VRM and differences for overclocking

Oct 7, 2018
Can someone explain to me exactly what VRM is/does and the different characteristics of "good" vs "bad" VRM for overclocking? I've seen things about phase design and cooling, but I'm confused as to what that means.


How technical do you want it?

Basically...on a modern motherboard a CPU VRM converts 12V dc, from the powersupply, to the voltage required by the CPU. There can be several but VCore, or voltage for the processing core(s), is generally the most important. VCore can vary a lot...from maybe 0.6V to 1.5V...based on load and frequency.

Simply put, it chops up the 12 volts by turning on a switch for a tiny amount of time. That means that for a given period of time the voltage is 12V for a small bit, then 0V the rest. Over time, it all averages out to the 1.3 volts the processor needs. With just one switch it's obviously a pretty rough voltage...so VRM's have more than one. Each switch is a couple transistors called FET's. Each pair of FET's, along with an inductor to average the current, is called a phase. The more phases, the more even the switching is and so the more level the voltage to the CPU core can be kept.

Of course, it's not always 1.3V...if the processor is working hard and needs 1.35V the switches stay open longer, if it's hardly working and only needs .8V they stay open less time. The changes are fast so the system has to be designed to handle that.

That, of course, is super super over simplified. If you want something a lot better, check this page:


An overclocked processor is operating much hotter and closer to it's stability limit. That means it needs more voltage to help keep it stable. Higher voltage and frequency puts a higher current load on the VRM so it's switches have to stay open longer and they heat up and also stop switching as well so the voltage fluctuates, with peaks that go above the set voltage. Since the processor is already operating at a higher voltage than normal those peaks can actually go into unsafe levels.

There are several ways to keep it safe. One is to cool the switches (FET's) with big heatsinks so they continue to switch on and off cleanly. Another is use higher quality FET's that don't get as hot during switches. Another is simply have more FET's...either more FET's in each phase or, even better, more phases. And of course the best way, and you see this in expensive motherboards intended for extreme overclocking, is to do all three.