( ESNUG 424 Item 2 ) -------------------------------------------- [03/09/04]
From: John Cooley <jcooley=user domain=zeroskew spot calm>
Subject: ( ESNUG 423 #13 ) Richard & Peggy's Take On The DVcon'04 CEO Panel
Hi, All,
I've had a number of people ask me what happened at the DVcon CEO panel
earlier this week. To see Richard Goering's report click on:
http://www.eedesign.com/news/showArticle.jhtml?articleId=18201864
and below is Peggy Aycinena's report on the same panel. Enjoy.
- John Cooley
the ESNUG guy
---- ---- ---- ---- ---- ---- ----
"The Halftime Program" by Peggy Aycinena
When you're late to John Cooley's Talking Heads panel at DVCon in San Jose,
CA, you can rest assured that ValleyPR's Georgia Marzalek will already have
figured out how to see over the heads of the 30 people spilling out into the
hallway and the 300 squished into the ballroom -- only 200 of whom had
places to sit. Georgia had dragged a chair from across the hall up to the
crowd at the doorway and was happily standing on it, perfectly able to see
over the hundreds of heads between where she was perched and the front of
the ballroom where John's annual circus was underway.
Inspired, I dragged a chair up next to Georgia's, and hopped up as well.
She was right -- the view was perfect. Two other guys caught on, joined us
with their chairs, and now there were four of us in the back of the crowd,
all perfectly able to see up to the front. But then of course, some wet rag
of a hotel authority had to come rushing over and insist that we get down
off those chairs immediately, or else!
How annoying -- it was impossible to see, let alone hear, standing on the
ground at the back of the crush, so Georgia tried a different tack. She
indicated that I should follow her lead as she bolted straight into the
crowd and relentlessly threaded her way between the bodies right into the
room. I did as Georgia did, and in a second we were both inside, standing
shoulder-to-shoulder with the crowds along the back wall. Having done that,
it was finally possible to fully concentrate on the antics of Cooley & Co.
The room was rocking with laughter. Panelist Brett Cline, VP at Forte Design
Systems, had just presented a dozen Krispy Kreme doughnuts to Ray Bingham
and had poured a fresh glass of water for the Cadence President/CEO, as
well. Cline and Bingham were two among the six sitting at the panelist table
on stage. Magma Design Automation CEO Rajeev Madhaven was also there, as was
Mentor Graphics CEO Wally Rhines, Synplicity CEO Bernie Aronson, and EDN
Magazine's Gabe Moretti.
Ringmaster Cooley was standing at the podium, orchestrating the show and
holding them all at bay with a whip and a chair. It was good-old-boy theater
at its best, and the crowd loved it. Standing, sitting, leaning, or
straining to hear - everybody in the room was having a whale of a good time.
Cooley threw out question after question in rapid-fire succession -- most of
the topics were available last week on ESNUG, including the bit about the
doughnuts -- and the panelists, relaxed and in their element in front of
the sweltering crowd, answered his questions in kind. The give and take took
a little over an hour to be given and taken. Here are some highlights:
Cooley: Ray, why don't you have a System Verilog product?
Ray: I don't actually know if it's needed, but our Incisive platform gives
us the ability to connect to a variety of tools and languages.
Gabe: It's not about the HDLs, it's about platforms giving engineers a
choice of languages to use at a particular place. Whether it's SystemC or
System Verilog, the breakthrough is in the platform.
Brett: System Verilog is what Verilog should have been 5 years ago. At the
end of the day, it's just another implementation language. RTL will be the
next Assembly language. Do something different, or you guys will die.
Wally: We're the only company shipping products in System Verilog. We've
just had our second largest order in the history of the company, and the
customer requested that we support System Verilog. We certainly can't argue
with customer demands, which is of course why we do SystemC, System Verilog,
Verilog, and VHDL.
Rajeev: If we have to support System Verilog, we will. It's not a highly
expensive thing to do.
Cooley: Ray, why don't you see what Wally and Rajeev see? [Laughter]
Ray: You know, it's not clear what market exists for a separate tool and
you can still play in the market without owning the market. Cadence isn't
in the business of System Verilog -- that's an incremental product -- we're
in the market for verification.
Cooley: Is Synopsys' paying $430 million for an IP company [MoSys] brilliant
or a disaster?
Ray: It raises questions about the earn-out plan. With that level of
commitment to IP, that particular type of IP, can you earn an EDA-market
style of return? Is having a lock on the market a positive thing? I've lived
with the hell of our customers looking at us as if we were trying to compete
with them. If the net [result of the MoSys purchase] is that Synopsys is
competing with its customers, then there's no benefit to be had in either
the horizontal or vertical markets for them.
