Usually synthetic benchmarks are considered a good way for comparing performance of various devices, but the benchmarks can be misguiding and affected.
Recently a user on Beyond3D noticed that the international version of Galaxy S4 with Exynos 5410 Octa processor runs at maximum speeds when popular benchmarking apps are launched, which makes them reflect better scores. AnandTech validated the claims and they are indeed true.
The GPU of the Galaxy S4 runs at 480MHz during normal operation, but when certain benchmarking apps ( AnTuTu , Quadrant, 3DMark Linpack, Benchmark Pi ) are launched, the firmware boosts the GPU clocks to 533MHz which is over 10% higher , this results in higher scores compared to what the device would normally show.
Other benchmarking apps like Epic Citadel and GFXBench 2.7.0 are not white-listed and are only executed with 480MHz clocks on the GPU.
The CPU follows the same optimizations and the Cortex A15 cores are made to run at maximum 1.2GHz clocks even when the app is launched but in idle state.
This means that it would not be fair to judge the real world performance of the device based on these synthetic benchmarks as they are influenced by Samsung's tweaks in the firmware. While Samsung may call it optimization to make the smartphone perform better in benchmarks, it is misleading users with affected performance scores.
Recently a user on Beyond3D noticed that the international version of Galaxy S4 with Exynos 5410 Octa processor runs at maximum speeds when popular benchmarking apps are launched, which makes them reflect better scores. AnandTech validated the claims and they are indeed true.
The GPU of the Galaxy S4 runs at 480MHz during normal operation, but when certain benchmarking apps ( AnTuTu , Quadrant, 3DMark Linpack, Benchmark Pi ) are launched, the firmware boosts the GPU clocks to 533MHz which is over 10% higher , this results in higher scores compared to what the device would normally show.
Other benchmarking apps like Epic Citadel and GFXBench 2.7.0 are not white-listed and are only executed with 480MHz clocks on the GPU.
The CPU follows the same optimizations and the Cortex A15 cores are made to run at maximum 1.2GHz clocks even when the app is launched but in idle state.
This means that it would not be fair to judge the real world performance of the device based on these synthetic benchmarks as they are influenced by Samsung's tweaks in the firmware. While Samsung may call it optimization to make the smartphone perform better in benchmarks, it is misleading users with affected performance scores.
