Frontier Journal
Exclusive Frontier Coverage on System Design
Vol. 2 No. 11 Nov 2005
GUEST EDITORIAL ¨C SoC Design success: Winning with Standards - OCP-IP
GUEST EDITORIAL SoC Design success: Winning with Standards Ian
Mackintosh President, OCP-IP It is inevitable that
the SoC Design industry will address increasingly brutal time-to-market cycles
and productivity demands by breaking away from the proprietary clutches of
individual suppliers. And, the road to this maturity and success will be paved
with open standards. Today,
most corporations grow by the pursuit of a product-driven strategy that begins
with the development of a Native product (see Figure 1). As competition
increases there is an inevitable progression to add ¡°Value¡± to this offering
which thins out competitive ranks and fuels the survivors¡¯ success and usually,
growth. Inevitably, consolidation continues in the segment, the product
offerings get more and more comprehensive and becoming what Geoffrey Moore has
coined as ¡°Whole¡± products. The path progression is generally predictable and
indeed brings much success to the short list of surviving companies. This
growth model holds well for most industries. The speed with which these
offerings evolve can serve early markets well, enabling many larger users to
abandon more poorly developed and high-cost internal offerings in their favor.
Other (often smaller) users are empowered by these offerings to enter markets
they could otherwise not address. But what comes next and is there a penalty
associated with this approach (See Figure 2)? If we consider the long-term effect on market growth, it is
easy to conclude that although this approach has real value to users (and certainly
fuels the corporate growth of providers), we are ultimately headed for a wall. The
reality is that the provision of these ¡°whole products¡± not done with the goal
of providing the best possible solution, but of capturing the largest
market-share with the minimum resources possible. Inevitably, these ¡®whole
products¡¯ take on such vast proportions that they are comprised not by ¡®best of
breed¡¯ constituent parts, but by what can readily be provided to minimally
serve the customers needs. The actual content of these whole product offerings
inevitably spread into new technical territories for their providers, areas
where their expertise can be weak and often second-rate. The result is
¡°adequate offerings,¡± that although generally superior to the weak preceding
internal offerings of the larger consumers, are still short of the mark in
providing excellent solutions to the market. Clearly
we need to mix and match ¡°best of breed¡± solutions from those suppliers expert
in the provision of the individual components that make up the infrastructure
surrounding these often excellent, core products. And the solution is quite
apparent; we must center product offerings around key standards. It is THIS
style of approach that ultimately dominates industries, matures them and
enables the required level of productivity and growth to be attained (See
Figure 1, ¡°Infrastructure¡± phase). The
analogies for the maturing SoC design industry are obvious. We have market
segments served by companies in every phase of development (again, see Figure1)
and the result is the infamous ¡°design productivity design gap,¡± which can best
be closed by collaboration on standards that will energize the growth and
productivity essential to address TTM and success-rate demands of the
electronics industry. Standards are central to optimizing SoC design success. There
are many signs of industry awareness and collaborations do exist. Although it
is not possible to single out more than a handful of truly ¡°universal
standards¡± in the SoC design space, it is clear that groups of companies are
now banding together to take on problems they see as ¡°bigger than themselves,¡±
or, certainly too large to take on without a divide and conquer approach. Some movements have been around longer
than others and have enjoyed differing levels of success and failure. Examples
are: STARC, OCP-IP, VSIA, Spirit, Si2, VCX and so on, all with rather different
missions, infrastructures and objectives. Most efforts represent the practical
alliance of an originally small grouping of Corporations, with their efforts
centered upon fast returns in a few areas, or establishing interface boundaries
between themselves and others in the group, to permit more efficient sharing on
¡°a couple of fronts¡± and solve some immediate problems, on a finite scale. Some
of these organizations have subsequently grown considerably. In itself this is
a positive approach and history says that as the resulting solutions are made
more openly available, they can often evolve into broader ¡°industry¡± standards.
