Intel granted overclocking blocking patent

Intel has been granted a patent outlining a technology designed to block attempts to over-clock its processors.

Intel applied for the technology to be patented in September 1999. The patent, number 6,535,988, was granted on 18 March this year, as reported by HardOCP.

The patent covers "a mechanism for detecting and deterring over-clocking of a clock signal for use in a processor, comprising: a detection circuit adapted to detect over-clocking of a clock signal for use in the processor based on a reference signal; and a prevention circuit adapted to prevent over-clocking of said clock signal by limiting or reducing performance of the processor in response to detection of said over-clocking of said clock signal".

In short, the chip dynamically compares its current operating clock speed to a reference signal, generated by an oscillating quartz crystal, and automatically reduces its clock speed if it finds it's running faster than it should be.

Interestingly, Intel didn't have enthusiast overclockers in mind when it filed the patent application. Its documentation refers to "resellers and/or distributors remarking processors at higher frequencies and then selling the processors as the higher speed part to charge for resale at higher prices".

Of course, Intel only has itself to blame. It admits in the same patent documentation that "processor manufacturers may be very conservative when rating such a clock frequency. For example, a processor which successfully operates during tests at 333MHz may be only intentionally rated (marked) at only 133MHz, 150MHz, 166MHz, 200MHz or 250MHz for different market reasons".

By the way, if those frequencies were low, don't forget that this stuff was written when "speeds of host processors can vary from 66 MHz to about 500 MHz".

HardOCP notes that an Intel spokesperson told them: "We have not made an announcement on any kind of implementation of this."

In other words, Intel may well be using this technology - it just hasn't formally admitted the fact.

It's certainly been a while since we heard of anyone passing off low-speed Pentiums as higher-speed parts, and the presence of the technology outlined in patent 6.535,988 may be why. But we suspect not - particularly since Intel appears to have a more lenient attitude to overclocking these days, if rumours surrounding the upcoming Springdale and Canterwood chipsets are anything to go by. ®

Related Story

Intel Canterwood, Springdale to be easier to overclock?

Related Links

Intel's patent, number 6,535,988
The HardOCP story