In this model, the voltage alternates from 0 volts up to the vCore value and back again each clock cycle. You’ll notice there is an instant transition from 0 Volts to the vCore value. In the real world, this doesn’t happen. Think of a garden hose, you could open the valve up to full to allow all the water through, but you don’t get maximum flow right away. It takes a second or two to get all the pressure and all the flow. The same thing holds true with voltages. Let’s see what it looks like graphically.

Here’s a more realistic example of what the voltage would look like. As the voltage changes, there is a quick peak and then it levels off at vCore (and conversely a drop before returning to 0 Volts). Now you can start to see why tolerances are in place. If the voltage is registered before it completely levels off, at least it will still register as a logic high or low as appropriate.

So, why do we need to know about this when it comes to overclocking? Well, let’s take the example above, but we will increase the frequency of the rise and falls to simulate an overclocked processor.

Anyone else see a problem with this? Because the clock cycles have increased in frequency, the voltage never gets a chance to hit the true vCore value before the new cycle starts. A minor overclock might not be too bad, since it would still more than likely peak close to the tolerance range, but the faster the processor is run from stock, the more likely the peaks won’t fall within tolerances. How do we correct this problem? Make chips with higher tolerances you say? Unfortunately, with current electronics technologies, tolerance levels are a relatively small band, so it’s impossible to set a tolerance for a 2 Volt gate, for example, from 1 to 3 Volts. A range of 1.8 to 2.2 Volts is a more realistic tolerance level. So, if we can’t adjust the tolerance, how can we fix the problem? Increase the value of vCore. Let’s take a look at stock speeds:

If we adjust the value of vCore in the BIOS, what we are doing is telling the system to output a higher voltage. The system now peaks and stabilizes at a value higher than what the chip actually needs to recognize a logic high. What is of note here is that the voltage now has to "jump" higher in the same amount of time. Now, watch what happens in the overclocked scenario:

In the overclocked scenario, the cycles happen too fast for the voltage to reach the new higher vCore value. However, the voltages are now peaking within the tolerance range of the actual vCore that the chip was designed to operate at, so the chip will interpret these peaks properly as logic highs.

Increased vCore Limitations & Hazards

Armed with this knowledge should your next step be to go into the BIOS settings and select the highest available core voltage setting the motherboard supports? No, and for two reasons:

1. Increasing core voltage drastically increases the amount of heat the chip generates. Do not attempt to increase core voltages unless you have sufficient cooling to compensate for it. Depending on how much of an increase is made with vCore, the thermal load may represent more than a standard air cooled heatsink can handle, requiring water cooling or more extreme cooling methods in order to keep the system in check.

2. Core voltage increases have a point of diminishing returns. While you might be able to vastly increase system speeds by applying some additional voltage, the more voltage you add doesn’t equate directly to faster system speeds. For example, a .2 volt increase on an Athlon XP processor may allow it to be overclocked 50% faster than stock. Applying an additional .2 volts, however, may only see system overclock speeds increase another 10% (assuming you had an extreme cooling solution to keep from frying the chip that is).

Conclusion