Have you ever wondered why neither Intel nor AMD have ever seriously tried to stop overclocking? There's been times when they looked like they would. There's even been times when I thought they would. So why haven't they? How Big Are We? Are they afraid of losing our business? No. The best guesstimates of the overclocking population seem to center around 1% of the computer population. It may be a bit higher; it may be a bit less, but that estimate is probably not too far wrong. 1% of the CPU population means about 1.25 million processors a year. Given the propensity of overclockers to buy lower-priced CPUs, I would be surprised if the average price paid per overclocked CPU was much more than $150. This means we spend about $200 million a year on processors. We'd probably not be far off splitting that 75% for Intel, 25% for AMD. That's about $150 million in revenue for Intel, $50 million for AMD. That's not a lot of money for Intel; proportionately more for AMD, but nothing earthshaking. So while it's not chump change, if overclocking were costing Intel or AMD serious money through returns or remarking, it's not enough to be a real deterrent to stop either from stopping overclocking. So what stops them? When The Cure Is Worse Than The Disease A lot of problems in this world don't get fixed because the systemic cure to a problem costs more than the costs of the problem. How much does overclocking cost Intel and AMD? If you realistically take all the factors into account, it's probably roughly a wash. Overclockers probably spend a bit less than they otherwise would without overclocking, and probably have a somewhat higher return rate. On the other hand, they probably buy CPUs more often than the average computer user. Stopping overclocking, though, would cost real money. The technology exists to stop overclocking dead in its tracks, to ensure that no processor ran faster than it should. Any real solution would require special circuitry that would require a good deal more additional design time and take up valuable silicon real estate. There's a good chance that such circuitry would make CPUs less reliable. More importantly, that change would have to made to every single processor made, even though 99% of them would never need this "cure." So to stop the 1% problem, you have to pay to cure 100%. Intel makes well over 100 million chips a year. AMD will probably make 30 million next year. If an overclocking bar cost just a dollar a piece (and I'd bet the real cost would be several times that), that's an additional $100,000,000 cost to Intel and $30,000,000 to AMD. Overclocking certainly doesn't cost either of them anything near that much. Of course, if overclocking ever became really mainstream, if not 1% but 10% or more did it, then we're probably looking at some serious revenue loss, since a greater proportion of overclockers would have otherwise have paid more for their processors. At that point, a cost like that might be justifiable. For now, though, it isn't. So Intel and AMD come up with "solutions" that really don't stop the problem, but make the task a bit harder (scaring off many wannabes) and are very cheap or practically free to implement (doesn't hurt the bottom line). Overclocking to Intel and AMD is like a stray fly buzzing around you. It's annoying, but not worth the bother to kill, so you shush it away. Get a swarm, and then you take real action.
