( ESNUG 440 Item 11 ) ------------------------------------------- [03/03/05]
From: John Cooley <jcooley=user domain=zeroskew spot gone>
Subject: Goering, Santarini, Peggy, & Video on the DVcon'05 Bigwig Panel
Hi, All,
For those curious about how the infamous DVcon Bigwig Panel went (where the
ESNUG readers got to submit their edgy questions to the EDA bigwigs), there
are 4 interesting ways you can look at this.
The first is Richard Goering's clinical coverage in:
http://www.eedesign.com/news/showArticle.jhtml?articleId=60401205
why I say "clinical" here is because Richard only reports the exact words
spoken at the event with any audience reaction omitted in his report. He
got his report up within 5 hours after the event.
The second is Mike Santarini's situational coverage in:
http://www.edn.com/article/CA506700.html
where Mike provides insights into the background and politics behind what's
going on at the panel plus some of the audience reactions. Mike's report
was up 2 weeks after the event.
The third is a video of the panel itself:
http://www.deepchip.com/demos/bigwig05.fhtml
And the forth is Peggy Aycinena's cheeky coverage of the panel purely from
an audience viewpoint complete with blow-by-blow audience reaction to every
question and response:
"Comedie-EDA" by Peggy Aycinena
Tuesday, February 15th
Ah yes -- the annual Cooley Bigwig Panel at DVCon. In year's past, it's
been the source of moments and memories that have long stuck with the
industry -- things that people have said, flame outs they've been goaded
into, and responses to outrageous questions lobbed at panelists in a way
that only Cooley can do -- and get away with. No wonder every year, the
lead up to the event gets more frenzied and the crowd in attendance more
intense.
This year, the DVCon 2005 Bigwigs Panel was no exception. There were
outrageous questions, pointed jabs from the audience, a lot of raucous
jokes -- mostly at the expense of one panelist or another -- and a nice
little flame out here and there. And as always -- the principle players
in this little Comedie-EDA were dressed to the nines -- Cooley in his polo
shirt conducting from the podium and the participants themselves lined up
along the table on stage like seven peas in a pod -- all in dark jackets,
light shirts, and nicely pressed ties. Given how little creativity goes
into the costuming for this annual melodrama, it's amazing how much
creativity emerges from the improvisational theater of it all.
John started by running down the list of panelists. He ticked off the
names of the folks sitting to his right:
Rajeev Madhavan, CEO at Magma Design Automation
Pravin Madhani, CEO at Sierra Design Automation
Jacob Jacobsson, CEO at Forte Design Systems
Antun Domic, VP & GM of Implementation at Synopsys
Robert Hum, VP & GM of Verification and Test at Mentor
Ted Vucurevich, CTO at Cadence
Gary Smith, Grand EDA Poobah at Dataquest
Gabe Moretti, Industry Luminary
John played it straight with most of the introductions. He did, however,
embellish Vucurevich's introduction with the caveat: "Ted's becoming the
on-demand stand-in for Mike Fister here, there, and everywhere."
That little jab, not surprisingly, got the afternoon off to a lively start
with a round of spontaneous cheering, stomping, and applause. If I were
Ted, I would have been highly flattered.
John did not choose, however, to comment on the fact the Verisity CEO
Moshe Gavrielov had mysteriously dropped off the panel, replaced by
everybody's favorite EDA editor -- Gabe Moretti. Gabe was probably more
interesting as a participant, anyway. When you're the CEO of a company
that's being acquired by another company, erring on the side of silence,
is always the better side of valor. Gabe, on the other hand, is nobody's
fool and speaks his mind clearly, honestly -- doesn't having his hands
tied by corporate policy or legal defensive ends.
So John, having warmed up his audience, launched into the EverythingEDA.
He asked Madhani why he didn't have any Indian customers? Madhani, who
obviously had seen the proposed questions for the panelists on John's
site earlier in the day, leaped over to the overhead projector and put
up his prepared foils. He had pie charts to show the ethnicities of his
employees and pie charts to show the nationalities of his customers.
