Sandybridge and Electromigration?
Well since I switched over to water, I've been experimenting with overclocking above my usual settings. However, electromigration can happen if I go above 1.4V on my chip. Does this mean my max voltage would be at 1.4V regardless of the cooling I'm using? I really don't want to damage my chip at all, but I want to go higher in my overclocks. Right now I can get 4.8ghz stable on my i7-2600k with exactly 1.4V, but does that mean I can damage my chip because I am running it at 1.4v?
Cooling does not stop electromigration. It occurs to some extent in the cpu at any voltage. The higher the voltage the greater the effect. Most who try for extreme OC do not care about cpu life. So what is going on is that as you raise the voltage you shorten the life of the chip. You will not damage it as long as you control the temps. UP TO A POINT. There does come a voltage level where the chip fails. Also though you cool the chip at the heat spreader, high temps still are generated in the silicone. Cooling does not stop the heat from being produced it just carries it away form the chip as fast as possible. Under high voltages the transistors produce so much heat that the cooling solution becomes inadequate. The chip itself may also not be able to transfer the heat to the cooling quick enough.
Starting out with suggestions and ending up with the actual bottom line, so here goes.
If you are just short term curious as to where you can take your 2600K do not exceed 1.520v to the CPU this would be more for bragging rights accomplishing a CPU-Z validation and some benchmarks for the 50X or so range.
For longevity and 24/7 daily use I would suggest you settle on a CPU clock that does not exceed 1.40v, Why? Because it would be even better longevity wise to stay below 1.40v, so if you can run 4.8ghz at 1.40v exactly, that would be an excellent clock for you to run, if that is what you want to do.
It's your hardware to do with as you see fit!
It's not just the 1.40v range that was deemed important it was also the additional heat the extra voltage brought to the table, and most of those past 1.40v limitation claims were using air cooling at the time, you now have better cooling than that.
It also matters exactly how you are overclocking in the first place too, as to whether you're controlling what your voltage is doing in the operating system, or it's out of your control being controlled by the CPU/MB.
Fixed voltage gives you better OP/SYS control of the voltage to avoid voltage spikes that would exceed your BIOS set voltage range.
Just for sheer curiosity I tested my wife's 2550K at 4.0ghz with the voltage on auto, and was shocked to see OP/SYS spikes to 1.520v at only a 4.0ghz clock, that should not be happening, but it does, and is reproducible results.
I've had my 2500K now for 16 months + have run fixed voltage at times all the way to 1.520v it runs as good today as when I first bought it, your 2600K can run the same clocks as my 2500K using less voltage because 4800mhz for my 2500K takes 1.415v, with SLI disabled and 1.425v with SLI enabled.
My suggestion is simply do what you want high clock wise to resolve your curiosity, then for longevity drop down to about a 4600mhz clock, that will even bring you under 1.40v and put you in an even safer longevity zone, you can always clock back up to whatever you need, save those overclock settings in your profile.
It's just not necessary to run those higher clocks daily.
The cooler you run it, the longer it will last, there's not that many applications that will take advantage of higher clocks anyway so why run it like that 24/7, test the performance for yourself and see if you think it's worth going past where you are now.
You'll get benchmark improvements sure, but you don't run benchmarks all day, they're basically for bragging rights.
Your 2600K has a 3 year warranty, if Intel was really worried about electromigration they surely would have only warrantied it for a year definitely not 3 years.
Additionally they would not have given motherboard manufacturer instructions of a 1.520v tested limit posted at the Intel website, which is referenced in my overclock guide.
I cannot speak for the rest of this place, but for myself, I have never bought and used any setup expecting to be using the same setup for 3 years, technology just changes too fast to be expecting to even be in the ballpark 3 years later.
So do what you feel comfortable doing amuffin, just stay below 1.520v even OP/SYS spiking and you'll still be below Intels warranty requirements. Ry