( ESNUG 384 Item 10 ) ------------------------------------------- [12/06/01]
Subject: Merger Proves How Piss Poor Synopsys Physical Design Tools Are
From: "Mike Dini" <mdini@dinigroup.com>
I don't get it at all. $830M cannot be described as 'road kill' and Cadence
still has many legal remedies available to block Avanti products from the
market. Whatever Avanti has that is good, I can develop from scratch in 18
months for a LOT less than $830M!
- Mike Dini
---- ---- ---- ---- ---- ---- ----
Hi John,
Anon please.
We're moving away from both Synopsys and Avanti. We have had very favorable
demos of both Ambit/PKS and Magma, but so far no complete tape-outs (just
re-runs of already-finished designs). IMHO, both SNPS and Avanti are
companies that were once industry leaders, but have lost their edge. It is
a sign of their weakness that they need to merge.
---- ---- ---- ---- ---- ---- ----
From: "Mark Garber" <mgarber@equipecom.com>
I own only the odd seat of H-Spice from either company.
On the whole, I view it as sign of just how unhealthy the industry of EDA
has become. Instead of innovating? Writing your own? Actually doing R&D?
EDA companies merge and acquire. This is a sign of continued stagnation
and lack of imagination and innovation.
In the long run, I believe it means there is just one less big EDA player to
acquire smaller innovators.
- Mark Garber
Equipecom
---- ---- ---- ---- ---- ---- ----
John,
Please keep me anonymous on this.
I feel that Synopsys would have been better off purchasing Magma even though
it would have been at a higher price tag. To date, Synopsys has not done
well in the physical realm:
- Chip Architect was a failure
- Route 66 was years late with questionable results
- Clock tree synthesis hasn't been there until recently
- PhysOpt fell short on all of my evaluations and Saturn produced better
results
It is obvious that Synopsys needed some help in this area and went shopping
to get it.
As for Apollo/Saturn, this tool seems to be stagnant in terms of growth and
capability as the quality of results are not keeping up with the results we
have seen from Magma. Magma is clearly a much more state of the art tool
and is producing results on designs that Saturn can not replicate.
The end result is that Synopsys bought a shot in the arm and are back in the
physical race, but are currently two or three lengths behind.
I'm curious as to how they will merge PhysOpt and Route66 with Apollo and
Saturn -- and what they will do with Chrysalis and Formality since both of
them are struggling to keep pace with Verplex.
In summary, the physical offering is back down to 2 real choices, the new
Synopsys/Avanti vs. Magma. Yes, there are a few other small players and (of
course Cadence) but none of these are very real or viable choices for
building a fast, reliable chip in 0.13um and smaller technology.
---- ---- ---- ---- ---- ---- ----
Hi John,
I consider this bad news.
I hoped that Synopsys will come up with some good new engines for clocktree
builder and a router that can handle the next generation design problems. In
my opinion Avanti has only technology that was state of the art 3 years ago
(even in Astro nothing is new). So a big chance lost. The Synopsys users
will have to struggle with the old limitations.
It shows that Synopsys is not a company for physical implementation. Even
with all the effort they have put into this they did not manage to get a
working hierarchy planer, a clocktree generator nor a router out to the
market. The future will show us if they learn from the Avanti people.
Synopsys has the tightest limitation on design size and some severe naming
problems. Avanti has some severe problems in maintaining the Verilog
hierarchy in the optimization steps. So the merger is for me resulting in
a product that is far behind Cadence and Magma current solutions.
Please keep me anonymous.
---- ---- ---- ---- ---- ---- ----
From: "Arvind Chopra" <arvind.chopra@philips.com>
I've always felt that Synopsys had better methodology, support, documenation,
and flow for their own developed tools compared to those they acquire. (A
case in point is the Pathmill/Timemill vs PhysOpt.) Instead I wish Synopsys
had continued work on their router and backend technology.
- Arvind Chopra
Philips Semiconductors
---- ---- ---- ---- ---- ---- ----
Anon please, John
Hmmm... my guess is that:
1. Route Compiler is routeless.
2. They needed to show growth, because their other products, except for
physical, are not showing any.
These factors forced the purchase.
---- ---- ---- ---- ---- ---- ----
Hello John,
It is the first time I am writing you. Could you please keep me anonymous.
Thanks!
I think it can be a good thing for Synopsys. It will give them a global
router that was really a lack for PhysOpt. We have been evaluating PhysOpt
for a year now and what we see is:
1) Avanti gives a better utilization of the chip than PhysOpt. For a given
utilization and chip frequency, PhysOpt gives a placed netlist that is
unroutable in Apollo. It was not the case by feeding directly the
netlist in Apollo. And this is important since during these cost
reduction times, we need to have the smallest die as possible!
2) We never achieved better results with the RTL to placed gates flow. We
are always using PhysOpt gates to place gates. And from a Synopsys AE,
most of the customer are using that flow. So I think that the actual
Synopsys promotion of the RTL to placed gates flow is only marketing.
3) Synopsys claims they can have a better estimation of the RC in 2001.08.
But what's the advantage of having a precide RC if the routing is only
a rough estimation? This is again a marketing feature.
4) Concerning the correlation between the frontend and the backend, we don't
see a lot of difference between using PhysOpt and DC. We are using
"in-house" wireload models because most of the wireloads provided by
foundries or libraries providers are much too optimistic. So again, we
don't see the advantage of bying such an expensive tool if we can achieve
out timing and constraints with DC.
As you see, we are not in favor of PhysOpt which, we think, is really pushed
by the marketing department of Synopsys. We hope that a lot of the issues we
actually have will be solved in the future thanks to their acquiring Avanti.
|
|
