| Cross Posted from DNT..
<<< ENUF::$1$DUA4:[NOTES$LIBRARY]DNT.NOTE;1 >>>
-< DIGITAL NETWORK TEAM NOTES FILE >-
================================================================================
Note 2088.1 AMD-Digital FDDI Chip Agr'mnt Annc'd 1 of 1
DELNI::CALLAHAN "Paul Callahan, NaCM, DTN 226 5171" 307 lines 10-OCT-1989 20:58
-< AMD-DEC FDDI Chipset Q&A >-
--------------------------------------------------------------------------------
AMD-DIGITAL FDDI CHIPSET AGREEMENT
<< QUESTION & ANSWER >>
~~~~~~~~~~~~~~~~~
CONFIDENTIAL 1-1-1
The Agreement:
Q. What is the duration of the agreement?
A. A minimum of 3 years, but it may be extended indefinitely.
Q. What does Digital get out of the agreement?
A. Digital is provided with additional sources for the production of
their FDDI chipset. In addition, Digital's agreement with AMD
will increase the availability of ANSI X3T9.5 FDDI compliant
chipsets which should drive down costs and subsequently lead to
the wide availability of FDDI performance in both high volume
desktop and bridge products.
Digital and AMD have a well established relationship based on
their joint development of the Ethernet chipset. This long
and mutually beneficial business relationship coupled with
AMD's leadership role in the development of FDDI chipsets were
the key factors in Digital's selection of AMD for this
partnership.
Q. What does AMD get out of the agreement?
A. The right to manufacture and market and develop designs
based on Digital's MAC and ELM ICs. AMD also gains a
systems partner with a tremendous amount of experience
in building and testing large networks, and one that is
committed to multi-vendor interoperability.
Q. How much money is involved in the agreement?
A. We don't comment on financial terms.
Q. Is this an exclusive agreement?
A. No.
Q. Does Digital get any rights to AMD's SUPERNET-3?
A. We don't discuss finacial terms of any agreements.
Q. Who is writing the software needed to support the Digital Chips?
A. AMD will provide software for all the FDDI circuits it
manufactures. Digital is providing portable SMT software to
AMD.
CONFIDENTIAL 2-2-2
AMD'S Position:
Q. What affect does the agreement with Digital have on AMD's plans
for SUPERNET-1 and SUPERNET-2?
A. AMD's commitment to SUPERNET-1 and SUPERNET-2 remains
unchanged.
Our priorities are:
1. Total support for SUPERNET-1 including executing cost
reductions, tracking compliance, and verifying interoperability
between users.
2. Development of SUPERNET-2, FORMAC-Plus chip for market
introduction in 1990. This product will be a highly
integrated, general purpose solution for FDDI equipment
vendors.
3. Manufacture of Digital's MAC and ELM chips.
Digital's MAC & ELM will be the building blocks for FDDI
developers who would prefer to customize or optimize their
designs around a modular chipset.
4. Developement of SUPERNET-3, which will be accelarated by the
use of Digital's modular MAC & ELM.
Q. Will AMD make Digital's MAC and ELM available to the
general marketplace or exclusively to Digital?
A. As part of the agreement, AMD will sell Digital's MAC and ELM
on the open market.
CONFIDENTIAL 3-3-3
The Circuits:
Q. Will Digital's MAC and ELM chips be interoperable with the
SUPERNET families?
A. Yes. Both companies strongly believe that interoperability
is critical to the acceptance of the FDDI standard. AMD's
technical expertise combined with Digital's experience in
building and testing large networks will ensure that
the two chip sets interoperate.
Q. Are Digital's MAC and ELM chips standard-compliant?
A. Both Digital and AMD are committed to supporting the X3T9.5
FDDI standard. If design elements are found that are not
standard-compliant, then, by agreement of both companies,
they will be modified to conform to the standard. [Optional
for Digital: Itt is our belief that the Digital chips do
conform to the ANSI standard.]
Q. Do Digital's MAC and ELM chips support Revision 5.1 of the SMT
ANSI draft standard?
A. Yes.
Q. What are the differences between the Digital chips and AMD's
SUPERNET?
