I believe we are living in the end times, an era which will see the termination of an age. The war between the two great enemies will consign reality as we know it today to oblivion.
The day is soon coming when AMD and Intel will make it impossible for all but the most affluent enthusiasts to overclock their CPUs.
A bold statement, I know.
Maybe even a touch provocative.
But if I am reading the tea leaves correctly, I truly foresee that overclocking as we know it today will cease to exist. There are several reasons why I believe that this is an inevitable outcome.
The "Core War" between AMD and Intel is the primary reason why I think that overclocking will someday be impossible for all but the most affluent enthusiasts. From the sidelines, it looks that it's far easier to simply keep on adding more and more discrete processing cores to a single CPU package than it is to design, test, manufacture, and sell poly-cored CPUs that have massive performance headroom left untapped.
Though there is still huge prestige attached to having CPUs which overclock like crazy, I think that there will come a point when you simply cannot have both a maximized number of discrete cores as well as overclocking headroom designed into the chip.
Something has to give.
If given a choice, I think that both AMD and Intel will favor simply increasing their products' "processor densities" (a term I'm inventing here, to describe the number of discrete processor cores within a given space). Given that it seems far simpler to just increase the number of cores than it is to cater to an enthusiast audience and design poly-cored CPUs with bags of overclocking headroom, the manufacturers will go where the dollars are more forthcoming. In other words, there are far fewer enthusiasts than there are "average" computer users. Not only that, but the corporate sector, which values stability far more than just outright speed, is a bigger revenue stream than the entire enthusiast market sector by orders of magnitude.
There are three major, significant, and inter-related consequences of having a chip with overclocking headroom. These are 1) increased power consumption, 2) increased thermal output, and 3) increased complexity and cost.
Not surprisingly, these three major consequences are also largely true of multi-core processor designs. There is no getting around the fact that higher performance parts are more hungry for power, run hotter, and are more expensive not just in initial purchase terms but more importantly in terms of actual use and ownership. It doesn't matter how you achieve the performance increase; the consequences are exactly the same.
Effective heat dissipation is one of the most crucial factors of a successful overclock. Increasing the number cores by itself necessarily increases the demands on a system's CPU cooling. A single-core chip runs cooler than a dual-core chip; a dual-core runs cooler than a triple- or a quad-core. Thermal output, therefore, scales upward as you increase the number of cores in the CPU package. Even at stock speeds, even at lower VCore levels per CPU, the more cores you have, the hotter your CPU runs. Forcing chips to run faster than stock will always increase the thermal output, and the increased number of cores merely exacerbates the problem of excess thermal energy that requires dissipation. The ultimate consequence is that even water cooling will no longer be adequate to maintain stable operation with a certain number of multiple cores. To maintain effective, stable overclocks, more exotic (and expensive) means of CPU cooling will need to be employed. The more cores you squeeze in, the more this becomes true.
Only the hardest of the hard core, those with the ability to spend money with impunity to do what is necessary, will likely be able to overcome these immutable realities defined by the laws of physics and chemistry. While we all might want to satisfy our need (or greed?) for ever-increasing performance, I have to say that most of us will have no choice but to admit defeat and accept that our wallets simply cannot afford to sustain our pursuits of performance.
But there are other reasons why I think overclocking as we know it will be forced to go the way of the dodo. One of these is that software these days is miles and miles behind the capabilities of modern hardware. Modern hardware is so advanced compared to the software available, the hardware isn't being used to its maximum potential. This means that hardware performance is inefficient.
Inefficiency in any machine is undesirable. Though the manufacturers will likely want the bragging rights to having the most cores, overclocking prowess is largely relevant only to the enthusiast sector. As I've said before, the enthusiast sector simply is not big enough to dictate market decisions completely. Overclocking only exaggerates the inefficiency inherent in the current gap between hardware and software.
Finally, in the minds of non-enthusiasts, bigger numbers are always more attractive. If we say that only enthusiasts indulge in overclocking, the number that becomes most relevant to purchasing decisions will likely be the number of cores under the hood. Clock speeds will matter, of course, but even today the focus seems to be fixed on the number of cores. The manufacturers will chase the money where it is easiest to come by.
I would be glad to be wrong in my analysis of the situation. Some might even say (quite rightly, too), that my analysis is pessimistic, even cynical. However, there is no question that no design conceived by humans can cheat its way past known physical, chemical, and thermodynamic laws. It's not my prognostication that stands in the way of the continuation of the lifestyle of the modern overclocker.