Originally Posted by steelbom
That's true, but those things that you mention would effect performance by a few percent at best, and if you look here
at the benchmark (scroll down) you'll see the differences are actually 20%.
It's the fact that it's in a benchmark that makes it cheating. (Or at least deceptive.) Isn't the whole point of the benchmark to give you an idea of how the device performs? If it does these "optimisations" then all you see is performance that can never be achieved in the real world. And it will make that smartphone look more powerful than other smartphones with similar processors. And then for those smartphones to appear to have similar performance, they also must do this as well. And each year new targets will be optimised for. Now when you go and compare performance across devices in a period of a year or two (or worse, even longer) it's going to be an incredibly difficult picture to read.
Futuremark themselves say "don't do this with our benchmark." It's against their... uh... ToS.
It's not cheating if there's a real benefit to it. If they did this for games, then that would be completely legitimate. But the thing is -- these "optimisations" can't possibly be sustained in an actual game. There's no way these devices could handle having all cores plugged in at max frequency (and the same for the GPU) for an extended period of time.
If these improvements were brought on by better drivers or software, like with graphics drivers on the PC from NVidia or AMD, then I'd welcome it! But it's not.
But they didn't tell anyone what they did, allowing people to be misled into thinking that said smartphone is more powerful.
I hadn't seen those geekbench scores. However, that's one benchmark. The rest were ~5 as measured on Anantech.
Benchmarks exist to measure the performance of a phone during normal usage, and a device should never treat a benchmark app differently than a normal app
Is flat out wrong. Benchmarking exists to show off your device. Android user on XDA bench using a similar technique of putting the governor to "performance" which does exactly the same thing. Yes the idea of a benchmark is to let you know how a device performs... in a benchmark. So pushing the cores to 100% is just fine. As I mentioned, XDA users do this as a rule. You benhcmark your device, you change the governor to performance.
Clearly, allowing the CPU to idle during benchmarks affects benchmark performance. It wouldn't reasonable to do this in apps, because apps don't need 100% core performance for the full amount of time it's in focus. It wouldn't be reasonable to do it in games, because they would overheat the device after long use and cause it to throttle.All this is showing is that if the device runs at 100% of it's ability, these are the numbers. How is that cheating a benchmark?
The optimisations are showing that it's better than other devices. The device sin that list could just as easily do the same thign by setting their governor as I mentioned. They'd give the exact same effect. Their scores would be higher than not using it.