( DAC 10 Item 7 ) ---------------------------------------------- [ 09/10/10 ]
Subject: EDA vendors react to the EDA360 paper before DAC
AN OPPOSING VIEW: I want to thank the Cadence rivals who took the time to
dissect the original 32 page EDA360 document and send in their take on it.
I also tip my hat to Bruggeman for having the guts to do this an open
discussion. Call me old school, but when I read a news story I like to
see BOTH the proponent's view PLUS the opponent's view so I can get the
whole picture.
"Could you please post this EDA360 vision paper in the DeepChip.com
downloads section and publish the responses you get about it?"
---- ---- ---- ---- ---- ---- ----
"We applaud Cadence's EDA360 vision for growing the entire EDA industry
'out of EDA' ($5B) and into Embedded Systems ($25B). It's not only a
matter of sheer survival, but an opportunity for our entire ecosystem
to deliver greater value to customers via complete application-ready,
programmable hardware/software platforms for SoC design.
EDA360 implies an FPGA mindset. Indeed, FPGA's are the ideal platform
for what Cadence calls System Realization, SoC Realization and Silicon
Realization. The trend toward programmability - what Xilinx has coined
the 'programmable imperative' - is accelerating while ASIC/SoC design
starts continue their steep decline, due to spiraling costs and
complexity at 32nm and below.
Cadence's EDA360 manifesto lines up well with our own Targeted Design
Platform strategy for delivering programmable solutions that help system
developers meet the demands of today's embedded systems - and the
company's timing couldn't be more apt. On the same day Cadence made
its announcement at the ESC Silicon Valley, Xilinx rolled out a new
ARM-based Extensible Processing Platform that will deliver unparalleled
integration, high-speed signaling performance, flexibility and
extensibility for embedded systems.
Like Cadence, Xilinx is committed to solving the productivity gap and
the profitability gap by working hand-in-hand with a broader set of
development communities to appeal to software-centric designers, beyond
our traditional customer base."
- Moshe Gavrielov of Xilinx
---- ---- ---- ---- ---- ---- ----
"My opinion is that this is:
1) A pretty gutsy move.
2) Correctly identifies some weaknesses in the current business
model for EDA.
3) Is a pile of mush with tons of squishy buzzwords.
The type of concept-to-system approach proposed in this document is
already accomplished by service companies such as Spectrum DSI. They
charge roughly 200K to bring a fully realized concept to market. The
customer brings them a napkin, and they build it, usually within 6
months. There is simply not enough profit at these levels to support an
industry infrastructure.
Mr. Bruggeman's clean and well shaped notion of the design process is
unrealistic. The idea that you would have design intent and constraints
at the the beginning of the project does not reflect reality. The design
process is, to steal from Thomas Hobbes, "nasty, brutish, and short".
There are many things that cause problems, and most are not related to
the software or the system. Significant features change daily, the
dream of the designer works great at system level or RTL, but then falls
apart when it hits the physics of implementation, the end customer
doesn't want to pay as much, re-specs the design, and so on.
Market forces have been driving the EDA industry to tighter integration,
higher performance, and lower cost. EDA vendors sell software today
that has many times the capability, for half what they charged 10 years
ago. I expect that trend to continue, as vendors add integration to
tools both up and down the design process.
I don't expect Mr. Bruggeman's ideas to translate into dollars for
anyone."
- [ An Anon EDA Vendor ]
---- ---- ---- ---- ---- ---- ----
"Read through all 32 pages of EDA360 vision paper. Seems like a summary
of well-known 'wish lists' from the design and product development
communities. This kind of "open source collaborative ecosystem"
discussed, as a concept, is nothing new. Yet, no such ecosystem has
ever developed. In fact, if anything, we've moved towards 'silos'
more than we've moved towards collaboration.
What will Cadence do to enable this vision? Why would other vendors
(EDA and non-EDA) join such a movement? The vision paper does not spell
out how the tool vendors will gain from this -- except making a tall
claim that EDA will suddenly go from being a $5 B stagnating industry
to a $25 B high-growth industry? Will EDA360 actually enable the tool
vendors to get a share of revenue that their tools enable? Honestly,
I doubt it, and hence, don't see where this sudden burst of growth
would come from.
I propose the following test case, and look forward to John Bruggeman's
response:
"I want to build a system capable of running, say 90% of the apps
for iPhone, without violating any patents, and within 12 months
from start to finish. I'm willing to license any IP that needs
to be licensed and is available in the market for a known price."
Can he come up with a concrete project plan employing the EDA360 vision?