Perhaps not the most comprehensive, organized and rapid approach, but better
than no standards at all! Interestingly,
it is possible to do MUCH better than this, if an environment exists where
standards can be rapidly laid down with some certainty of adoption. It seems
that much of the burgeoning ASIAN SoC design community may well be able to take
advantage of the newness of their industry to lay such foundations, providing
they are fully aware of lessons learned about the real nature of standards and
what it takes to get them rapidly deployed and adopted in a high-tech
community. The success of standards depends on how they are defined,
implemented and the nature of participation in these activities. Let us
consider some pointers surrounding these issues. Firstly,
Standards need to be treated like products. More than this, they should be
defined as products that fulfill a market need, rather than be created because
they are readily implementable and seem important. There is a fundamental need
for true product marketing in this definition process and it is essential that
the whole activity is ¡°market driven,¡± rather than ¡°product driven,¡± as is
typical of the high-tech arena. So, a standard should exist in the context of
the many other standards that should make up the entire ¡°suite¡± of standards.
More than this, standards must been seen in the light of their application and
use. It is insufficient for a standard to be ¡®just a paper document,¡¯ as this
will bring no assurance it can and will be used productively. More ideally, a
standard is a specification AND the tools, support, training and often
multi-faceted, broad-based infrastructure essential for its productive and
prolific deployment to the community it serves. Next,
it is essential we are sensitive to the balance and mix of infrastructure we
provide to surround any given standard. Just because one standard really needs
an introductory training program, hot-line support, exhaustive documentation
and compliance tools, does NOT mean that every standard requires the same. Some
standards are incredibly simple and self-evident, so require less hands-on
support, but almost ALL standards will have users who need to ramp fast and get
answers quickly and cannot afford long delays to get answers. So, efficient
support is normally key, but the mix and extent of infrastructure surrounding a
standard can vary wildly. This of course means that the way infrastructure is
developed around a series of standards must be well-planned ahead of time and
the supporting organization must have the inherent flexibility to configure and
deliver the very varied infrastructure surrounding individual standards. Finally,
the nature of cooperation and collaboration in defining, developing and
delivering standards is key. Participants must be committed and available.
Standards are not things that are ¡°nice¡± to have in an efficient SoC design
process; indeed they are the essential foundations upon which all
else is built. In a more productive world, engineers either develop standards
or use them. It is imperative that the brightest and best engineers are
available to build these foundations and that they are capable of working in a
collaborative and productive fashion with individuals they might in many ways
see as competitors. Sharing with a competitor is a mature behavior that must be
FULLY embraced for all parties to grow, progress and thrive from the
collaboration. It is particularly important that those leaders contributing
such skilled professionals to the combined effort not only make them fully
available, but also commission their work with a productive sense of it¡¯s
importance and the high value of a shared and common outcome. Clearly,
standards are essential to a rapidly advancing SoC design industry. Their
nature is such that in most ways they are ¡°just another type of product¡± and so
should be subject to similar product marketing practices. More than this, when
standards are planned a great portion of their successful deployment will be
determined by the accompanying supporting infrastructure and the suitability of
the supporting organization to deliver this in an appropriate manner. As
standards become the foundations for successful industries, so then must the
developers and providers of these standards be sensitive to the value of
collaboration in moving all participants forward to success in a shared
objective. In
summary, standards themselves are often misunderstood and indeed many existing
¡°standards bodies¡± often fail to recognize the basic nature of their work (much
of which is described above) and indeed continue to founder on some or many of
these fundamental principals. The good news is that awareness is growing and
organizations will adapt to the basic principals and structure that it takes to
successfully develop and deploy their standards. It
is clear that for industries to grow, companies must move away from a model
that is inward-looking and centered purely on internal resources, to a larger
outward approach of collaboration that should be centered on open standards and
broad industry support. Individual standards might vary widely, but is clear
that the SoC Design industry will not attain the productivity required for
their demanding markets, without paving the way with well-defined ¡®products¡¯
and very effective deployment organizations.
Dr. Danny Rittmen¡¯s Column ¨C Nanometer Reliability
Me Too Is Not My Style (Part IV) - Acer Group
A Practical Guide to High-Speed Printed-Circuit-Board Layout - Analog Devices
Verification Methodology Manual (Part III) - TransEDA
The Hacker Social World and Floss (Part VI)