What Overclocking Is

I thought you might find it interesting to see, not so much my answers, but what kinds of questions (and underlying assumptions) get asked.

I’ve made some slight changes for purposes of clarity.

Introduction

I’ve written quite a bit, probably more than you’d want in certain areas,
mainly because some of your questions carry underlying assumptions that
really aren’t true. To a large extent, they seem to reflect what the CPU
manufacturers would like people to believe, which isn’t really, or even at
all so. That’s understandable, after all, they’re the ones who talk about
it.

The reality is overclocking simply takes advantage of how CPU makers design
and manufacture their chips. It is greatly encouraged by dubious price
structures. Adopt a more rational pricing structure, and most of it would
vanish, leaving a much smaller core of hobbyists.

Excluding the possibility of price reform, overclocking is inherently a
self-limiting phenomenon. So long as it is a fringe activity; it’s not
worth the effort and expense to stop. Should it ever approach becoming the
norm (which is highly unlikely) then most of it could be stopped at a cost.
The approach of the CPU companies seems to be to discredit the practice with
words, but little action, which on the whole is fine by us; this isn’t
exactly a struggle for human rights. 🙂

________________

Question: I’ve read a lot of overclocking sites recently, and while they have loads
of details on how to do it, I don’t get a feel for when it all began?

First, let me explain what overclocking is.

Imagine a factory making auto engines. This factory makes only 200
horsepower engines. It turns out there aren’t enough buyers out there for
200 horsepower engines, but there are plenty of people who’ll buy cheaper
150 horsepower engines. Pretend it doesn’t cost any more to make a 200
horsepower engine than one of 150..

What you do is take the engines you can’t sell as 200 horsepower engines,
and sell them as 150 horsepower engines rather than have them sit in the
factory.

This is what the CPU makers, whether it’s Intel or AMD, basically do. They
need at least some of the more powerful CPUs, but unlike most other areas,
making a less powerful CPU doesn’t cost them less. So they try to make all
of them 200 horsepower, but sell most of them as 150.

Overclocking is simply taking that 150 horsepower engine and running it at
200. It takes advantage of any difference between the stated and actual
potential of a CPU.

This has been going on a long time. We recently asked our audience and
found people were overclocking long before the personal computer. A couple
overclocked mainframes and minicomputers; quite a few more overclocked
calculators. So the roots of overclocking are deep.

Personal computers have been overclocked for almost as long as they have
existed, even before the IBM PC. Probably the first overclocking effort to
get some attention was to change crystals on an IBM AT. This got you from 6
to a whopping 8 or 9Mhz.

The popularity of overclocking in the pre-Internet days waxed and waned with
its relative difficulty. Changing crystals and soldering components is much
more difficult than flipping a switch. Over the past three years or so, it
has been relatively simple to do, and has yielded significant low-risk
rewards, so more are interested than when the rewards weren’t as great and
the risks higher.

The 800-pound gorilla of overclocking today is the Internet. You no longer
have individuals or groups working in relative isolation. You no longer
wait for a magazine to come out. You no longer wait for a monthly meeting
or yearly convention. You no longer wait, period, for information, and it
can now come from anywhere..

If someone on the other side of the globe discovers something Monday
afternoon, the overclocking world knows about it by Tuesday morning. It
doesn’t take weeks or months for information to spread, which means it
doesn’t take weeks or months for people to act on it, or go on from there.
It collapses the time function, which greatly accelerates the pace of
change. It’s a great example of what happens when you give any group
instant cheap global communications; it changes the pace of change.

Do you know who or which groups kicked it off, where, or when?

Probably since the early days of radio, you’ve had people tinkering with
electronic components. When computers became cheap enough to tinker with, a
lot of these folks moved over. When doing it got simpler, a lot more people
joined them.

Is there a typical overclocker?

No.

You have two overclocking groups, the hobbyists and the pragmatists. The
hobbyists do it for the joy and challenge of pushing a piece of machinery to
its maximum, and while most are frugal, saving money is not the top
priority. Overclocking is usually a pretty cheap hobby over the long-run
compared to, say, bar-hopping.

The pragmatists view overclocking more from a cost/benefit analysis: I can
save X by spending Y and doing this and that.

Many overclockers start off as pragmatists, then drift into the hobbyist
camp.

