( ESNUG 495 Item 1 ) -------------------------------------------- [11/11/11]
Subject: Designer seriously questions Aart pushing an All-SNPS design flow
> Aart said it is no longer good enough to sum up unique values for each
> stage of a best-in-class EDA tool flow.
>
> Sum != Product (by product he means if any one of the inputs
> is a 0 then the output is a 0).
>
> Aart pitched for why a total Synopsys toolset is more valuable than
> best-in-class EDA tools. Typical SNPS sales talk.
>
> - from http://www.deepchip.com/items/0494-01.html
From: [ The Horse With No Name ]
Hi, John,
Keep me anon for fear of retribution from Synopsys. Their sales guys called
my company after my last letter questioning a SNPS business practice.
Let's take Synopsys' single tool implementation flow.
1. ESL: Cossap, Cocentric System Studio, VP Designer, from Cadis and
CoWare and Virtio acquisitions.
2. Fuctional Verification: Vera, & System Verilog: Vera acquired from
Sun Microsystems and System Verilog from Co-Design Automation, Inc.
3. Verilog Simulation: acquired from ViewLogic which acquired Chronologic.
4. Logic Synthesis: Design Compiler (the original Synopsys product)
5. Formal Verification: Formality (along with technology from Chrysalis
which was acquired by Avanti)
6. Static Timing: Primetime, toss in acquisitions from Motive and
Extreme-DA (Goldtime).
7. Physical Design: IC Compiler. IC Compiler is a derivative of the
Avanti acquisition, which itself was the merger of two small
companys (ArcSys & ISS, with some acqusitions later). How many
ICC users currently use Azuro along with ICC?
8. Floorplanning and Hierarchical design: Similar to above -- an Avanti
acquisition, along with some technology from Monterey Design which
acquired its floorplanner from Aristo.
9. RC Extraction: Similar to above, STAR-RCXT came from Avanti through
the ISS merger. They have also recently acquired extraction tech
from Extreme here.
10. Physical Verification: Hercules, again from Avanti through ISS
11. IR Analysis: PrimeRail is EPIC and Apache and other small
acquisition, plus some organic efforts.
12. Spice: 1/2 from Avanti and 1/2 from Nassda.
13. TCAD/Yield: Numerical Tech, Avanti, ISE AG, HPL, Sigma-C
14. FPGA & Prototyping: Synplicity & ProDesign Electronic GmbH
15. IP: 1/3 DesignWare and 2/3 Virage, Leda, Mosaid, nSys
There's probably some hidden 1-5 man acquisitions I don't have listed above
as well that never got announced.
What were to happen if way back in the early 1990's, the industry went with
a single vendor flow, and relied on Synopsys solely for innovation? What
would Synopsys' product offerings look like today in the IC implementation
space? Verilog simulation? Even in their core area of logic synthesis,
where would Design Compiler be today without meaningful competition from
Ambit, Get2Chip and now Oasys?
Would we all still be stuck on version 3.4b of dc_shell like we were until
Ambit came along?
Let's look at the above flow without Aart's many acquisitions:
1. ESL: no offering
2. Fuctional Verification: no offering
3. Verilog Simulation: no offering, prob still pushing VHDL simulation!
4. Logic Synthesis: dc_shell / PhysOpt
5. Formal Verification: Formality
6. Static Timing: Primetime. (You think they would have multi-cpu yet
though?)
7. Physical Design: probably physOpt would have CTS and a router by now
8. Floorplanning and Hierarchical design: no offering
9. RC Extraction: no offering
10. Physical Verification: no offering
11. IR Analysis: no offering
12. Spice: no offering
13. TCAD/Yield: no offering
14. FPGA & Prototyping: no offering
15. IP: a much weaker DesignWare offering
To prove my point, check out the SNPS Wikipedia page:
http://en.wikipedia.org/wiki/Synopsys
It lists the 29 major acquisitions that made Synopsys what it is today:
1994: acquired Cadis
1997: acquired EPIC Design Technology Inc.
1997: acquired Viewlogic Systems, Inc.
2002: merger with Avanti Corporation (ArcSys & ISS)
acquired Co-Design Automation, Inc.
acquired inSilicon Inc.
2003: acquired Numerical Technologies, Inc.
2004: acquired Accelerant Networks
acquired Analog Design Automation, Inc.
acquired Monterey Design Systems, Inc.
acquired Cascade Semiconductor Solutions Inc.
acquired Integrated Systems Engineering AG (ISE)
acquired LEDA Design, Inc.
acquired Nassda Corp.
2005: acquired HPL Technologies
2006, acquired Virtio
acquired Sigma-C
2007: acquired ArchPro Design Automation Inc.
acquired IP assets from MOSAID Technologies
acquired Sandwork Design.
2008: acquired Synplicity
acquired ChipIT Business Unit from ProDesign
2009: acquired Chipidea from MIPS Technologies
2010: acquires VaST Systems Technology Corp.
acquires CoWare Inc.
acquires Virage Logic Corporation
acquires Optical Research Associates
2011: acquires nSys Design Systems
acquires Extreme DA
This does not include everything SNPS has acquired, nor does it include all
companies which themselves had acquisitions, etc.
To be fair to the other 2 big guys have been fairly acquisition prone as
well. (I couldn't find enough info on Mentor to list it.)
CADENCE
The Cadence flow, is nearly 100% a 'design by acquisition' flow on the
implementation side. (Cadence itself being the merger of SDS and ECAD.)
Without acquisitions we'd still be at a Synergy/Cell3/Dracula/Verilog-XL.
They've had meaningful acquisitions such as CCT, CadMOS, Simplex, SPC,
Plato, Get2Chip, Verplex, Denali, Clearshape, Quickturn, Azuro, etc.
Many of these CDNS acquisitions were in what I would consider mainstream
technology and tools that were small startups getting traction as point
tools... tools that if everyone believed in single vendor flows, would
not exist.
MAGMA
Magma, I'll give them credit -- in the IC implementation space, they are
pretty much organic growth guys, and rely much less on acquisitions
(Although their cash cow FineSim was an acquisition as was Quartz DRC/LVS.
I'm not sure if Tekton/QCP are the results of small acquisitions or they
hired the team organically -- their R&D lead is the former Pathmill lead.)
Even if I'm a strong believer in a single vendor flow, is it good for our
industry? Where would our EDA tools be today if our user strategy relied on
only Synopsys, Cadence, Magma or Mentor as our sole source of innovation
and product offerings?
Much of the in-house innovation that did happen at SNPS/MENT/CDNS/LAVA came
from the threat of being displaced by an EDA start-up's superior point tool.
How can Aart argue against designers using best-in-class point tools at
every stage in their design flow when Synopsys itself is 90% acquisitions
of the best-in-class point tools of their day?
The only time EDA tools seriously improve is when a small start-up that's
completely free of empire bureaucracy and politics hyperfocuses on making
a much better point tool for a known design problem.
Aart should recognize and honor that, instead of scold and sell against it.
- [ The Horse With No Name ]
Join
Index
Next->Item
|
|
