( ESNUG 576 Item 1 ) ---------------------------------------------- [09/22/17]
Subject: Anirudh, Sawicki, Hogan go at it over the CDNS Pegasus DRC launch
DAC'17 Troublemakers Panel in Austin, TX
Cooley: Anirudh. So, I'll just read the question. "Cadence, under
your leadership recently announced you're going after Calibre
in the DRC LVS market. Are you crazy?" [audience laughs]
"Does Cadence have an employee drug testing policy and why aren't
Cadence executives being tested for psychedelic drug use?"
Sawicki: I just want to make clear that I didn't have to give that
question. [audience laughs]
Hogan: I'd answer the drug question.
Anirudh: Well the funny thing is I think I've been asked that question
before. [audience laughs]
Cooley: Oh really?
Anirudh: In a different context, different products. I think so far, I
like our batting average. In terms of DRC, first of all, let's
remember that was Cadence's first big product [Dracula] a long
time ago, right?
Cooley: Yeah but that's like 20 years ago.
Anirudh: But still it's a critical piece of the puzzle. And also, what
I've said before, at lower geometries it's very important to
integrate DRC with implementation. So, I believe at this point
Cadence is the only company that has a very strong analog
implementation portfolio with Virtuoso, and a very strong digital
portfolio with Innovus and we need to complete that picture with
Pegasus which is a DRC signoff tool.
So that's the reason we are very serious about DRC and when I
joined Cadence with the help of Lip-Bu Tan, the first team we
built was the Pegasus team. It's a brand-new team. They've been
working at it for almost 4 years.
Cooley: So, you were scheming at this for 5 years?
Anirudh: It's called planning, John, I would not call it scheming.
[audience laughs]
But it's very important to have. And I think with DRC, there
has not been a real new tool in a long time. There is
opportunity there, and there is integration with implementation
that needs to be leveraged. So that's the reason we at Cadence
are doing DRC.
Cooley: So, Joe are you ready to capitulate the market to Anirudh?
Sawicki: Oh absolutely, it gives me a chance to go home for a while.
Yeah, I can't help this one. You [Anirudh] served me a softball,
and I think it's appropriate -- Pegasus is a mythical beast --
I'm having trouble finding it in the market. [audience laughs]
Anirudh: You know I have heard the same thing before, so let's not get
ahead of ourselves. We just announced the Pegasus product, we
are very confident, it's a new architecture, and it scales to a
1,000 CPUs.
This is Cadence -- it is not like any startup company doing
DRC. I think DRC people say it's a difficult market, how are you
going to succeed?
It's very different from a startup doing it, because we have the
implementation infrastructure with Innovus and Voltus. We have
all the PVS run sets. And the CDNS sales force is well-trained
to sell DRC and the integration with physical. So, I think it's
going to be very different this time around.
Cooley: How can you have a run set for a new tool?
Sawicki: So you have a 10nm runset for Pegasus?
Anirudh: Pegasus, just to be clear, runs all PVS runsets.
Cooley: So, the old Cadence PVS runsets?
Sawicki: Certified?
Anirudh: So PVS has certified run sets at 7nm, 10nm, and all the way
up. So, the run sets are available from the foundries.
Sawicki: But you and I both know, you have to certify a run set on a
specific DRC tool. It's not just a runset certified across any
tool.
Anirudh: Yes, so Pegasus will run all PVS run sets.
Sawicki: Still a different question.
Anirudh: So, I think when you have a new binary, yes, we will certify the
runsets. But most of the work -- I don't need to tell you that,
Joe -- most of the work is producing the runset. Of course,
when you have new versions, even if you have different versions,
of PVS, we will run the runsets again to certify them.
To me that is not the hard part, the hard part is actually having
the runset in the first place. At this point, because of Cadence
we have runsets at several foundries.
Cooley: Several foundries meaning TSMC, GlobalFoundries, UMC... I don't
mean to name names...
Anirudh: Yes, TSMC is a big partner of Cadence and they have PVS runsets
all the way from 7nm to the previous nodes, 10nm, 16nm...
Cooley: Joe?
Sawicki: Yes, they do -- but only for Cadence PVS.
