Nvidia’s GTX680 uses a system of ‘dynamic’ overclocking, otherwise known as ‘Boost clocks’.
This Boost clock mode uses a variety of factors to determine whether it is a good idea to run at higher speeds, or not as the case may be. It analyses power consumption, GPU load, temperature and memory load, among other factors.
The driver is coded to make on the fly decisions about what clock speed is safe in comparison with heat output and power use. The automatic overclocking algorithms need to be coded with a variety of safety parameters.
Above, the KFA2 GeForce GTX 680 LTD OC. We can see that the boost speed is set to 1,268mhz (from a default of 1,059mhz). This is the ‘average’ clock speed that the core will run under during typical gaming load. The clock speed may actually exceed this speed depending on the given situation.
For overclocking today we used MSI’s Afterburner software. We spent a long time playing with the card and analysing how far we could push it without encountering instability.
We weren’t expecting any headroom, especially as this card is supplied in such a crazy state of overclock. We did however manage to increase the voltage and push the card to a staggering 1,285mhz core and 1,658mhz memory (6,632mhz effective).
Overclocking the card even further produces a score of 11,394 points, not too far from a GTX580 SLi configuration. We can see just how far Nvidia have moved since the last generation of hardware.