He pointed out that with 24% Indian employees in his company, he had one
point less in the ethnicity category than the panel itself, which he
said was 25% Indian. He said, "Ethnicity's not important, but to answer
this question -- I had to go cube by cube among my employees and ask,
'What are you?'"
You could ask why he had to respond to an inane question at all, but the
audience reveled in the naughtiness of it -- on account of sex, politics,
religion, and race are definite taboos at dinner parties and technical
panels. But, of course, nothing's ever taboo on a Cooley panel, which
is of course why everybody comes.
So John said, "So, Pravin -- you don't have an Indian strategy at all?"
Pravin said, "Okay -- so clearly, we're going to get dinged both ways."
John said, "What, you don't understand the culture?"
Pravin said, "Right -- I've just been here too long, man."
John said his next question was for Gary Smith. "Gary, why are your numbers
such crap?"
Of course, besides the taboos on sex, lies, videotapes, and race -- John
was violating the taboo of ever challenging the veracity of Dataquest
numbers. After all, this is bible -- and the stuff by which the entire EDA
industry navigates its course, year to year.
Gary's responded: "What? Did I offend some marketing weenie? I mean -- John,
you yourself use my numbers."
Wow -- that brought the house down!
John said, "Yeah, but why do you parse the tools into such weird subsets?"
To which Gary responded: "Well, people buy the report. You should buy the
report too, John, if you're interested. Maybe then, you'd understand."
After the laughter, Gary had lots more to say. He offered a detailed
discussion of simulation, hardware, ESL, software, algorithmic engines,
Celoxica, silicon performance, Simlink, the co-paradigm of HW/SW,
concurrent compilers. In fact, his answer was so detailed a response to
the broadside attack, that John and his audience stopped listening. They
didn't want facts. They wanted blood and humiliation. What fun was this?
So John moved on.
John said, "Rajeev? Was it a bad idea after all, to send that letter to
Synopsys?"
Rajeev Madhavan surprised everybody (not) by saying that his lawyers had
advised him that he wasn't free to comment on anything legal. That meant,
of course, that he wasn't going to offer up any trash talk about Synopsys.
But he did say that they'll always do what they have to do to protect their
IP, and that this industry has to turn down the number of fights they have
with each other, and that Magma is proud that they themselves have never
started a litigation, and they hope they never have to. He ended with:
"All Wall Street sees when they look at all of us is a bunch of people
fighting on the least provocation".
That was boring. We were all there to see a brawl, so who invited Rajeev?
So, John said, "Antun?"
Antun Domic said, "Take a look at it from Synopsys' side. We've had only
one serious litigation -- Nassda. And anyone could see that we had an
extremely strong case there. What's happening there will be in the best
interest of the customers. Synopsys did proceed to defend its position
after we received the Magma letter. But nonetheless, these two cases do
not constitute a large number of litigations."
The audience groaned. Whether it was because Antun was right, or Antun was
in denial, or Antun's lawyers had clearly given Antun permission to speak
was not clear.
In any case, Rajeev responded that he'd seen all he needed to see about
excessive litigations when he was involved at Ambit, and that the idea
that there aren't a lot of litigations in EDA is not true and violates
of all the we hold sacred in EDA -- which is that we litigate too much.
Everybody in the audience seemed to agree.
John asked Jacob Jacobbson to comment. Jacob said, "We want to side with
Rajeev, because EDA does indeed have the image of being very, very patent
infringement happy. The question is how to sort things out without a legal
fight. Our image here in EDA is definitely tarnished."
Mike Santarini, newly minted EDA editor at EDN, piped up from his front-row
seat in the audience. "Well, maybe people in EDA ought to stop stealing
stuff if you want to stop seeing so many lawsuits."
Yes, sir! That's what the audience wanted! Blood. The audience stomped,
hooted, and hollered.
There were more exchanges at this point between John and Jacob, which
allowed Jacob to wax poetic about verification, SystemC, and behavioral
design. He ended his musings with, "We can be successful where Synopsys
can not."