What are the biggest technical challenges to cooling the processor?
(Presumably there must be a lot of OC’s who have flooded or frozen their
computers? It looks like a risky business).

Like a person, a CPU needs energy to work, and since it’s not perfect, it
emits a waste product. A CPU eats electricity, and emits heat as waste.

As time goes by, CPUs are asked to do more and more work faster and faster.
As with most things, this usually means more power. Unlike most things,
though, at the same time, they also need to get smaller and smaller. The
general trend has been to shove more and more power into smaller and smaller
spaces. So they get hotter and because they’re smaller, it gets harder to
cool them down.

The options you have to cool a CPU are pretty much the options you have to
keep yourself cool on a hot summer’s day.

You can decide to just not do very much. You can decide to fan yourself, or
buy a mechanical fan to do it for you. You can decide to go to the pool or
the beach, or just sit in the bathtub all day. You can buy an air
conditioner, or, if you’re really extreme, you can build or buy a room-sized
deep freeze and lock yourself in.

Each succeeding step takes more effort and expense. You determine whether
or not it’s worth the bother.

The vast majority of overclockers are at the mechanical fan level. A few
have reached the bathtub level. Even fewer have done with equivalent of air
conditioning, and only the most extreme have gone beyond that.

When people ask me, “How risky is overclocking?” the real questions that
needs to be answered is “How far are you willing to go for how much, and how
much of a fool are you?”

Overclocking can be as risk-free or as risky as you make it. It’s like
going to the zoo; you can just watch the lions, or you can climb over the
walls for a closer look. If you do the latter, you may well get eaten, but that hardly makes all
lion-watching a perilous business.

As a general rule, you can get about 80% of the total benefits from
overclocking by taking a few minimal, risk-free steps, and that’s just what
the vast majority of overclockers do. You buy proper equipment, put it
together, and let it fly, and if you’ve taken reasonable care and have
realistic expectations, it will usually work. Even when it doesn’t, it’s
not “kill or be killed,” you still have a perfectly functional computer
system; it just won’t be quite as fast as you had hoped. I’ve put together
about a dozen systems, and every one of them have reached my goals.

A small minority do more than that, and sometimes do take extreme steps.
Some people like climbing hills; others like climbing Everest. The latter
is certainly riskier, how much more risky largely depends on how well you’ve
prepared. However, considering all overclocking risky because of this is
like citing Everest fatality figures to someone escorted novice
hill-climbing day trip.

However, an idiot can turn crossing the street into a suicide mission. If
it’s just a matter of time and circumstance before your life becomes a proof
of the negative personal aspects of natural selection, overclocking can be
very dangerous. But then, what isn’t?

If you view overclocking as a Nietzscheque exercise of sheer will
unencumbered by reality, you are quite likely to get yourself into trouble.
You just can’t intimidate what is organized sand, but that’s not the sand’s
fault.

What does it feel like when all the effort pays off, and your overclocking
rig works?

Depends on the type of person you are. At the least, the kind of
satisfaction you get from figuring out a puzzle. At the most, a sense of
personal triumph. Some no doubt find they’ve put together a status symbol
on the cheap.

Finally, I understand that some manufacturers (Intel?) are locking their
chips.

Since August 1998, Intel has blocked one avenue by which one could
overclock. However, they’ve left others wide open, so this “locking” has
been no more than a minor inconvenience; it hasn’t stopped significant
overclocking of its chips at all.

But overclockers seem a smart bunch. Do you think they will always find
ways to overclock processors, or will the opportunities dry up as
manufacturers clamp down?

First, there are two aspects to overclocking. What I’ve been talking about
is an individual either doing the deed himself or knowingly hiring someone
else to do it for him.

What also happens, though, are people and companies who overclock a
computer, sell the results without making the buyer aware that he is buying
an overclocked machine, and pocketing the extra profit. That is called
“remarking” and no overclocking enthusiast supports that practice; that’s
simply fraudulent.

The CPU companies quite legitimately are opposed to remarking. People are
paying an inflated price for what they are getting under false pretences.

The situation is different for an overclocker. At first glance, it looks to
be the same, in both cases, the company is getting less for the product than
they might otherwise. However, if overclocking were completely blocked
tomorrow, the vast majority would not spend more for their product; they
would simply accept less.