A. Clearly one of the things that is not different is their
compliance to the ANSI X3T9.5 standard. SUPERNET-1, SUPERNET-2
and Digital's MAC and ELM are all compliant with the ANSI FDDI
standard.
Digital's MAC and ELM chips will be the modular building blocks
for FDDI developers preferring to add their own system
interface. AMD's SUPERNET-2 will provide a highly integrated,
general purpose solution.
In addition, AMD will include the Digital MAC and ELM chips in
future versions of their highly integrated SUPERNET chip
series.
Q. Will Digital use SUPERNET-3?
A. Digital is looking for multiple sources but can't comment yet.
CONFIDENTIAL 3-3-3 (continued)
The Circuits:
Q. When will SUPERNET-3 be available?
A. SUPERNET-3 is not yet defined. It will be brought to market
when FDDI is needed in extremely low-cost applications.
Q. What's wrong with SUPERNET that AMD isn't using it for the
core of a desktop solution?
A. AMD expects that SUPERNET-3 will incorporate key elements of
the SUPERNET-1 solution and Digital's MAC and ELM.
Q. So, what did AMD need Digital's circuits for?
A. AMD wants Digital as an FDDI chip customer. AMD and Digital
have had a long and mutually beneficial business relationship
in Ethernet.
Digital supplies not only circuits, but years of systems and
networking experience. This systems and networking
experience will be a valuable aid to AMD in ensuring a
standard-compliant and interoperable FDDI solutions.
In addition, the modular design of Digital's MAC & ELM will
accelerate the design process for SUPERNET-3.
Q. Will SUPERNET-3 interoperate with the Digital's MAC and ELM chips?
A. Yes.
CONFIDENTIAL 4-4-4
The Market/Standards:
Q. Isn't AMD confusing the market by offering so many FDDI
solutions?
A. No. In fact, AMD provides the widest range of silicon based
FDDI solutions. (refer to Positioning Statement)
Q. Isn't it more likely that the software for Digital's MAC and ELM
would support SUPERNET-3 more easily than the SUPERNET-1 software?
And isn't that a reason people will buy the Digital circuits in
preference to SUPERNET-2?
A. No. Digital's MAC and ELM represent core circuits. AMD will
surround these with its own systems and software interfaces.
Q. There is no indication that the market is large enough
to support so many solutions.
A. We expect the market will see explosive growth once the FDDI
standard is finalized next year. At that time it will be
important to have a portfolio of products that are
differentiated by application.
Q. Why did Digital design its own core when FDDI implementations
are expected to be available from several merchant IC vendors?
A. Digital's design is a natural outgrowth of their system
knowledge and extensive simulation work with FDDI technology.
In addition, Digital's experience with Ethernet has shown that
making chips widely available reduces costs and makes
technology affordable.
Q. Are the Digital MAC and ELM chips the same as those offered by
Motorola?
A. Yes. Digital is working with multiple merchant IC vendors to
provide core technology to ensure interoperability among FDDI
products.
CONFIDENTIAL 5-5-5
Q. Why isn't Digital listed as a member of the Advanced Networking
Group if it is committed to the standard and to interoperability?
A. Ask Digital.
Q. How will the agreement with Digital affect AMD's position in the
ANSI X3T9.5 committee meetings?
A. Each party is free to vote independently from the other. The
agreement does not require any collaboration on open standards
issues.
Q. Aren't these chip sets coming out on top of each other pretty
fast??
A. Not really. The market demands new developments much more
quickly than it ever has--witness the speed with which the
80486 was introduced after the 80386 was introduced. We
believe it is prudent to provide a solid foundation for an
upgrade path and to make details of that upgrade path
available to our customers.
Q. Will there be an alternate source for the SUPERNET chips?
A. Nothing in our agreement with Digital prevents us from adding
a second source to the SUPERNET family.
Q. So, AMD will be able to second source SUPERNET-3?
A. Yes.
Q. Will there be a second source for any SUPERNET chips?
A. No second source partner has been identified at this time.
AMD may develop a second source in the future.
Q. Will the Digital chips be second-sourced?
Nothing in our agreement precludes this. AMD's agreement with
Digital is non-exclusive. Digital may be working with other
vendors. Ask Digital.
