As per a recent report, the much-coveted Samsung Galaxy S4 shows off enhanced performance when being tested through a handful of benchmarking apps, holding the company responsible for this tampering. It seems that the tools falling on this list by Samsung include GLBenchmark 2.5.1, AnTuTu, Linpack, Quadrant and more.
From what can be gathered, the smartphone’s GPU as well as processor sprint at higher clock speeds while running the apps outlined above. For instance, a post on AnandTech has pointed out that the frequency of the Exynos chipset-driven model’s GPU scaled up to 533MHz when GLBenchmark was used to analyze the device, whereas it wouldn’t cross 480MHz in the case of other tools.
As for the CPU, the phone employed its Cortex A15 chipset flashing a speed of 1.2GHz as soon as the GLBenchmark service kicked off and switched over to the Cortex A7 cluster with a frequency lower than 500MHz when used for testing through GFXBench 2.7. The aforementioned article also spoke of similar aftermath for the Qualcomm processor-powered variants wherein the phone’s performance was maximized while running the apps mentioned above.
What currently has caught everyone’s attention is that a string of code has pointed in the direction of a certain BenchmarkBooster program within the Samsung Galaxy S4, which is probably the root cause of the varying performance. The South Korean manufacturer has brushed off these allegations through a blog post today, but conveniently didn’t address the existence of the specified code.
All it said to clarify itself is that the GPU’s frequency which scales up to 533MHz under normal conditions is reduced for some apps running in full screen, to prevent an overload. However, when used for regular functions like camera, video player and certain benchmarking apps, it delivers maximized performance.
Considering the scenario, it becomes a little difficult to tell whether the company is indeed blameworthy for messing with the Galaxy S4 benchmark readings and if yes, it certainly couldn’t be the only one.