Cadence has sold 300 licenses of Ambit, including
40 during the last quarter. In addition they sold 11
licenses of PKS. At $100K a pop for DC, 300 seats of DC
amounts to $30,000,000. That's US. 11 Physical Compiler
seats at 250K results in $2,750,000. It seems to me
that Synopsys needs to start making money the
old-fashioned way: good products sold by people with integrity.
Tall order, isn't it.
Regarding the Ambit threat and NT: I hate to say
this, but you are a full 100% wrong (not even
99.99999999 but a full 100% wrong) on this. The NT port does
not require rewrite in C++. The 3X improvements were
due to algorithmic and architecture changes and had
almost noting to do with the NT port. Although SNPS may
have planned continous improvements, what suprised
people (including Madhavan) was the ability to speed up
3X on what was considered a relatively stable and
mature tool, which normally means that there is not much
scope for improvements. Again the point was that do not
count SNPS R&D out.
In general, porting is a
task taken care of by porting specialists who have
done this work before; they do not and NEED not
understand synthesis at all. No software company will put
their core technologist on a port; their time is just
to valuable for that. Specially so in the EDA
industry, and the synthesis market, where the NT port is
largely symbolic; very few people actually run synthesis
Cooley: He may be hand in glove with
SNPS but that is the best we have right now. He is
known to post negative stuff about SNPS too. With the
absence of any other information I would trust his
comments, since he gathers feedback from the user base. In
the internet age any strong bias or misinformation
can not last long; there are simply too many
information resources availble.
endorsements: The issue was the stability and quality of PKS vs
PC. As you yourself state Matrox engaged with PC for
18 months, that is more than the time Ambit has had
to work with Cadence tool. It somehow validates
Cooley's statement about PC being 6 months ahead of PKS in
terms of the state of the tool.
AVNT and PKS: I
did not understand what exactly you wanted to say.
Essentially PKS is shutout from AVNT accounts due to
political, support and flow issues. This provides PC
"terra-firma", something to stand on while the PC vs PKS
1."nVidia may be a Synopsys pal. But what about
Matrox had engaged with SNPS for
over 18 months prior to the announcement - long before
PKS. In fact, Matrox did not benchmark PhysOpt (no
Physical Compiler) against anything else (PKS or Magma).
So their investment in the technology, and
subsequent PR announcement, is meaningless.
released a new version of Design Compiler which was three
times as fast the previous one."
Yes, SNPS came out
with a quick release of DC after what appears to be
the Ambit threat. What outsiders don't know is that
SNPS was already engaged with a major personal
computer manufacturer to port DC to WinNT. This meant that
old DC code had to be re-coded to C++. This was many
months prior to Ambit's threat. The release of DC with
3X improvement was a result of the portation to C++
- NOT a response to Ambit. SNPS benefitted from the
3. Cooley is a paid
"employee" of SNPS. His comments are worthless (see my
4. "AVNT customer cannot even
engage in PKS evals."
AVNT can't even support their
own Jupiter evals, never mind getting included into a
PKS or PC benchmarks. Thay have no AE's that know
synthesis. The good AEs get gobbled up by competitors after
they realize that they can get MBOs/bonus at any other
EDA company. So, AVNT has to home grow their own
synthesis AEs. How can they compete against the SNPS AE who
has 7-10 years experience? NOT.
strategy is to screw the whole industry"
agree with you more on this one.
>1.nVidia may be a Synopsys pal....
still too early to tell anything about the PC vs. PKS
battle. Both CDN and SNPS have internal test cases and
know which types of designs will work for its product.
One real customer doesn't make any difference at this
>2. True, its a very close battle for logic
synthesis. But your info is wrong: Madhavan had already been
booted way before the Ambit synthesis tool was
>3. Cooley has no objective credibility anymore.
There have been strong reports about PKS in beta stage.
Also, check out the Mitsubishi press release last
>4. I agree with you on this one (for
>5. CDN's drop in synthesis prices was intended to do
damage to SNPS margins, and to take the focus off of
low-end logic synthesis and onto physical synthesis (they
should have waited until PKS was ready - marketing
screw-up here). Notice that since the price drop, CDN's
market valuation has resurpassed that of SNPS. So far,
CDN's strategy has worked. Lower valuations makes those
place&route acquisitions a little tougher, no?
1.nVidia may be a Synopsys pal. But what about
Matrox endorsement? Both these companies are in the
fore-front of designing multi-million high performance
ASICS; perhaps the toughest designs in the business.
Sure they would have some reason to put their eggs in
the PC basket.
2. About the Synopsys culture:
When Ambit came out with their new tool (simpler to
use and faster, with competitive QOR), SNPS released
a new version of DesignCompiler which was three
times as fast the previous one. Madhavan(Ambit's
orignal CEO and currently Magma's CEO) himself
acknowledged the surprise which Ambit folks had about
Synopsys' ability to turn around so fast and suppress the
threat to their market share. So do not bet on them
giving up their share of the market.
article about PKS being 6 months behind PC: Cadence
bought Ambit about an year and a half ago. Given all the
distractions which follow a cash merger (Read developers busy
buying their new houses and cars with their millions)
and the turmoil within the Cadence organization, I am
not surprised at the statement. Integrating two
different tools and flows into a reliable and effecient
solution does require a signficant engineering effort and
validation. On the other hand Synopsys has been working on
their Physical products for a number of year and is
bound to have some engineering lead. The six month
estimate is possibly on the more conservative side. That
is why we have not seen a single customer
endorsement in terms of tape-out or flow documentation
inspite of the Cadence spin machine working over time on
killing Synopsys' cash-cow (latest from the earnings
release " ..11 new orders..."; no mention of number of
customers, number of seats and the actual sale price or
revenue (perhaps they paid the customers with free
licenses for other tools to try PKS!! (just
4. AVNT factor: Presently AVNT customer cannot even
engage in PKS evals.(read somewhere, but expected since
PKS is supposedly tied to Cadence phsyical design
tools). Unless PC screws up big time, Synopsys has a
virtual hold on these accounts. This itself provides
enough momentum to sustain PC vs PKS in the
5. Realities of the newly forming market: The
reality is that Cadence is bent on giving their tools
free whever they are unable to compete on quality
(Ambit, new simulator price cuts etc.). This screws up
the entire EDA industry; many semiconductor companies
have higher EDA budgets then they are actually
spending. Wonder why? Because they feel that tools out
there are not adding the value they need. Cadence
strategy of price cutting ensures that the resources
needed for developing new tools which can add real value
and initiate extensive re-tooling in semi-companies
are not coming out at the pace at which they possibly
could. Cadence strategy is to screw the whole industry,
just look at the valuation of EDA vs the rest of the
market (even S&P 500 has higher P/E, I do not even want
to talk about the Nasdaq 100).