The plan should clearly identify the choices for the HW and SW IP pieces,
the integration plan and the verification plan -- including all the
appropriate tradeoffs that I should consider. The project plan can
borrow from or use Android in some manner, but should not lead to a
virtual iPhone on top of Android -- it should lead to a new, more
optimized hardware and software system than the iPhone itself, since a
very large number of applications are already defined.
Will wait to see if John can defend in practice what he preaches, or if
the EDA360 vision paper just an exercise in marketing hot air."
- [ An Anon EDA Vendor ]
---- ---- ---- ---- ---- ---- ----
"I must admit I really enjoyed reading the EDA360 document from Cadence.
As you know, I was a VP Sales at Cadence and one of the reasons I left
was that Cadence had no vision and was only focused on increasing the
current quarter revenues, no matter what the cost. It is great to see
that there is some deeper thinking going on now.
I am confident that eventually the industry will have no choice but to
go in this direction, where the SW applications define the HW and the
HW is designed to serve the SW. It makes a lot of sense and I believe
it is the only way to deal with the new challenges. I have been working
in the past 18 months with IPsP, a company promoting exactly this vision.
I believe they could easily become part of the backbone of EDA360.
Last summer I sent Lip-Bu Tan 2 emails on this topic and was ignored.
Maybe this document is the answer..."
- Coby Hanoch of EDAcon Partners, Ltd.
---- ---- ---- ---- ---- ---- ----
"Please keep me anonymous; I'm a veteran EDA R&D developer.
There is nothing new here, and there is very little money. Furthermore,
Cadence's rivals are all trying to do this stuff already, as is Cadence.
The adjacent market segments he talks about in his paper are either
already being targeted by major EDA firms (yes, ESL is coming really
soon now), or are composed of people in the habit of paying zero or
very little for tools (embedded software developers). The big players
are all already trying variants of this guy's proposal, and it's
not paying off. Higher levels of abstraction! IP reuse! Platforms!
Sound familiar?
So, can you address a much larger market by targeting people who are
merely integrators? Well, there's certainly a market there, but it isn't
huge, or ESL would be worth more money than place and route. Hint: it
isn't. Certainly it's worth producing "System Realization" tools, and
there's some market there. But we're talking tens of millions, not tens
of billions. It's the people Cadence used to target with SPW and the
people who bought CoWare. You can't making them pay big for integrating
their software development with your platform, because they'll be pulling
in a Linux kernel and the GNU tools, or using Wind River's tools.
Entering a segment where you're competing against free or low-cost tools
isn't the easiest way to focus on profitability."
- [ An Anon EDA Vendor ]
---- ---- ---- ---- ---- ---- ----
"Keep me anon.
I think EDA360 has more to do with what's going on with Cadence than it
does with the industry.
10 years ago in 1999, Cadence had $1.1B in revenues, Synopsys had $806M
and Mentor had $511B. Last year, Synopsys had $1.36B, Cadence had
$852M, and Mentor had $802M. Even given the two rough recessions in the
last decade, Synopsys and Mentor have both grown 60% to 70%. Cadence
has blown up yet again and is adrift.
If you strip out the $133M of low margin services Cadence had last year
and just focus on products they were number 3 in EDA last year. Mentor
had $30M in services, making them number 2 in product sales last year,
ahead of Cadence.
And Cadence has been laying people off like mad. On their fourth
quarter 2008 earnings call they said before the lay offs went into
effect, they had 4900 employees. Since then, they've fired nearly a
thousand or about 20% of their whole workforce.
I think EDA360 is more an attempt to rally Cadence employees than it is
a call to the industry.
Just look at the EDA360 white paper. Cadence talks about the importance
of IP. Synopsys has Designware. Cadence talks about embedded software.
Mentor has that. They talk about the need to jointly develop the
hardware and the software -- which you can do on emulators that Cadence
and Mentor both sell. And so on. The paper is mostly vacuous platitudes
about system design filled with things that are already being done or
worked on.
Looking at more than IC design may be an epiphany for Cadence, but
systems customers have been doing this a long, long time.
They end the paper with a bunch of silly obvious stuff like: "Use
single-pass design" and "Intelligently partition large digital blocks"
Really? I guess the plan to fail a couple of times and stupidly
partition won't work, huh?
As a vision to rally the troops, I can see why Cadence would do this.
Their troops must be pretty demoralized after all the blood in the
halls. But the only substance here is already being done -- mostly by
Cadence competitors."
- [ An Anon EDA Vendor ]
---- ---- ---- ---- ---- ---- ----
"I'd like to be anonymous.
My feedback about the EDA360 Vision is from the vendor side of my career,
although I was a user for several years so I'll also include some of that
perspective as well.