Rajeev: If the move means that Synopsys is willing to compete with its
customers, then most of their customers will say their competitive delta
is gone.
Ray: Where's the leverage in providing IP?
Cooley: Well, I think of both Synopsys and Mentor as being in IP. So, why
isn't Cadence there? [Laughter]
Ray: Yes, we have IP, but it's non-differentiated IP that zeroes in on
partnerships.
Cooley: What's going on in the synthesis market?
Rajeev: We're not in RTL synthesis. Design Compiler is almost a thing of the
past -- Design Compiler for synthesis has lost its value. The back end is
where the final creative solution will be. Here's a good opportunity for
Magma to get into the market and blow Synopsys out of the market. The game
has changed. People and their inertia is the only reason that they're still
using Design Compiler.
Cooley: Gosh, Ray, maybe you should be investing in Magma. [Laughter]
Ray: Well, maybe Magma's just watching us. [More laughter] You don't invest
in RTL synthesis -- you invest in synthesis, because physical synthesis is
changing the way people design. The old tools are getting long in the tooth.
Cooley: Yeah, Ray -- that's why you make the investment. [More laughter] So,
where are we with OpenAccess?
Wally: We're aggressively working on things there.
Rajeev: Whether it's OpenAccess or Milkyway or Oracle, when the customer
asks for it, we'll spend a couple of months and meet that demand. But it's
really just a way to go backwards in time.
Ray: Rajeev had done a job on validation with a unified database.
Rajeev: Actually, Magma doesn't use a unified database. We use a unified
data model; an important difference. In fact, there are big religious
differences between a database and a data model.
Gabe: Yes, but a data model works best in a closed system, and Magma has a
closed system. Okay, so what happens if a customer wants to open that
system to a third-party vendor?
Rajeev: Thank you, Gabe, for that marketing opportunity. [Intense laughter]
If you want to write a Tcl router around the model, you can do that. It only
requires a little work, and we'll do it when we're asked. We can write out
into any format -- LEF, DEF, GDSII, or our application.
Gabe: And when you do that, you have to deal with the interface.
Ray: It's a matter of what the customers are doing, the success our
customers are seeing; the most widely advertised success of late being HP.
It's actually very interesting. From our perspective, we need to look across
all levels of our business. It's clear that Rajeev's focus is a bit smaller
than ours. [Lots of laughter]
Rajeev: Ray, I challenge you to name a tool that you have that we don't have
at Magma. We've spent $100 million developing our tools and the companies
are using them.
Ray: Then I guess the customers are misguided. [Really lots of laughter]
Cooley: Bernie?
Bernie: I guess I've been away from engineering too long, because I want to
know how do we enlarge the business for EDA companies. Is there money in the
ASIC space any more? The number of ASIC starts is down, but there should be
a bigger market for the ASIC guys. Both NEC and LSI, for instance, are
moving to 90 nanometers. What we're seeing now are design wins in the
structured ASIC area, where the customers can go before going to the ASIC
itself.
Wally: I agree that platform-based ASICs are certainly a way to reduce
costs, but the jury is still out on structured ASICs. We're attacking some
of the fixed costs in the tooling. At the same time, we're seeing Japan,
for instance, going with platforms.
Cooley: Ray? Rajeev?
Ray: We should be concerned with growth and long-term trends. Certainly
there's a trend in the ASIC market today which is causing real pain.
However, although ASIC starts are down, the content is not down. The teams
are not getting smaller, because the chips are getting bigger.
Cooley: So, eventually, we'll all just be designing one Big Chip? [Huge
laughter]
Ray: I don't know. You need to pay attention to your strategy and your
focus. The money seems to be going to SoCs. The SoC opportunity is a large
growth market.
Rajeev: The number of design starts is down, but the number of gates is
increasing. Our licenses are proportional to the number of gates on a
project. So, if the total number of gates is going up, the amount we're
being compensated is going up. Of course, we're working to bring the cost
of silicon down. The ASIC versus FPGA thing is an immaterial argument here.
It's always true that some parts of a design will change again and again.
That's forcing customers to do a mix of structured ASICs and standard cell.
If you make a religious battle between FPGAs and ASICs, nobody wins. The
whole thing requires a medium ground where you can mix and match.
Cooley: Ray, should I move to India or to China to find work with you?
[Ripple in the crowd]
Ray: Which job are you applying for? [Huge laughter]
Cooley: Okay, then with Cadence doing so much training in other places,
should Silicon Valley just move? [More laughter]
Rajeev: In the last 100 jobs we've hired into, we haven't taken even one
job from anywhere and moved it to India. For Magma, it's because we're
doubling year over year, which we're achieving through some things in
India, some in China, and some in Silicon Valley. We need local support
and big chips are going to be done everywhere. But it's a myth, the idea
that we're paying $3 per hour anywhere. In fact, the peripheral
infrastructure needed for us to support a remote developer is such that
Magma is not getting any savings at all from off-shore outsourcing.