Cooley: So [Joe], you seem to be like smoldering on this, like there's an
Achilles heel here or something.
Sawicki: No. We had this discussion a couple of years ago, when it
was Cadence PVS at that time that was going to take over
Calibre because "you needed to the have the integration with the
design space."
And the one thing I still stand by, is introducing new tools in
EDA is a really weird beast.
It takes an enormous amount of energy to develop the tool in the
first place. It takes an enormous amount of energy to get people
to start looking it. It takes an enormous amount of energy to
get your first couple of customers to drive that through.
And if you've seen over the years, it's hard to move major
positions. And I've always said that you need two things. One,
you need some kind of discontinuity, and two, you really need the
other guy to mess up.
When Calibre came on board, yeah you had the discontinuity of
design size and hierarchy, but Cadence Dracula did some things
at the time -- and I don't think anybody's still here from back
then -- that really helped Calibre a lot.
Hogan: That would be me. [audience laughs]
Sawicki: And I appreciate that. [audience laughs]
[ Editor's Note: Jim Hogan was in charge of Cadence
Dracula/Diva when Mentor Calibre stole the DRC
near-monopoly away from Cadence. - John ]
Hogan: Oh, and BTW, and I'm sitting here...
Sawicki: I look now, it's like you know you saw the announcement, and
it's like "oh, the default tool does not scale." Well that's
delusional. We're scaling out to... I've got tests with certain
customers out to 1,000 CPUs.
Now let's be clear. No customer on the planet wants to have to
use 1,000 CPU's on a runset! We got people out to 7nm test
stuff going on right now. About the most we got from our biggest
customer is about 200-240 CPU's.
No one wants to have a lot of CPUs. What they want is an
overnight DRC turnaround time and that's what we've been
concentrating on for years. So, I am comfortable in our
Calibre technical positions. They've talked about before
that our competitor for ages has not been another DRC tool,
it's been the node.
Cooley: Really?
Sawicki: Yeah, these things are just... you turn around every node and
everyone's got to verify twice as many shapes. And that's a pain
in the butt. But the design rules are increasing by anywhere
from 2X to 4X on every node because we've been pushing the
physics to where it just makes accomplishable shapes so difficult
to do.
We end up having to deal with what are like 8x computational
issues going from node to node. That's been our competition.
When someone else comes onto the board, we'll play in the game,
we'll see how what happens. Everyone should be trying to push
into these things, God knows I'm trying to push into some markets
we're not winning either. It's just not as easy as saying "I got
a rule deck, I got a tool, and by the way the other guy sucks."
Cooley: Ok, Jim, Pegasus verses Calibre. Who's gonna win?
Hogan: Well when you own the majority of market share, you've only got
one way to go, and that ain't up. So, you know what? Joe and
his team have done a great job. Kicked my butt. Congrats,
right? I might have been thinking about something else at the
time.
Cooley: That was 20 years ago.
Hogan: Yeah it was 20 years ago. But with that said, you ought to
ask Antun Domic of Synopsys about what he [Anirudh] did to
place and route. So, I wouldn't just ignore Anirudh. It's
gonna be a good race.
And you know, bringing new technology to market at a time of
technology disruption both in software and hardware -- you can't
ignore that, right? There's an opportunity.
So, Joe, you're going to do a great job of fighting, I know, and
trying to maintain your market share. But this guy [Anirudh] is
hungry and he wants it. So, I'm anxious to see what's going to
happen. I wouldn't discount him.
Anirudh: Joe, okay, I have 2 messages. First message for you, please
don't worry about it. It's not a problem. [audience laughs]
For the rest of you, please come talk to us if you want a next
generation DRC tool. [audience laughs]
Sawicki: I'm looking forward to seeing the first benchmark.
---- ---- ---- ---- ---- ---- ----
Related Articles:
Anirudh, Sawicki, Hogan go at it over the CDNS Pegasus DRC launch
Dean Drako on IC Manage PeerCache P2P caching EDA tool accelerator
Prakash sees 3X revenue in 2017 by crushing SNPS Atrenta Spyglass
Amit on Solido machine learning, lib characterization, and Siri
Join
Index
Next->Item
|
|