John said, "But even Mentor has abandoned SystemC -- so why persevere?"
Jacob said, "Abandoning SystemC is really weird. Especially when the
momentum for the language is clearly accelerating."
Then he too leaped over to the overhead projector and put up a foil that
indicated there'd been 200,000 (that's two hundred thousand!) downloads
of the free SystemC simulator. John -- The Knight in Shining Armor for
EverythingHDL -- was ready: "Come on, Jacob. Are you asking us to believe
there are 200,000 SystemC customers in the world?"
Jacob said, "Yes! Those are 200,000 true downloads of the SystemC
simulator -- and 7000 people have taken our class so far, which is in
itself a sign of the momentum." He went on a bit more, but Mentor
raised a hand.
Robert Hum said, "Mentor has not abandoned SystemC."
John said, "Yeah, but for synthesis? You don't do it. I know 'cause I
phoned up Mentor and asked."
Robert said, "We've got CatapultC working in the synthesis area, but
nobody wants SystemC in the state space. You don't need SystemC to
do that."
John asked Ted Vucurevich into the fray. "Ted?"
Ted said, "The picture's bigger than just synthesis. People are saying
that you've got to get the software guys into the conversation.
Unfortunately, this requires breaking with the past and trying to sort
things out. How to get from point A to point B is the question, plus
everybody wants one language. But, we need synthesis for hardware and
we need synthesis for software. And we have to look at the adoption
rates and the manufacturabilitiy. So, I would be very careful around
anyone who says they're already using SystemC."
So Antun replied: "SystemC is finding its place at higher levels of
description. To use it as an HDL is not warranted. To try and put on
one more language to do RTL synthesis doesn't add value. For companies
like Forte -- it may pan out. It may not."
Okay -- so, Mentor, Cadence, and Synopsys had had their moment -- now for
the editor & analyst.
Gabe jumped in before Gary could advocate: "We're still in the pioneering
stages of the ESL market, but we're having difficulty defining ESL. Look
at Gary. He's redefining the term every 6 months because there's a demand
from people who want to design at higher levels of abstraction. But, the
EDA vendors are trying to respond to the demand by just reworking the tool
chain. They want to live in the middle."
There wasn't a huge response to Gabe's comments. After all, for the bulk of
that audience have careers and capabilities invested in EverythingHDL, so
why should they celebrate anybody or anything that might upset the
EverythingHDL apple cart?
John changed the subject. "Rajeev and Robert -- are you pissed that you
missed buying Verisity?"
Excellent! More sass! More humiliation! More blood! The audience perched
back up on the edge of their chairs.
Rajeev said, "I can't comment. But I congratulate Moshe."
That seemed a little useless. Moshe wasn't there. But, once again, John
missed the chance to talk to the elephant in the middle of the room -- why
Moshe was a no-show. Meanwhile, Rajeev wasn't done: "The Cadence
acquisition will bring parity to the industry. Verisity needed better
channels and Cadence can supply that."
Robert, however, said, "I didn't realize this was a rumor mill."
Gosh, where's he been? Has he never seen The EDA Improvisational Theater
and Rumor Troupe in action before?
John was not dissuaded: "Are you telling me that Mentor didn't bid for
Verisity?"
In the time-honored tradition of anyone who ever sat before Joseph McCarthy,
Robert replied: "I can't remember."
Excellent. Blood flowing freely. The room was awash in laughter and
irreverence.
Robert said, "Testbench-based verification has been in the industry for a
long time. There's all sorts of stuff built around that. People are saying
that testbenches are a fact of life, they have to run them to make things
work. Verisity only has 12 percent of the total available market. Only
12 percent are using e or Specman. The rest of the industry is saying no
to the technology. But the industry needs to standardize things. Standards
drive adoption. We want to standardize things. The industry wants to
standardize things. The battlefront has moved beyond testbenches. Now,
asserting based verification, coverage-based verification, and formal
techniques will be where the new products will come from. For Verisity,
however -- we're happy that they have a home. But we're not particularly
torqued about it."
