NVIDIA’s Kepler-based GPUs Rumored to Feature Dynamic Overclocking

Posted on March 12, 2012 10:30 AM by Rob Williams

It’s being rumored that NVIDIA’s upcoming Kepler-based graphics cards will feature dynamic overclocking, much like Intel’s ‘Turbo’. On Intel’s side, if a particular process needs a little more oomph, Turbo kicks in and gives certain cores a clock boost. NVIDIA’s implementation will almost be a verbatim replica of Turbo, as once a game can take advantage of additional power, the GPU will get a clock boost of around 5~7%.

As someone who doesn’t particularly care for GPU overclocking, this is a feature I’m looking forward to – and to be honest, I’m a little surprised that it’s taken so long to come to fruition given just how overclockable current GPUs are. That of course means that it’s highly likely we’ll see a similar feature in AMD’s next-generation GPUs, Radeon HD 8000, or whatever they might be called.

The way NVIDIA will judge whether more performance is needed will be based on a power limit, and it’s mentioned that not all games will cause this turbo mode to kick in. As an example, a game like Minecraft might not require any sort of boost, while Metro 2033 no doubt would.

How this dynamic overclocking will affect real overclocking is at this point in time unclear. It can be assumed however that Kepler will be heavily overclockable, as the 5~7% clock boost NVIDIA provides its turbo mode is likely very conservative. Kepler might be one of the most overclockable architectures ever released. Let’s hope so.

Source: Heise (German)

Recent Tech News
Recent Site Content