end of
this article). He talked to us about the evolution of electronics design,
the advent of global engineering teams and the enabling forces of EDA tools.
Q: Jani, you've seen some
amazing change in your career. What were some of the highlights?
A: The last 25 to 30 years have
seen phenomenal change in the semiconductor industry landscape. The fundamental
trigger happened with digital semiconductor circuitry but specifically the microprocessor
that revolutionized the whole electronics industry. I give credit to the
microprocessors from Intel and Motorola and the advances in microprocessor
integration since then, with wireless and other functionality being added.
When we started talking about
Moore's law in those days, I thought maybe it would be interesting for three to five
years and some things would sustain. But the surprising thing is Moore's law
has held for such a long time. The level of integration was not imaginable
then. We weren't thinking of process nodes that we have today. Frankly, for me,
it was a pleasant surprise. It's making
design simpler than it otherwise would have been.
When I did my first memory
board design, it had 4K of memory on one DRAM, and 256K used to be
considered a significant memory size. Today, we're looking at 32 and 64Gbytes! The
way memory technology has evolved so rapidly is amazing. We couldn't have
tapped the power of microprocessor without this.
Q: Talk a little bit about
the EDA evolution you've experienced in that time and how it contributed to
A: Design automation is more
efficient and productive than when I did my first design. Then, we handcrafted
the design. We would need to go through the data sheets and understand characteristics
and see how to interface each of them. From there, the EDA revolution brought
us a lot, not only in terms of models for each semiconductor gate and
functionalities that could be integrated but the level of verification that
could be done in advance because things were getting complex and expensive. Things
needed to be simulated beforehand. Such designs were possible only by
evolution of EDA.
Mixed-signal design tools
have been a huge innovation for design. It used to be just analog design and digital
design, and we couldn't imagine integrating them. Mixed-signal technologies
that have evolved have been very beneficial. That helps us integrate several once-separate
products in a single chip and create, for example, a watch. That integration
has happened because of mixed-signal technology.
Q: Given all that you've
seen, what do you see for the future of design?
A: I see the immediate wave
for the semiconductor industry will be from the Internet of Things. Home, road,
automotive all are being semiconductor enabled, and they're all connected through
the Internet. That will really multiply our opportunity. Mobile will increase
semiconductor consumption even more. And today machine-to-machine and
person-to-machine systems are becoming the next level of semi consumption.
The major challenge that I see
is process nodes are touching the limit. Are we going to be successful with
stackable approaches to design? Is there a new kind of semiconductor tech that
will happen in the future that we can take advantage of? The other challenge
that I see is the chip designs have become extremely costly. Think of a semiconductor
design in the context of its node and densities: You're looking at an $8-$10
million investment, at least, on the table. Startups are becoming rare, and the
level of investment for a chip is costly. Is there something new coming that
will make this affordable?
Take for example 3D printing.
You could prototype your chip at your office in a small fab. That might take
another 10 to 15 years but we need to think how are we going to improve productivity
and cost per design.
Q: How have engineering teams
evolved over that time?
A: The engineering team of
20 to 25 years back had to worry about the basic characteristics of transistors
and gates, and how to make them interface properly. Today, the focus isn't so
much on the functionality of those gates and transistors because they're preassembled
as IP blocks. Today's challenge is innovation: How do you differentiate your product
with respect to your competition. What's your secret sauce? How do you put
things together so others can't easily replicate your design? It's evolved from
basic processes and functionality to innovation and functionality.
Q: That, and we design around
the clock, around the world today, right?
A: Without the kind of
evolution in communications technology and growth in huge bandwidth networks,
we couldn't work as one global team. The critical success factor that has
defined global electronics development has been communications and bandwidth as
well as EDA tools evolution and licensing models. I see that around-the-clock
designs are happening at one end of the world and verification happens
somewhere else. It's as if all the teams are working under one roof.
That is a gift. That has
happened because of strong EDA design.
