Lessons from Discarded Computer Architectures - History of Computing: Learning from the Past - page 203

The general application of parallel processor architectures depends upon theoreti-
cal advances that can establish the importance of parallel universality [24].
Others have investigated parallel processor arrays. These include Intel, whose
iWARP product was evaluated for use in the international space station [25] where it
gave computational speed increases of an order of magnitude over a single processor
equivalent. Product development led to a commercial supercomputer product, the
Intel Paragon XP/S [26].
Despite these advances, parallel computing has not achieved the hopes of the fifth
generation computing project, and applications appear to remain restricted to special-
ised (but important) situations. Interest has moved to quantum computing, where the
superimposition of energy states are expected to be processed using optical methods
to solve problems extremely quickly.
5 Conclusion
There appear to be four lessons to be learned from this story of discarded computer
architectures. Firstly, the BBC microcomputer was quickly supplanted by the open-
architecture IBM PC and the closed architecture Apple Macintosh. However, the
RISC processor designed for the BBC microcomputer's successor has been widely
adopted because of its very low power consumption. In a world faced by climate chal-
lenge and a huge growth in the use of mobile computing devices, this has been a win-
ning strategy.
The second lesson can be drawn from the way personal computer operating systems
have been largely linked to processor families. Most of the main lines of growth have
been limited to von Neuman uni-processor architectures. Even the development of
quad or more core processors are still just variants of this sequential flow machine,
albeit allowing a few more threads to access different parts of memory simultaneously.
The lesson of how to break away from this design template has yet to be learned.
The third lesson relates to the Transputer and ways in which these chips could be
wired in arrays. Even declarative languages such as Prolog were not ported success-
fully to this architecture in such a way to facilitate problem solving which became
mainstream. We appear to lack an understanding of how to implement solutions using
these kinds of techniques.
Finally, our hopes of leapfrogging this difficulty rest with novel computing tech-
niques such as quantum or biological constructs. It remains to be seen how successful
these will be - and they may need to take heed of the other three lessons to achieve
their goals.
References
1. Nichol, J., Briggs, J., Dean, J.: PROLOG in education. Educational Review 39(2),
137-146 (1987)
2. Nichol, J., Dean, J., Briggs, J.: Teachers encounter PROLOG. Journal of Computer
Assisted Learning 2(2), 74-82 (1986)
3. Hornby, T.: Acorn and the BBC Micro: From Education to Obscurity (2007),
http://lowendmac.com/orchard/07/0228.html (February 12, 2010)