Originally Posted by
iddqd
Overclocking in general is sorta dying, in no small part due to cheaper and better profiling tools available at the foundry. 15 years ago, these tools didn't exist, and nobody really knew how good their silicon was (well, without spending more money than the profit you'd get from selling the chip, so you'd rather not find out). In the interests of making sure things work, they would market their silicon as if they were all the worst-case scenario in silicon quality. This resulted in companies leaving a lot of performance on the table, which you could then help yourself to by investing a lot of time into finding out just how good your chip was.
Things are different now. Profiling tools got really good, and really cheap. The fab knows EXACTLY how fast the critical path in your chip is. Armed with this knowledge, you can actually design your own auto-overclocking circuit that will push the silicon right to its theoretical max for every imaginable path through the logic. There's no way a human can compete with a machine like that. We aren't seeing this quite yet, by the way, but the first steps are already there, and I promise you it is being worked on. For now, the auto-overclocking technology is good enough to do a better job than most (but not all) overclockers. And it's only the first step.
It's bittersweet. On the one hand, the fun in trying to get a free lunch out of your expensive new toy is gone. There's no free lunch; if you want a faster CPU (or GPU, or whatever), the only answer is more money. On the other hand, the 99.9% majority of consumers that either don't know about the existence of the free lunch, don't know how to obtain said free lunch, or simply can't be bothered will benefit from the hardware doing it by itself.