Misc:
Q. When will Digital have an FDDI product and what will it be?
A. Ask Digital.
|
| Can anyone help us with benefits of the modular DEC design, in
competition with the AMD Supernet-1 chips? This chip set is being
offered in many competitors' products now (e.g. UB, cisco) and I'd like
to give customers as many reasons as possible to wait for the DEC and
AMD Supernet-3 products. If the chip sets are all "X.3T9.5 compliant",
why should a customer wait for DEC's to be announced?
|
| Are these other vendors willing to make their devices meet the standard, once
the standard is finally approved, especially in relation to SMT, at no
additional cost? Will these vendors replace or correct components that
operate improperly from the standpoint of data integrity and chip design
problems?
If you look at Ethernet, it is clear that DEC is commited to delivering a
fully functional inspite of design problems in virtually every LANCE
implementation on the market. DEC has spent significant amounts of money
replacing adapters that we sold that later proved to have design problems
in these commodity parts. Do these other vendors have similar track records?
(I'm positive the answer is NO.)
|
| AMD's chipset is the only currently available `FDDI' chipset. It is currently
used in early implementations of FDDI bridges, routers and system adapters.
It is a modular, five chip implementation, consisting of Buffer Controller,
Data Path Controller, FORMAC (fiber optic ring MAC) chip, FDDI Encoder/
Decoder and a Data Seperator. Functionally, it's probably equivalent to
Digital's initial Lab implementation of FDDI in a six-chip model, although
performance and scale evaluations between the two were never made.
Digital's first implementation was carried out in six chips, because at the
time, the standard components (PMD, PHY, MAC and SMT) were all very `soft'.
Engineering wanted to be able to modify designs in a controlled and staged
process. The result was development of a second-generation design which
is the basis for all FDDI testing currently taking place in Digital Labs.
It remains to be seen what will happen in the marketplace when new silicon
implementations of FDDI become available. Will the AMD SuperNet I still
be competitive? Are there any fundamental problems with such things as
station support (that is, numbers of devices operating simultaneously on
a ring), scalability, ability of this implementation to properly handle
timing on the ring, etc.? These questions remain unanswered to date, but note
that AMD is hard at work on SuperNet II. "Interoperability tests" seem to show
that AMD-based devices in a limited test environment can be good citizens
on a small ring and can even interoperate at a low level. Is this news?
Would it not be more of a surprise if AMD devices didn't work together?
Vendors anxious to put FDDI or FDDI-like products to market have had to build
devices around AMD silicon. Digital, on the other hand, has sought to
design and build it's own chipset. This was an intentional effort that
was based on the need to have a robust, high-performance and reliable chipset
implementation. The trade-offs then were: finished (non-experimental) silicon
vs. quick time-to-market.
License agreements should say something to potential FDDI customers. Why are
Motorola and AMD building to the Digital design? It's because Digital has made
a tremendous investment in the base-level FDDI technology and knows what will
be necessary (at the silicon, device, system and architectural levels) to make
FDDI work in both a workgroup and a backbone network environment. We should
be cautious not to bash AMD in any way. They are are partner under the
license agreement, and we would do well to respect that partnership. But
we should caution our customers about commiting to FDDI too early in this
game. Spending money on FDDI should result in network investment, not write-
offs on prototypical equipment that becomes obsolete in the near term.
Is the X3T9 standard complete? No. Will all FDDI silicon implementations
play together? Who knows? The real answer will not be known until
independent interoperability testing can be conducted and verified by
unbiased parties (standards bodies, rather than silicon vendors, for example).
Customers should wait because there is considerable uncertainty about the
longevity of these early device implementations. FDDI devices are costly now,
and may remain so for many months to come. Before spending considerable sums
on early FDDI implementations, customers would probably be better served by
careful planning for the higher speed technology. But they should wait
to see what the major `system vendors' (companies that design and build sys-
tems, network devices, small and large-scale networks, companies that develop
operating system and applications software, companies that provide complete
services - in other words, companies that are commited to doing it ALL) do
before they make a decision on procurement of FDDI.
|