[Boos and hisses from the audience] Look, if we want to get a top graduate
from one of the campuses of IIT in India, we have to pay quite a bit. And,
it's not a factor of 1 to 5 [compared to a U.S. graduate], it's maybe only
a factor of about 1 to 2.5, or so. We haven't moved a single job from
Silicon Valley to anywhere else in the world. And like Wally said, we're
growing furiously and we can't grow like that if we don't hire everywhere
in the world.
Cooley: Ray, is Cadence laying off here and hiring in China?
Ray: Within which time frame? [Laughter and hisses]
Cooley: Like last month, or in the last few months?
Ray: Okay, there's no ducking the answer to these questions. Quite honestly,
there are two reasons for looking at China and India. There are very
qualified and talented people there who can be employed at a fraction of
the cost. But second of all, the customers are demanding that you build
capability in those locations, in both R&D and access. These are demands
which I believe any company must respond to. The question is then: How does
that work overall? The highest paid software job 30 years ago was for a
COBOL programmer. But people don't employ that technology today. The fact
of the matter is that, whether it's India or Mexico or China, that's the way
the world economy works today. If you try to control a market -- whether
you're a company or a nation -- you'll lose in the long run.
Cooley: Are you selling the tools at lower cost in China or other markets?
Ray: I don't think that's accurate.
Cooley: Or is it because so much of the software is stolen in those markets?
Rajeev: None of us here, that I know of, are selling tools cheaper in China.
That doesn't mean, of course, that piracy isn't an issue.
Cooley: Wally, you're being very quiet. [Laughter]
Wally: I've been giving this speech for 25 years. If you go off-shore in
design or R&D just to cut costs, you'll be sorry eventually. In the early
1980's, we opened a center in Bangalore, India. The cost for a developer
there was one-twentieth of what it is today. Every year, the salaries there
continue to go up. If you only play in one economy, you'll lose money. We
have to play in all economies, to be competitive. And, there are outstanding
graduates and customers in every part of the world. [Applause]
Cooley: [Looking at a slide up on the screen] Cadence has 473 employees
working with an H1B visa. Synopsys has 208. Mentor has 145. Magma has 50.
Synplicity has 43. Forte has 3. Reed Elsiver [publishes EDN] has 55. Are you
all telling me that none of these jobs could have been filled by qualified
American citizens? [Huge restlessness in the audience]
Brett: Actually, Forte only has 2 H1B employees. One has a Ph.D. in EE and
is highly skilled and the other one is a very driven AE. If we were to find
people in the U.S. with that same level of skill and commitment, we would
hire them.
Bernie: In 1997, 1998, 1999, and 2000, we tried, but were unable to hire any
developers here -- up until the crash in 2001. In those early years, we
hired in Bangalore, and in Portland, and in the South of France. In 2001, on
the other hand, we were suddenly able to hire 42 qualified employees,
compared to the less than 40 that we'd been able to hire in the previous
4 years. Clearly, times have now changed. We did hire a lot of H1Bs before
the crash. And right now we're actually hiring, but we're hiring in
Sunnyvale, not in Bangalore. We would actually rather hire here because when
new employees come on board here, they can hit the ground running. There is
more of a training issue in India, but with all of the people going to India
right now, that may change within the next 3 years or so.
Gabe: The issue of H1B visas is a spurious one. There are some very strict
rules set out by the government that you must follow, principally the
requirement that all employees receive equal pay for equal work. If you
break those rules and they catch you, watch out! There is another
consideration; I would prefer a system where it's possible to hire qualified
foreign professionals to work here, because when they work here, the jobs
stay in the U.S. Also those employees pay taxes and contribute to the U.S.
economy.
Cooley: Given the situation today, would you advise your children to go into
engineering?
Wally: Engineering is a good and productive career.
Brett: Absolutely.
Bernie: It's the single most exciting profession I know of. I see lawyers
and doctors getting bored with their careers. [Laughter] But working in
engineering allows you to meet the brightest people both here in the U.S.
and everywhere in the world. I would most certainly recommend this career
to my kids.
Gabe: If you want to deal with interesting problems, engineering is it. If
you want to be respected, and, on average, you want to make a fair amount
of money, go into something else! [Laughter and applause, long and loud]
- Peggy Aycinena
Editor's Note: This week Peggy moved from http://www.EDAtoolsCafe.com to
http://www.SoCcentral.com -- her colorful columns can be seen there now.
|
|
