July 20, 1998
Can computers ever be fast enough?
By Bob O'Donnell
Last week's news was filled with reports of the dire straits that the semiconductor
business is currently in, as well as predictions that the chip industry's woes are far
from over. Numerous reasons for the slump were cited including the Asian financial crisis,
overproduction and excess inventory, and the surging interest in low cost, low margin
microprocessors and memory modules.
Along with that news came commentary from industry analysts and other industry watchers
that computers are reaching the point where they are "fast enough" for most of
today's applications. Their argument is that there really isn't a compelling need to buy
computers that use faster, more expensive processors -- which are usually more profitable
for semiconductor manufacturers -- so IT departments and consumers aren't buying them.
This, in turn, is affecting and will continue to affect the near-term and long-term
profitability of companies that manufacturer microprocessors.
The question I have is, are computers really fast enough? Will they ever be fast
enough?
I have to admit that I'm really torn on this question. On the one hand, I hate to
promote the idea of needing to always buy newer, faster hardware, just because it's
faster. And the truth is, there are very few programs that really tax even today's
entry-level PCs, let alone midrange and higher-end models.
Some people point to voice recognition products, such as Dragon's NaturallySpeaking,
IBM's ViaVoice, Lernout & Hauspie's Voice Express, and others as being examples of
software that pushes the frontiers of computer hardware. But even these programs typically
only require a 166-MHz MMX processor, which you would be hard-pressed to find on any new
desktop computer, regardless of how inexpensive it was. Even most low-end notebooks have
at least a 166-MHz MMX processor. The real limitation here is the amount of memory the
system has. Most of these programs require at least 32 MB of RAM and are happier with even
more. Again, however, even entry-level desktops have 32 MB of RAM, and 64 MB is not at all
uncommon for low-to-midrange priced computers.
So, are today's computers fast enough? Well, as someone who spends a good portion of my
day in front of a computer, I grow frustrated even with tiny delays. I don't want the
computer to interfere with my thought processes at all, so if I have an idea or decide to
do something, I don't want the computer to get in my way. In fact, until absolutely every
action I take can happen instantaneously -- whether it's booting a program, opening a
large file or switching between applications -- I won't consider a computer to be fast
enough. And I don't know about you, but I've yet to see or work with a system that even
comes close to this ideal.
Admittedly, there's a lot more to making a computer fast than just having a faster
processor. In fact, it might be that today's processors really are fast enough to meet my
goal, but the rest of the system surrounding the processor isn't up to the task. New
memory technologies, such as the forthcoming RAMBus, should help in this regard, but there
are lots of pieces inside the computer that need speeding up, particularly hard drives. As
fast and efficient as today's drives may be, they're still one of the bigger bottlenecks
in current computer designs.
Another big problem is Internet connection speed. Unless you're blessed with access to
a T1 or T3 line in a business environment or a cable modem in a home office environment,
the width of your data pipe can have a tremendously negative impact on your overall
computing experience. Unfortunately, as we all know, there are no easy or quick answers to
this problem, yet it's one that has to be fixed to speed up today's computers.
Bloated, inefficient software is another problem that people will point to as keeping
us from my ideal of a completely transparent computer. Again, there aren't necessarily any
easy answers here, although I'd argue that if the computer's hardware were really fast
enough, it could overcome even this problem.
The question of how fast computers need to be is not a new one. People have been and
will continue to debate this issue for many years to come -- despite the fact computers
have continued their Moore's Law march to higher and higher levels of performance. I'd
love to reach the point where computer performance was no longer an issue, but from what I
can see, we aren't there yet.
©
Copyright 1998, by InfoWorld Publishing Corp., a
subsidiary of IDG Communications, Inc. Reprinted from InfoWorld,
155 Bovet Road, San Mateo, CA 94402. Further reproduction is prohibited.
