Well, here we go.
Over at the IMP3Net forum, it’s been revealed that the typical speed of the Rockchip 3288 is 1.2GHz, not 1.8GHz [Google Translate]:
According to his [user 2222’s] actual [AnTuTu score], 3288 actual frequency 1.2G, even the big game occasionally 1.5G, overclocked to 1.8G entire security Bunny
In other words: A user with the handle 2222’s AnTuTu score — which was just 30,000-range — revealed that the 3288 achieves its full 1.8Ghz speed only when AnTuTu is running. It generally throttles down to 1.2GHz and will occasionally boost to 1.5GHz as needed for games.
So where do we really go with AnTuTu scores or any CPU benchmarks? Even the K1 in the Xiaomi MiPad will throttle back when it senses the temperature is getting high.
Do we measure by peak frequency or by typical frequency? And is peak frequency the actual peak — or just one momentarily used to fool AnTuTu?
How is it then possible to compare the iFive Mini 4 to the Xiaomi MiPad — two machines that, until now, were thought to have very similar overall AnTuTu scores?
Mike Cane’s xBlog 
Antutu scores have no real life impact anymore.
Most if not all of the new CPU/Chipsets/Devices provide an extremely smooth android experience.
I’m an advocate of just simply judging the Random I/O R/W speeds and the GPU scores. Rand IO affects performance. GPU affects gaming performance.
I guess that’s what A80T is doing as well, does not explain the 60K+ score completely, but I guess it’s a mix of this + the trick to get 20K+ on A31.
It’s sad, I think they are going to lose the trust of their customers by doing this, true performance of those chips is perfectly fine (currently 30K is excellent for budget tablets),
if they spend their time optimizing their android build and cleaning their roms of all their bloat they would get real better antutu score and more sale.
Dear chinese SOC makers you are digging your own grave.
It is not actually cheating
So, Antutu is measuring a real hardware in a fare manner
If you would want to use the tablet on maximum frequency, just root the tablet and use e.g. SetCPU
but your tablet would be horribly hot
so, benchmark measures the maximum result possible and it is fare
in comparison with Cube’s cheating, when they were scoring some unbelievable scores with an outdated Mediatek
but still, xiaomi mi2 with its s4 pro and 20000 and antutu just beats any “non-brand” chinese tablet in terms of smoothness
We need a measurement that shows typical use, not maximum use which is possible only when being benchmarked.
@Filipp, Yes. It’s called cheating because under real world usage, it supposedly never reaches 1.8ghz.