AMD is working on dynamic overclocking of graphics cards

With the launch of the Geforce GTX 680, Nvidia introduced a technology to automatically adjust the clock frequencies according to load. GPU Boost uses dedicated logic to read the power consumption and temperatures every hundredth of a second, to see if there is room to increase the clock frequencies (in steps of 13 MHz).

The technology is reminiscent of similar features in x86 processors from AMD and Intel, with the important difference that Nvidia’s technology lacks a fixed ceiling and thus can raise the clock frequencies significantly, if the conditions are right.

AMD’s counterpart came with the Radeon HD 7970 GHz Edition, which delivered Powertune with Boost. The technology is different in its design and gives all applications access to a maximum clock frequency, in this case 1,050 MHz. The drivers regulate it all and set it to lower frequencies if the graphics processor’s temperature or power consumption risks becoming too high.

Techpowerup is now reporting on technical documentation from AMD, which suggests that the Radeon maker should introduce a more potent variant of the technology. The document contains two designations that indicate whether the graphics processor’s or memory’s clock frequency can be changed, and a third that measures the activity and load. These are three essential ingredients for dynamic overclocking.

In practice, this means that graphics processors from AMD, just like Nvidia, could offer maximum available performance while keeping the graphics card within exposed power consumption. However, it is unclear when the technology will be sold in commercial products. A possible alternative is at the launch of the next generation Radeon.