We are in a period of rumors, between the Consumer Electronics Show and the Mobile World Congress. It’s just horrible to navigate the tech media, among the rumors about the latest benchmarks Galaxy S7, stolen photos of Xiaomi E5, sketches LG G5, the supposed disappearance of the headset port on the iPhone 7 … Indeed, they are often clues and relevant information that lead to debates, sometimes very interesting. But there are also less interesting things and above all really exciting, as the benchmark results.
They may be authentic, but these results are not captured less boring. It is obvious that some users worldwide are currently testing these devices that will be, for most, announced at Mobile World Congress.
Like many geeks that we are, you feel compelled to launch “a small benchmark” to your new device or installation of a new update – I have the same reflex with speedtest on networks (supposedly handled by mobile operators). Nevertheless, these are numbers that are difficult to interpret, since it is a subtle mix of parameters that can be used under predetermined conditions.
To see more clearly
Synthetic benchmarks like AnTuTu GeekBench and simply measure the raw performance of the SoC, which does not necessarily represent the performance obtained in practice. Applications benchmarks like PCMark and 3DMark or GFXBench meanwhile trying to simulate real uses such as scroll on a webpage or viewing a 3D scene, but there is a way: cheating
With these tools, we can shake the devices. They are relevant for us to check what is often observed in use – that’s why we use in our devices tests
Why we still bored.
With the Galaxy S7, we quickly saw disembark many benchmark results. The best known is that of AnTuTu, which submitted the hypothesis that Samsung had collaborated with Qualcomm to integrate the new Snapdragon 820. This architecture, we know from last year, Vincent already tested Qualcomm platforms
More on the same topic. Qualcomm Snapdragon 820: our performance test, in …
More on the same topic: 7 features the Qualcomm Snapdragon 820 that …
These famous platforms are often Development Kits, well optimized and designed to take advantage of all the functions provided by the SoC. A kind of showcase for Qualcomm. Compare these results with those that are found in the supposed Galaxy S7, contribute nothing.
Especially the devices currently on the network, such as the Galaxy and LG S7 G5, are not units finals. We are far from commercial versions that come out during the first half, especially the announcement several weeks and several months of the commercial release. Not to mention the fact that these results can be manipulated, tracks of cheating have already been identified -. As HTC One M8 and the Samsung Galaxy S4
The cheat the most common are that which consists, for OEMs, to detect the launch of the benchmark (with the name of the process) and pass the SoC in high-performance mode to increase the operating frequency, which does not reflect the “normal” functioning of the chip. HTC and Samsung have for example been caught in flagrante delicto cheat on 3DMark. It is for this reason that tools like GameBench emerged
Finally, these tests often turn in optimal conditions. Testing units are rarely used for everyday use, some apps are installed and often very few services are in the background.
Yet benchmarks have some relevance
In our tests, the benchmarks have a real relevance. We test devices with commercial versions, in the vast majority of cases. When we look to use – once our apps with our SIM card installed – a certain fluidity, or conversely slowdowns, and we call on the benchmark tools. These tools allow us especially to check our observations, such as latency observed on certain games or slowdowns observed in transition animations.
I recommend you read the excellent record of GameBench, allowing you to better identify the different benchmark tools, and utility.
In any case, the S7 and Samsung Galaxy LG G5 remain no less exciting, but the benchmark results are useless and annoying. You disagree
?
No comments:
Post a Comment