Investigation firms discovered OnePlus and Meizu cheating their benchmark scores to trick users and critics. Benchmark scores help users make informed buying decisions, and without the truth, more and more consumers will be purchasing the gadgets that don't really fit their needs and wants.

An XDA Developers investigation, working with Primate Labs, discovered a conspiracy that goes deep in the benchmarking infrastructures that tweaks the scores of Android devices. New reports reveal OnePlus and Meizu cheating the benchmarking scores of their OnePlus 3T and Meizu Pro 6 Plus a little bit higher than what they truly got, NDTV reported. OnePlus admitted the accusations and vowed to fix it, but Meizu declined to give any comments on the issue.

How It Affects Users' Trust

Fairness is important in an the Android device industry, where benchmarks give users an informed buying decisions on what devices fits their preferences appropriately, Android Central reported. With this controversy, it would be hard for users to trust these benchmark scores. Experts advise that it is still best to not just rely on these scores, but check the entire device when deciding which device to buy.

How They Cheated

XDA Developers and Primate Labs obtain a certain Geekbench that triggers cheating, which helped them discover big performance differences resulting to higher overall temperatures. This was the same results that helped them uncover OnePlus and Meizu cheating their benchmark scores. These phone makers activate performance modes on their products as they enter CPU and GPU benchmarks.

According to the investigators, OnePlus targeted benchmarks by name entering a different CPU scaling mode to tweak their benchmark scores a little higher. Reports claim that OnePlus tweaked its CPU governor code into something more aggressive turning it into an artificial clock speed in Geekbench that originally wasn't there in the hidden Geekbench build. It was based on the package name of the app and not on its CPU workload to fool the hidden build.