Subject: Yikes! CoWare/VAST gives SNPS a near monopoly in SystemC models

On 02/09/10, Jay Vleeschhouwer of Ticonderoga Securities issued a comment:

>  Synopsys has announced two small acquisitions (VaST Systems, CoWare)
>  to build its position in electronic in system-level design and
>  "virtual prototyping".  The combined transaction values, we believe,
>  represents only a small portion of the company's nearly $1.2 billion
>  in cash.  VaST will be immaterial to results while CoWare, a roughly
>  $30 million company, may have some impact.
>
>  We maintain our Neutral rating.
>
>      - Jay Vleeschhouwer of Ticonderoga Securities


From: Simon Davidmann <simond=user domain=imperas not mom>

Hi, John,

Thought I'd comment on recent happenings in the virtual platform business,
but first I figure I must explain to you what VP's are because you always
look confused about it whenever you see me at DAC.

In a nutshell, virtual platforms are big C models of a complete system
or of a chip.  In the software industry most virtual platforms are built
with just C using standard C compilers and readily available SW debuggers.

For hardware designers, the EDA industry has gone down a different route
with Synopsys and CoWare promoting a C++ class library called "SystemC"
with an additional compatibility layer called TLM 2.0 added on as the glue
to interconnect these C based components.  Ergo, models tend to be in C
and the glue to interconnect them is either C or in SystemC TLM 2.0.

Like simulation for verification of your chips in RTL; "Virtual Platforms"
are just C models of your chip that you can use to verify the operation of
your chip with software running on the embedded processors in your chip.

More technically, a virtual platform is a collection of C models of the
functionality of your entire chip all brought together in a C simulation
environment with tools to help with your verification - the focus tends to
be for software bring up and overall software/hardware verification.  All
the C models tend to be optimized for runtime speed.

There are two very important aspects to a virtual platform; availability
of high speed models, and the ease of putting models together.

Virtio/Vast/CoWare:

In the past, most models came from independent companies.  This is what
Virtio, VaST, and CoWare provided - now all part of Synopsys - the models
being developed on a contract basis.

Interestingly, this was also Virtutech's business model - service - but
their focus was on full chips for semiconductor companies to be used by
systems companies - and Virtutech are now part of Intel.

Where does this leave the Virtual Platform market?  It's now one dominant
player, Synopsys, with a lot of expensive staff, three different approaches
to building SystemC models for customers yet less competitive choice for
the customer.

This now leaves two independent companies in the Virtual Platform space:

    - Carbon with its cycle accurate model building solutions,

and

    - Imperas with our Open Virtual Platform that provides ultra
      fast instruction accurate models. 

The interesting differentiator between these two and Synopsys is that Carbon
and Imperas provide solutions to their customers that enable their customers
to build the models and platforms themselves, whereas the Synopsys approach
is a lock-in-the-customer service solution.  (Imperas also provides vendor
verified open source processor models through OVP.)

No question this week, with Synopsys buying both VasT and CoWare, it draws
a clear line for users who want to build their own simulation platforms for
high speed software verification can now only choose Imperas or Carbon --
or they can get locked into the Synopsys empire.

I'm reminded of 20 years ago when I was in Cadence.  Back then Joe Costello
was building a service business to help his customers use complex EDA tools
called Spectrum Services.

Here we are now in 2010 with Synopsys arguably now the world's largest EDA
tool vendor putting significant resources into building a service business
around the new Virtual Platform market.

Now there's less choice.  Synopsys is going to be expensive with services.
Didn't we do this before?

    - Simon Davidmann
      Imperas Software, Ltd.                     Oxford, UK

         ----    ----    ----    ----    ----    ----   ----

CoWare owns $20M SPW market.  

Synopsys COSSAP only took $10M of same market.

SNPS buys CoWare for $20M, gives SNPS $30M majority in Virtual Platform
plus Sony, Freescale, NXP and Toshiba which are SPW customers that
Synopsys didn't have before.

This plus VAST buyout gives SNPS a near monopoly in SystemC models.

    - [ An Anon DeepChip Reader ]

         ----    ----    ----    ----    ----    ----   ----

VaST was old dying technology from 1996.  With around 50 people, they were
never profitible but ran at close to break even.  They couldn't raise any
more VC funding.  Labat shopped VaST around for 5 months.  At best, they
got $3M or $4M from Aart to give to the investors and $0 for the employees.


    - [ Another Anon DeepChip Reader ]

         ----    ----    ----    ----    ----    ----   ----

This VaST acquisition was negotiated directly between the VaST investors
and Synopsys.  VaST employees found out about the deal after it was done
from the Synopsys press release.