Q: Jumping off from that, how
do you see EDA tools and services evolving in the cloud-computing era?
A: The processing power is
amazing. We will be able to upload the design, do synthesis and other functions
much faster and more efficiently. Design teams can be more productive. The
cloud is a miracle and that's the evolution that has happened.
Q: EDA vendors haven't soared
into the cloud yet because there are concerns about the security of designs
being worked on in the cloud and concerns about the types of EDA tools (and
business models) that are best suited for cloud applications.
A: Security is not something
new; it has been a concern since the dawn of the Internet. I have seen
organizations hesitate to go global because they were worried about connectivity
and the public network. Today, there's a private
cloud to immediately alleviate those concerns.
Today, a semiconductor company
that has teams designing globally, can think of a private data center and private
cloud to start with. Downstream, as the public cloud becomes a reality and
evolves for trivial and mission-critical apps and as security provisioning on
the cloud mature, these concerns will go away. I think it's a matter of just three to five years.
Already the private cloud is
a reality in company like Mindtree. We're looking at some apps that are
critical in moving into a hybrid cloud: some run on private and some run on
public. These concerns are natural. We have faced in the past and overcome it.
Now it'll be to the cloud.
As for what EDA tools are
best suited to be cloud-based, I believe that processing-intensive tasks can be
done on premises but collaboration and integration can be leveraged through
the cloud after a basic level of processing and synthesis is done. Data bottlenecks
can be overcome, and bandwidth also goes higher in the future.
Q: Back when you were starting
out, did you envision the role Indian engineers and Indian design engineering
would play today?
A: I've seen two eras. Until
1990, India was a closed economy. Most of the things needed to be invented here.
We used to design our own databases and operating systems and our own compilers
because the Indian economy was not open. That hurt progress at time. But if India
didn't have a closed economy from '79 to '91, we wouldn't have learned the
technology on our own. We would be assemblers of computers. We had to
understand how computers are designed, put together, how tools are made, how compilers
are designed. We built skills and capabilities.
But at the right time—in '91
and '92—the economy opened. There was more import/export freedom. And then networking
technology took off, as did Internet and client-server technology. The world
got flattened. Whatever knowledge we got we were taking it to the world, collaborating
with worldwide leaders and providing design services. And we were able to
attract investment here, and IBM and
Compaq and others built design centers here.
That happened because of
talent and cost structure, and at the same time networking. Today, India has the
talent and capability to do complex designs. We were able to make things happen.
It is something that's a pleasant surprise for us. We're enjoying fruits of
investments in the early years.
Q: What do you want to see
from EDA vendors in the future?
A: That's a tough question.
How do we bring down the fabrication costs? I don't know whether EDA can help
on that front or the semi-equipment industry.
My major worry is that in recent
times the semiconductor industry has slowed down because the cost structure is
unviable for smaller companies to innovate. That kind of innovation has slowed.
How do we bring down the cost structure in terms of making new chips happen?
Whether EDA does that or some miracle elsewhere happens remains to be seen, of
course. That could revolutionize things. Otherwise it'll be more incremental
innovation and re-engineering.
In his role as CTO, Jani directs
Mindtree's technology vision. Jani has built the Product Engineering Services business of Mindtree
since its inception. A strong technologist, Jani created a deep research
division that works on building Ready to Integrate (R2I) Intellectual
Properties and Ready to Brand (R2B) Products in future technologies.
Prior to Mindtree, Jani
spent 19 years with Wipro, leaving as the Chief Executive of its Global R&D
Jani is the President of
the Indo-Japan Chamber of Commerce and Industries (IJCCI), Karnataka. He is
the founding member of India's Semiconductor Association (ISA), serving its
Executive Council for the first five years and later as Chairman and Advisor of
Jani holds a Bachelor's
degree in Electronics and Communications from the National Institute of
Technology (NIT), Trichy, India, and a Master's degree in Electrical Engineering
from the Indian Institute of Technology (IIT), Chennai, India.