John yawned and said: "Ted? Are you getting squishy? You're saying nothing."
Ted smiled: "We've had 50 or 60 acquisitions in 12 years. This one is
interesting, but so have the others been interesting. We wanted to make
sure that this was the right thing to do and it was. We're gaining an
excellent Israeli team -- and we saw the chance to integrate great people
and great technology."
John would not be put off: "Specifically, Ted. Are you going to kill e?"
Ted invoked Axis, Palladium, Incisive, customers, simulators, emulators,
and bottom lines. John yawned again, and allowed Rajeev to jump in and
clear things up: "I was active in OVI when VHDL and Verilog were fighting."
Rajeev said there would be similar dynamics while people sorted out between
the different verification languages, but he was sure that eventually there
would be a single language that would emerge to save the day.
Nothing was cleared up, however, so Gabe said: "In 5 years as an editor,
I've seen that the people in the Palladium group would rather commit Hari
Kari than admit the technology was no good."
So Ted said, "Axis starts as an accelerator for simulation. Palladium is
inside the accelerator for emulation. They're competing for design space,
but both are excellent technologies that can drive the market. Eventually
things will ferret themselves out."
Ferrets are good, so John turned the screws, once again, on Antun.
John, "Antun? How many customers have you lost to BlastChip?
Antun was sure: "Very few. There's lots of talk about other options for
synthesis, but we've also seen dramatic improvement in Design Compiler.
But yes, it's true -- we do see some competition."
John: "What about people posting better results with Get2Chip?"
Antun: "We're never going to have 100 percent of the customers. But we're
extremely confident in DC. Our numbers continue to be excellent."
Rajeev got in a marketing pitch: "The job you need to do in synthesis is
very out of date. The value is going down rapidly because much of the
decision must be made at the backend. Now it's about how much should be
done at the front end versus what's done at the back end. That means a new
flow and a new opportunity."
Antun was not intimidated: "We must be careful with things you say and
about what can be done. We agreed 5 years ago -- things change and everyone
needs to do more. But you have to be very careful here. Astro,
Primetime -- we have seen what changes. Customers have seen this in many
cases. And synthesis -- Synopsys or not -- must be used to its best
opportunity. But, I do agree we need to do more."
The audience slumped in their chairs, while the battle for hearts and
minds continued.
Rajeev was still in the zone: "You don't tape out a netlist. You tape out
a GDS II. None of the other things matter. As time passes, what matters
is how do we maximum first silicon? We've had a front-end tool and a
back-end tool, so there's been two strings. Can we to one? Absolutely,
but there's inertia! And the move from front end to back end won't
happen overnight."
More minutes ensued as the technology banter zigzagged back and forth; all
in fun, at least for those sitting at the panelist's table. Clearly neither
Ted, nor Antun, nor Rajeev were going to back down. That was a good thing,
because the audience loved the rancor. Probably for the same reason that
people like to go to cockfights -- they're a little tasteless, and a lot
lively.
John called in the arbiters: "Gary? Gabe? Is there any truth in anything
these guys are saying?"
Gabe started: "Synthesis tools have become content sensitive. But, Gary
and I are wondering -- the ASIC counts are going down, yet apparently
everyone is buying more synthesis tools. Something doesn't compute here."
A'stomping and a'hooting ensued. The cockfight was ended because the cops
had been called in to put a stop to the thing.
Officer Gary said: "The seat count's going up even though the number of
design starts is going down. That's because we sell to sectors, not to
designs. Also -- the price wars have been horrible!"
Rajeev and John duked it out for a couple minutes over whether Magma was
bigger than Cadence and Synopsys combined in the digital P&R with
synthesis market. Of course, Ted and Antun didn't want to walk away, so
they jumped with 4 fists swinging as well.
And in a heartbeat, they were all yelling: Rajeev, John, Antun, Ted; nobody
could be heard. The audience was hooting. Somebody thought about calling
hotel security. The videographers looked nervous. But the thing was settled
when --
Ted said: "We've had 100 90-nanometer tapeouts!"
