Common sense would tell you not to enhance your riding lawn mower's motor so it goes 190 mph, right? Well, sometimes that common sense doesn't follow over into the computer world. The once garage hack is quickly becoming a technological hobby and even art form. Overclocking refers to a term that means taking a processor or other component and making it run faster than it was originally designed to do. In this article, I'll explain how it works and why it might not be the greatest idea.
Of Mhz and Ghz
Intel or AMD engineer CPUs to run at particular frequencies. These frequencies, which like radio frequencies, are referred to in terms of megahertz (Mhz) and gigahertz (Ghz) control the number of cycles a CPU uses to process information. Generally speaking, a CPU with a higher clock rate will perform operations faster than another one. This is especially true within CPU families. So for example a 3.0 Ghz Pentium 4 Intel CPU will perform faster than a 2.8 Ghz Pentium 4 Intel CPU.
As a side note, other factors control how fast a CPU processes information such as internal caches, chip design and so forth. Therefore clock speed may be misleading. For example, an AMD CPU may only run at 2.2Ghz but still outperform a 3.0 Ghz Intel CPU. For the purposes of this article, we're not going to discuss or look at these factors. I'm only going to concentrate on clock speed.
When Intel and AMD create CPUs, they don't have assembly lines dedicated to specific CPU running at specific frequencies. They don't create 2.2 Ghz processors one day and then 3.0Ghz ones the next. Instead, the silicon is all created at the same time and then after the chip creation process is finished, the chips are tested at different speeds.
Finished chips are sorted based on how they perform on these tests. CPUs are never sorted by their maximum performance, but rather by a performance level that won't stress the CPU that's well below the chip's maximum level. That ensures that the CPU won't fail over time. For example, during testing, a chip may start showing stress at 3.0Ghz. Based on a formula, the maker may label and ship this chip as a 2.2Ghz CPU to ensure that it works properly.
That's where overclockers come in. Knowing that any given chip is running well under it's maximum capacity, overclockers make changes to systems to push a chip more towards its theoretical limits.
Doing so has some danger. However, when there's a price difference of several hundred dollars for essentially identical silicon under the case, not to mention the geek-chic of it all, it can be attractive to overclock a system.
In the old days, motherboard frequencies were controlled by a discrete clock chip. It controlled the frequencies for the entire system. Increasing the speed of a motherboard was no harder than going down to Radio Shack and purchasing a new chip. You'd pull out the old one and put in the newer, faster one. When you rebooted the computer you were off.
As motherboards and CPUs became more sophisticated, so did the ways you'd have to overclock a CPU. Sometimes you'd have to move jumpers on a motherboard. Other times, you'd only have to make some changes in BIOS.
When you overclock a system it makes the chip run hotter. This is where the real danger comes in with overclocking and what usually causes a CPU to fail. It's not so much the speed of the data going through the CPU as it is the heat generated. CPUs which are shipped at lower speeds generate too much heat to run at higher speeds. In the old days, you could solve this heating problem with a simple CPU fan or heatsink. That doesn't work as well anymore as CPUs are coming from the factory with fans and heatsinks already.
Additionally, Intel and AMD would prefer you pay those several hundred extra dollars for the faster CPUs, so they're doing some things that are making it difficult for overclockers to work their magic. There are still some ways to get it done however.
Overclocking in the 21st century
Because heat is still a problem with overclocked CPUs and basic cooling won't help, new ways of fighting it in overclocked CPUs are popping up all the time. Some of them involve bringing back technology from the mainframe days such as water-cooling. Other companies offer even wilder technology such as using dry ice or liquid nitrogen. Even other companies have created CPU chillers that immerse the entire computer in an inert fluid that dissipates heat.
Although the thought of running tubes of water through your computer may run chills thru your spine, it really does work. Not to mention it does look pretty cool.
Beyond the challenges of cooling, there other issues to face when overclocking a system. Most of these revolve around voltage issues, as well as specifics about each CPU. Unfortunately because CPU families change so rapidly, no one single article will tell you how to overclock your specific system. However, there are sites on the Internet that you can visit and search for those specifics.
Reasons not to overclock
Overclocking may sound like a good idea, but it isn't necessarily. For example, one of the main reasons people use to justify overclocking is the difference in price between CPUs. That may be true, but you may wind up spending as much or more in components to cool and make the system work at higher speeds than what the price difference was in the first place. Also, don't forget that CPU prices fall like rocks. You might not be able to afford a certain speed today, but that same speed may be much cheaper in just a few months.
Secondly, even though the CPU should work at the higher speed, overclocking may cause processing errors or shorten the life of the CPU. As with most things, when you tinker with a system's CPU, you pretty much be assured that your warranty will be voided.
Other major drawbacks to overclocking include:
- Higher electric bills due to having to run systems at a higher voltage
- Increased noise of cooling components
- Failure of associated components such as RAM that can't keep up
- Increased maintenance of overclocking components