|
 Our claim to innovations here is
the formal definition of the concept of platform and
the definition of platform based design as a
cornerstone for a refinement-based design
methodology. Two of the partners are well known for
their work in this area.
To realize our vision, we have on the hand to develop
formal techniques at the abstract level so that verification
is started early and with the correct set of tools and
methods.
On the
other hand, we have to think of ESW and hardware
architecture in a unified and harmonious way.
In addition
to the pressure for system designers to choose flexible
means of implementation, we are witnessing also a growing
attention of the IC manufacturers towards chips that can be
shared across several designs so that the development cost,
which is also increasing by leaps and bounds as
manufacturing processes evolve below the .2 micron barrier,
could be amortized over a large number of units.
This
alignment has resulted in the birth of platform-based design
where re-use and programmability are the name of the game.
The concept
of platform is related to System-on-Chip (SoC) but it is by
no means a synonym. Indeed, a platform can actually manifest
itself as a chip-set, perhaps a System-in-Package (SiP), or
an embedded system-board or higher-level object. We define
the basic concepts of platform-based design later, but here
it suffices to note that in platforms, the indispensable
presence of programmable cores requires chipmakers to
acquire a new (for them) expertise in ESW.
In this
case, they need to provide the users of their platforms
means for mapping quickly and effectively their desired
functions onto the components of the platform. Today, most
of the platform providers rely upon third parties to provide
a tool chain for programming the cores, but there are many
signs that this business model needs to change. Third
parties are often too small and not equipped to provide
support for hundreds of users and applications and they are
not necessarily interested in developing tools for new
architectures.
Hence, we
are witnessing on one side an increasing consolidation in
the industry where companies like WindRiver are growing by
acquiring other Real-Time Operating System companies to
capture a sizable chunk of the market that will allow them
to think strategically about this very appealing and growing
market. On the other side, we see semiconductor companies
acquire more competence in software tools.
The
role of hardware architecture platform and software
“middleware” has to be clarified. Indeed both
concepts share the idea of isolating higher levels
of abstraction from implementation details. However,
they have not been placed in a unified contest where
this core idea is generalized and provides the basis
for a design method that promises to reduce
time-to-market in a substantial way. Indeed, we will
provide ways to link the control algorithm
development with its software implementation and
RTOS in our methodology.
|