Some would have you believe that overclocking causes a high rate of CPU
failure, and that adds to the manufacturer’s cost.

Reasonable overclocking does no such thing. So long as you do not push a
CPU beyond it’s limits, there is little to no difference between the
“regular” chip and one that is overclocked. In fact, because the average
overclocker makes more of an effort to keep his CPU cool than the average
computer manufacturer, many overclocked CPUs will probably last longer than
many “regular” chips with poor cooling and inadequate insulation.

CPUs are currently designed to last far longer than their effective useful
life, anyway.

Of course, a fool can manage to burn out a CPU. However, most overclockers
are not fools, and it takes considerable effort by even a fool to do this;
there’s a lot of margin built into CPUs. In fact, the CPU companies
themselves have used the same tricks used by overclockers to get chips
running a bit faster than they were designed to be.

There’s no denying that CPU manufacturers incur some additional warranty
costs due to overclocking, though not as much as they might have you
believe. However, overclockers tend to also upgrade their machines and buy
new CPUs more often than the average computer user. They effectively pay
less more often. Most of those who are regular readers of our site have
bought two CPUs the past year. They would not have done so if they had not been
overclocking. Between the two, I figure it’s a rough wash.

So the CPU companies have an undeniable and legitimate interest in stopping
remarking, but not overclocking, which is a conceptual annoyance, but not a
financial drag. Can you stop one without stopping the other? Yes.

Intel has placed code inside each CPU that unalterably identifies the rated
speed of the processor. They have a small program which reads this code and
tells you what the official speed of the processor is. This allows positive
identification of any remarked chip, while letting overclockers do what they
like.

It would be a simple task for AMD to do the same, but they have not done so.
It is very difficult to take AMD claims of trying to stop remarkers
seriously when they have not taken this simple step.

Is overclocking itself a major threat to the revenues of CPU companies?
Now, no. There was a surge of interest in overclocking a few years ago, but
growth appears to have stabilized; this is not something that is growing at
an exponential rate.

Could overclocking ever conceivably become a big threat to CPU
manufacturer’s revenue in the future? That is very unlikely for a simple
reason.

The average person buys a computer made by a big computer manufacturer. The
big computer manufacturers have invariably installed equipment missing certain features necessary for
overclocking. To make such a machine overclockable, you’d have to rip the
machine apart and put it something through which you can overclock. The
average person simply will not do that en masse.

Given that, the issue becomes whether or not the CPU manufacturers will ever
become so irritated by the number of overclockers actually out there that
they’d take dramatic action.

Let’s say you don’t want your neighbors to drop in. You can ask them not to
do it. You can lock your doors and not answer the doorbell. You can have
Rottweilers prowl the property. You can plant mines in your yards, put
motion sensors all over the place and create automated fields-of-fire. What
do you do? That all depends on how determined you are to keep them out and
how determined how many of them are to get in.

It would be almost impossible to make overclocking literally undoable. It
would be a much simpler task to make it difficult enough to deter the vast
majority of overclockers, leaving only the most determined hobbyists. If
the threat were big enough, and a CPU company were willing to pay the price,
it’s certainly doable. However, is it and are they?

The CPU companies appear to be ambivalent about overclocking as it stands
now. On the one hand, they hardly like the concept that at least some of
those overclocking are spending less money on their products than they do
otherwise. On the other hand, they certainly don’t like the idea of
spending any significant amount of money on 100% of their processors just to
stop 1-2%.

The policy up to now has seemed to be to talk a big game against the
concept, while only taking actions with negligible cost which discourage the
average person from trying this. Should overclocking ever become a much
bigger factor than it is, I’m sure more drastic action would be taken. I
doubt it will come to that.

A big incentive for overclocking over the years has been the rather lopsided
pricing structure of CPUs as has been dictated by Intel. The price of the
current fastest processor is far, far higher than can be justified by the
increase in performance. When you find out you can get the same
performance of a CPU that costs hundreds of dollars more with minimal
effort, that’s very tempting.

A very significant development in just the last few months has been AMD
levelling their price structure to more closely approximate the improvement
in performance. You will have many more casual overclockers if it’s a
matter of saving $500 as opposed to $100. AMD’s pricing actions will
probably reduce overclocking more than any antioverclocking technical
measures they might undertake.

Email Ed