What makes one CPU a better overclocker than another?

[Josurr]

Its well known that not everything has the same overclocking potential whether its a CPU, a GPU, or even a monitor, this phenomenon is sometimes referred to as the "silicon lottery". Now my question is why? Why are some CPUs capable of reaching such a better overclock than others? I recently saw a video of an Intel lab and it was about how the process is so tightly conducted, everything is done pretty perfectly, to the last little detail, so if that's the case why is there a wide range of potential when overclocking CPUs?

[ttam]

The wafers can't be completely uniform, the manufacturing process can't be 100% uniform. You have to realize we're dealing with some of the smallest scale things in the world. 28nm is 28 *billionths* of a meter. The center to center distance of features of a processor have only around 56 silicon atoms of width between them at that point. (that's center to center, the space in between edge to edge would be fewer than 56 atoms). 22nm is only around 44 Si atoms.

 

Things are so small that an extra atom in the wrong place can produce different characteristics. It's amazing that we get things as consistent as we do.

 

Your motherboard can also play a factor in this.

[ragedev]

Basically, this should explain some more 

When chip manufacturers like Intel, TSMC, UMC, GF, etc. make wafers, there are slight variations in material quality across the wafer surface, there are local variations in how the lithography, metal vapor deposition, photoresist chemical deposition, etc. are done and this can yield a significant contrast between how good the best chip of a given batch will perform vs how bad the worst chip of the same batch will perform.

To maximize profit, chip manufacturers test and sort chips based on various criteria such as leakage current, power draw at key frequencies, salvageable defects, etc. to decide which product range the chips fit best in. That's binning.

When you buy "K-chips", you are playing the lottery: you are guaranteed a chip that performs at least up to stock K-chip standard but you have absolutely no way to know beforehand how much farther beyond that your specific chip can go under any given circumstances beyond stock conditions. That's the chip lottery. Some i5/i7 3xxxK/4xxxK may max out at 4.2GHz while others may hit 5GHz. Some may require 60W@1.2V to get to a given clock rate while others may require 100W@1.35V.

 

Quote from http://www.tomshardware.co.uk/answers/id-1753815/silicon-lottery.html

 

One of Linus's videos explaining binning processes in component manufacturing should clear things up 

http://www.youtube.com/watch?v=B14BiB-Bv3s

 

The binning process applies to GPUs, RAM and much more which is why we get different variations of the same core such as GK104; GTX 670 & GTX 680. The GTX 680 is just higher binned. The same happens with RAM, 2400 MHz & 2666MHz memory are just higher binned than 1600MHz.

 

Intel takes advantage of the binning process by offering CPUs such as the 3960X and 3970X. These chips are expected to be able to overclock to higher frequencies than the 3930K as they are almost 'perfect' chips out of the 39xx series which is why Intel charges double the price.

[Josurr]

I see, this sheds a lot of light on the subject. Thank you very much

[ForsakenLive]

Basically, compare the Silicone wafers to Brownies. When you make brownies  you got those fluffy tasty and big ones on the middle, and those that are not uniform, that lay across the corners, this are usually less fluffy and sweet. You can't really have 100% perfect brownies.

CPU wafers are almost the same, the transistors are extremely small, and even the tiniest difference will set them apart. Sometimes you find you can't deliver that much power to the cores, or set them as high. This are binned and separated, the defective ones are thrown away.

When AMD prepares their 8 core processors, they separate the 8320s, the 8350s, the 9370s and the 9590s according on how much power and clockspeeds they can get from each.