I don't get it. It appears that some people in Cadence wrapped an
elaborate 'vision' around concepts that people in the EDA industry have
known and talked about for 2+ decades ? hoping to quickly fill a hole
in its product mix.
Executing early software on virtual hardware has been a primary goal for
the functional verification vendors since well before Verilog came along
in the late 80's. The weird part is that much of Cadence's Palladium
success over the years is tied to being able to do just that. And before
that, how many generations of emulators/accelerators have we seen from
the likes of Zycad, Ikos, PiE & Quickturn and many others? Cadence's
Palladium seems to have captured more than 50% of that space.
On the surface, it's not obvious what Cadence stands to gain from
claiming that a sudden emergency exists. Productivity and profit have
always been important to users, although EDA vendors don't get the type
of access to the senior system execs that they'd like to so they don't
get to talk much about profit. I'm sure that profit is not something
new to the customer execs. Getting EDA tool users, CAD managers and
engineering managers to discuss their profits with EDA vendors won't
happen. EDA vendors can't change the scope of their customers' jobs
that much.
That brings us to the 'Open IP Integration Platform'. Last I looked
Synopsys was #2 or #3 in semiconductor IP and Cadence hadn't yet cracked
the Top 20 ? that's a big gap. However, Cadence does have most of what
an 'IP Integration Platform' would include. What more could it to do
open up its tools? OpenAccess, LEF/DEF and Verilog are mostly open and
cover a lot of the design space. Synopsys still controls synthesis &
timing and Mentor practically owns physical verification. Could this be
a play to get everyone to drop the remaining walls? Nope? Something
tells me that Cadence is relieved that the OpenAccess/PDK fiasco has
died down a bit. The last thing they want is to restart the 'Open PDK'
debate. IPL & TSMC might crack that open over time. And, they didn't
bring in John Bruggeman from WindRiver just to open more of their stuff.
I think the 'vision' is more about the 'Open IP' and far less about the
'Integration Platform' part. Let's face it, Aart has been steadily
building an IP portfolio starting with DesignWare over the last 10-15+
years. Since then, Cadence has had 5 CEO's with at least as many growth
strategies. When I had close friends there ~6 yrs ago, the sentiment
was that getting into the IP business was considered competing with
customers unless it was used in a narrowly defined services engagement.
That seems to have changed over the past year or two -- Cadence bought
a few verification IP companies, Synopsys entered the custom IC tools
business and Synopsys bought the ChipIdea custom IP group from MIPS.
It's no secret why IP is so important to EDA companies. It's tuned to
work extremely well with a particular vendors' tools, it's a great money
maker all by itself, and it helps EDA vendors lock-in customers. It
might even eventually help EDA vendors have that coveted 'profit'
discussion with customer executives. The independent IP companies tend
to tune it to the market leading tool in each space -- Synopsys's DC and
PrimeTime, Mentor's Calibre, etc. It's also a major threat to any EDA
vendor that doesn't either have enough of it, or it doesn't favorably
exploit the unique capabilities of its tools.
So what's an EDA vendor like Cadence to do? Go out and buy a bunch of
IP companies, focusing a lot on custom IP to protect Virtuoso, build
their own IP or appeal to the masses to demand that all IP be open.
Since there aren't many IP companies left to buy, Cadence doesn't have
enough capital left with which to buy much & building their own could
take a decade or more of consistently executing a committed plan while
deciding on many other things not to do, there's only one thing left
for a 'rock star' marketing guy to do.
I don't know that much about the software tools market, but I don't
think that EDA will take this bait."
- [ An Anon EDA Vendor ]
---- ---- ---- ---- ---- ---- ----
"I'll grade EDA360 the way my Marketing prof would in biz school.
Message: "F" - It's unclear what your message is. Need elevator
pitch, not 32 pages.
Delivery: "A" - Great idea to use Cooley's forum to get your
message out like that. Never been done before.
Impact: "C" - Most EDA people know the term "EDA360", but have
no idea what it is. Fantastic penetration.
Overall EDA360 is a "C-" to "D+", which is not bad for a first attempt
by a new Cadence CMO new to EDA. Have a clear message next time."
- [ An Anon EDA Vendor ]
---- ---- ---- ---- ---- ---- ----
"It's good to know that Cadence mgmnt wants to do something different,
that they are thinking hard how to change flatten EDA trend. Only a
much bigger pie can make us better, this "us" includes Cadence,
Synopsys, Mentor, etc.
Apart from the argument back and forth, apart from the legal blabla,
I hope everyone in this industry, especially the C*Os can think hard
how to make EDA a better industry and how to make EDA help IC more."
- [ An Anon EDA Vendor ]
Index
Next->Item
|
|