    - [ Even Another Anon DeepChip Reader ]

         ----    ----    ----    ----    ----    ----   ----

From: Bill Neifert <bill=user domain=carbondesignsystems not mom>

Hi John,

Synopsys is certainly shaking up the virtual platform (or what you like to
call the SystemC models) marketplace.  From my perspective though I would
guess that these two purchases were made for very different reasons.

VaST had a big concentration in automotive accounts and in Japan.  When the
financial meltdown took place it put a tremendous financial strain on the
automotive market and greatly diminished Japanese sales.  I can't imagine
that this meant good things for VaST from a sales and finance perspective.

It appears to me that Synopsys was likely able to buy VaST on the cheap,
giving SNPS a huge amount of system IP that they weren't really targeting
with their current Synopsys Innovator product line.

When you look at what Synopsys did when they acquired Virtio, first thing
they did was spin out models and create its DesignWare System Level Library.
They also ported them over to be functionally equivalent to the RTL models
in its DesignWare IP Library.  This gave SNPS a chance to sell system models
that could run in ANY virtual environment and then back that up with RTL
models that were compatible.  I imagine they'll do the same thing now with
the VaST IP library.

I view the VaST sale as more of Synopsys trying to establish more dominance
in the IP space than about what it does for them in system modeling.

The CoWare sale is a bit more of an enigma to me.

From the outside it would seem that CoWare has almost complete overlap with
tools that are already at Synopsys.  SPW would seem to compete with System
Studio. CoWare Platform Architect and Synopsys Innovator are in direct
competition with each other.  Innovator is typically used at the higher
level of abstraction and Platform Architect used at the lower level.

Maybe CoWare had a few bad quarters and needed to get purchased to not go
under?  Or perhaps the exact opposite was true and CoWare had a few good
quarters and decided to strike while the iron was hot?  I know that given
the big expansion that Carbon observed in the virtual platform market last
year it seems far more likely that CoWare was weathering the economic storm.

There are three more assets CoWare possesses that Synopsys doesn't:

   - a large virtual platform services team
   - their Processor Designer tool, and
   - a large number of Japanese virtual platform customers

I find it unlikely that Synopsys is going to go through all of this turmoil
to acquire a services team, and Synopsys already has a substantial presence
in Japan so it's hard to imagine that they'd buy CoWare just for customers.
That leaves CoWare's Processor Designer.  Processor Designer seems to fit a
niche of designs that don't fit well into typical behavioral synthesis.

This could be a play to give SNPS even MORE leverage in the IP space as it
now gives them the capability to deliver SystemC models plus implementation
RTL for a whole class of designs where customers might otherwise buy a
custom processor.

Mentor has many offerings but has never made a major play, probably due to
the lack of IP.  Now Synopsys is arguably out in front with respect to the
big three and from what I've seen with the customers I've visited, Cadence
seems to be sitting this race out.

Fewer choices in the marketplace is rarely a good thing for customers but
at the end of the day the models that go into the virtual platforms are far
more important than the platforms themselves.  The TLM-2.0 standard is
making those models portable across any virtual platform.

If the Synopsys acquisitions mean that it now has sufficient market presence
to thumb its nose at TLM-2.0 and use its own proprietary interfaces then
this is truly a sad day for system modeling and the customers that use it.

If however, Synopsys keeps following the interoperability path then I view
this more as the end of the "my platform is better than yours" wars and the
start of a new era of competition where it's more about the value of the
models that can go into those platforms.  Synopsys has positioned itself
well for this new IP-dominated fight as well, but that battle seems to be
just getting started.

    - Bill Neifert
      Carbon Design Systems, Inc.                Acton, MA

         ----    ----    ----    ----    ----    ----   ----

From: Brett Cline <bcline=user domain=forteds not mom>

Hi John,

The recent Synopsys acquisitions of CoWare and VaST are interesting.  While
consolidation in the virtual platform segment is probably needed there is a
hidden scary part of this story that is critical to understand; it's not
about the products, it is about the models.  Each of these companies has
significant investments in C and SystemC models and now (with their past
Virtio acquisition) Synopsys will have a lock on that space!

Synopsys was one of the cofounders of SystemC way back in 1999 and it openly
supported SystemC until 2002 when it went on hiatus to build System Verilog.
As the official "SystemC poster boy", I'm flattered Synopsys is back in the
SystemC game, but we do need to keep an eye on this model monopoly thing.

And I have to laugh at the Mentor Catapult C guys.  Even they have finally
realized that they need SystemC for synthesis despite years of telling us
all why SystemC wasn't needed!  ...  So now comes the ultimate backpedal
marketing shuffle about why they weren't wrong for the last 5 years.  I
wonder how Thomas Bollaert is feeling about his comments at DAC 2009 that
"SystemC will eventually go away" -- OOPS.  Welcome to the team.

    - Brett Cline
      Forte Design Systems                       Acton, MA