Page 184 - Visions of the Future Chemistry and Life Science

Exploring human organs with computers 165

limited mainly by acceptable analysis times. The speed of inexpensive
commodity microprocessors has increased exponentially since their intro-
duction three decades ago, doubling every 18 months. If this were to con-
tinue, by 2010 they would be 100 times more powerful than today’s, and
they would be cheaper in real terms. Unfortunately, physical limitations
to both transistor density and switching speed will almost certainly limit
increases in the power of individual microprocessors.
An alternative is to look to the Internet, whose growth is sure to con-
tinue unabated, driven by factors as diverse as minimising the drudgery of
grocery shopping to the widespread adoption of working from home, as
people strive to avoid the damaging social and environmental effects asso-
ciated with commuting. This leads naturally to the concept of distributed
parallel-processing techniques, which divide the task of analysing the
finite-element model between several processors that are housed in separ-
ate computers in different locations. By utilising commodity computers
we benefit from the economies of mass production that are associated with
sales of tens of millions of units annually.
Distributed parallel processing also provides the potential to utilize a
wasted resource. Many people have a computer in their office or in their
home that spends more than 99 per cent of its time doing little more than
providing low levels of background heating and noise. It makes sense to
give them something to do when they are not being used as expensive type-
writers or handheld calculators. The utilisation of only 50 commodity
computers would, with virtually no capital investment, provide a distrib-
uted parallel application with the processing performance that a single-
processor computer will not be able to match within the next 10 years.
And, of course, as individuals computers are upgraded the distributed
application will have immediate access to the increased power.

9.9 The year 2020

It is difficult to predict developments beyond the first decade of the twenty-
first century. However, there are two arenas in which modelling and
biology may converge even further, namely developmental biology and
carbon-based computing. Developmental biology is an area of experimen-
tal research that is expanding rapidly. Current tissue-based work on coch-
lear regeneration highlights the difficulties of artificially controlling the