Introduction
Moore's law states that the
increase in complexity of microprocessors and other types of
semiconductor integrated circuits, will double every 18 months.
History shows that so far this has been the case and in fact
the observation made by Gordon Moore has become a self-fulfilling
prophecy. This doubling in complexity has resulted in a phenomenal
growth in the amount of computing power available on a single
microprocessor chip. However, the cost of developing these chips
and providing production facilities is also growing, and the
profits from one generation of chips are needed to finance the
development of the next generation.
This article addresses a number
of issues surrounding Moore's Law. One of these is the extent
to which Moore's Law is driving the development of applications.
Another issue is the importance of Moore's Law in different industrial
sectors. The question of whether in some sectors there are more
important issues other than available computing power is considered.
The article also addresses what will come after the microtechnologies
of silicon, when it is no longer feasible to reduce the size
of a silicon transistor. The article deals with the topic of
nanotechnologies, seen by some as the step that eventually lies
beyond the micro.
Main Issues
A new era is emerging in which
computing is becoming invisible - microprocessors are embedded
in many objects and interactions with these devices needs to
become more user friendly and less reliant on keyboards and other
traditional interaction devices. Achieving a situation where
computers are ubiquitous and interaction is natural and effortless
requires a massive research effort.
Applications centred around
broadband residential applications include home entertainment,
home automation and learning. The vision here is one of more
applications and communications with silicon technology everywhere
and always switched on. Moore's Law appears to be fundamental
in achieving this vision, by providing increased computational
power for signal processing and applications. The goal is a system
on a chip, and one of the main drivers for this is cost reduction.
Achieving a system on a chip requires more complexity on chips,
in other words Moore's Law is an enabler.
Applications in the automotive
industry are however different. In the automotive sector there
are many application specific components, with up to 50 per cent
of the electronic content in cars falling within this category.
There are also many different types on applications in cars,
including motion sensing, ignition, in-car entertainment, navigation,
and environment sensing. Many applications have the potential
to either improve safety or fuel consumption. However one of
the main challenges is reducing warranty costs. Cars represented
a harsh physical environment and there is a significant need
to reduce component failure. There is a problem referred to as
a design gap. Design productivity is increasing more slowly than
the complexity and capabilities of the technology. Moore's Law
is not the main issue in automotive applications, there being
other more significant problems to address.
Smart card applications provide
the basis for secure transactions. Smart card technology provides
the personal enabling factor for a networked society. Smart card
technology, however, is about three to five years behind the
technology that is used in personal computers and other advanced
applications. The risk of damage to the smart card electronics
means that the usable space on a smart card is limited. There
is a need therefore to increase the complexity of the usable
space to provide new functionality and better security. Therefore,
Moore's Law is relevant, even though the smart card industry
is still working with earlier generations of microprocessor technology.
What will happen when the physical
limits of silicon technology have been reached? One possible
answer lies in nanotechnology. The physics were different on
the nanoscale. Surface effects predominated over bulk effects
and the materials are also less stable. Nevertheless, there is
significant application potential, for example in creating drag
free materials for aerospace, or carriers for other materials,
or super strong materials. There are already companies selling
products based on nanotechnology. Manufacture of nanomaterials
is a significant challenge, as is the scaling up of processes
for mass production.
The cost of developing nanomaterials
is a possible. The development of new generations of silicon
technology is paid for from the profits generated by earlier
generations of technology. This raised the issue of how the development
of nanotechnology would be funded. This may not be a major problem
as many different industries are interested in the technology
and what needs to happen is that different contributors should
work together. The important point about nanotechnology is that
it is relevant to industry in general and is not just an issue
for the semiconductor industry.
Conclusions
and Future Directions
In the area of broadband residential
access, the key issues are achieving improved signal processing
and development of systems on a chip. System on a chip implies
increasing chip complexity and also better design methods, especially
at the system level and achieving hardware and software co-design.
Moore's Law is therefore important in this area. In the automotive
field some of the main challenges are better design methods and
improved packaging. For this industry, Moore's Law is not the
main concern. In the smart card industry Moore's Law is of relevance
as there is a need to add more to a limited area on the card.
Looking towards the end of Moore's Law, when the physical limits
of silicon have been reached, nanotechnologies may provide a
way of enabling the continuation of this law. It is likely however
that silicon will still be used in conjunction with nanotechnologies.
It is certainly the case however, that many applications still
need Moore's law, and will continue to do so for some time to
come. |