Rajeev said: "Well, we've got 60 percent of that market!"
Antun said: "Well, we've got 50 percent of that market!"
John said: "Well, add 60 to 50, and that means Cadence has minus 10 percent
of the market. Is that right, Ted?"
Bedlam.
Ted went philosophical: "What does it mean to say that you're part of the
flow?"
John appealed to Gary.
Officer Gary said: "It's the number of tapeouts versus dollars spent.
Rajeev's done a good job of holding discipline -- of selling tools for
what they're worth. Many others are giving away quite a few tools for free.
Who's got what share of the market? It's a 1/3, 1/3, 1/3 to each vendor."
John turned to Pravin: "Do you guys actually have tapeouts?"
Pravin said: "We're bringing a solution to market that integrates into any
scenario and gives an overnight productivity boost, as design starts move
to 5 to 10 million gates. We can lay claim to some tapeouts because we
provide that functionality."
John was nonplussed: "So Pravin? Who do you want to get acquired by?"
John's mob went loud and unruly.
More blood followed -- Ted versus Rajeev versus Antun. Blathering about
OpenAccess. Who was interoperable with who. Who truly cared. Who didn't.
Accusation followed insult followed nonsense.
Now things were pushing up against the 5 PM finish time. John addressed
rumors about DAC. He turned to Ted: "Ted, tell us more about Cadence's
situation at DAC."
Ted responded with confidence: "We're starting to see a transformation.
The focus is more on the technology than on trying to sell things. In the
past, DAC was about telling our friends and neighbors what each of us is
doing. But, last year we brought in the technology and we had chalk talks
with real customers about what we're doing. People thought that was really
cool. We believe it was much more effective. But the booth won't change."
People in the audience whispered. Wasn't the word already out on the
street? Cadence's booth had shrunk from a 100x100 in 2004 to a 40x40 in
2005. Had Ted not heard?
John was quick: "So are you guys reducing your presence at DAC or not?"
Ted was also quick: "I suspect the number of people there will be less."
Good Citizen Robert jumped in on behalf of Mentor: "EDA is a very global
industry. So, DAC is a good clearinghouse for our customers to meet each
other. DAC is very, very important to us -- the technical papers, the
sessions. And DAC can't exist without investment from the industry.
Also -- DAC is a place where we're always happy to be getting resume
from people at Cadence."
Stomping. Whistling. Guys hanging off the rafters banged on the ceiling
in ecstasy.
Good Citizen Antun threw in a Synopsys benediction: "We're seeing changes
at DAC, but there's a big amount of money to be made at DAC. Yes, the
conference is changing, but we believe it's improving."
John drilled down into the topic: "Jacob? Will you be increasing your
presence at DAC? Will you have 3 people this year at DAC, or will you
be upping that to 4?"
Jacob spoke up on behalf of Forte: "This year we'll have much more
opportunity to talk about production. So, we're definitely stepping up
our DAC presence."
The stomping and rafter banging continued.
So, John let Pravin have a turn: "Pravin? How many Indians are you
bringing?"
Good Citizen Pravin was ready on behalf of Sierra Design: "We're a small
company, but we take every opportunity we get to talk about our technology.
Our booth at DAC is going up in size by 25 percent."
John turned to Rajeev: "Rajeev?"
Good Citizen Rajeev said for Magma: "We're still at the stage where DAC
has value for us. You get to see people and participate in round tables;
and it's a great networking opportunity!"
So Gabe gave it one last shot: "Ted, how can you talk with customers and
potential customers if you don't have demo suites on the floor at DAC?"
Ted looked everybody in the eye and said, "Our floor space will be the same."
And then -- perhaps on cue -- Gabe asked: "Ted? Are you still going to
invite John and me to your party?"
Cooley didn't wait for Ted to answer.
He proudly announced: "I never get invited to the Cadence party!"
The John Cooley Fan Club went wild. Something had to be done, and although
John knew he could have had more, it was past 5 PM and he had to reign
them in.
Cooley said, "Thanks for coming, everybody. It's time for a beer."
Index
|
|
