Today’s tutorial: How to rig Android benchmarks – Courtesy of Samsung and HTC

How do you make your handset stand out in today’s highly competitive market? Rig your benchmark results of course! That way - on paper at least - your latest smartphone appears to blow away rivals.

Should you have any moral issues? Not since it appears to be the industry standard.

It's undeniable Samsung has driven innovation in processors; for years even supplying them for Apple’s range of iDevices.

The Exynos 5 processor is an Octa-Core, but doesn’t use all eight cores at once. Four are high-power, and four are used for power-saving when under low intensity applications. In order to gauge the performance of these processors; reviewers look to benchmarks including GeekBench and SunSpider.

With the myriad of devices available - especially on Android - many consumers simply look at raw scores; heading towards the highest performer. You don’t have to be a marketing expert to know more sales, equals more cash - and every company likes more cash.

This whole “benchmark boosting” fiasco started on the Beyond3D forums. A member posted how the benchmarks show much higher than real-world tests in apps and games; causing website AnandTech to further investigate.

AnandTech's Klug and Anad Shimpi noted: "Running any games, even the most demanding titles, returned a GPU frequency of 480MHz." Whilst the GLBenchmark software –used to benchmark speeds – triggered a GPU clock that ran at 533MHz.

On further examination; they found software code hidden which would boost the results when benchmarking software was used, giving an unfair (and unrealistic) advantage.

This would explain why reviewer Brian Klug, before these revelations, posted: “The Galaxy S4 manages to outperform the HTC One by around 17% here. Again, it's unclear why we're seeing greater performance than clock scaling alone would provide, but the net is that the Galaxy S4 does deliver better GPU performance than other Snapdragon 600 based devices today.”

But, is HTC immune? Will owners of their flagship One smartphone be sitting smug?

Nope. Over at xda-developers (have to love these forums!) a member found a library in HTC Sense which will bump the CPU to max speed when a benchmark is found running. It doesn’t look quite as shady as Samsung’s, but still questionable tactics.

What do you think about these major manufacturers rigging benchmark results?

Related Stories

Leave a comment


This will only be used to quickly provide signup information and will not allow us to post to your account or appear on your timeline.

31 Jul 2013, 6:41 p.m.

Triggering the SOC to Tim at 100% isn't rigging since the SOC naturally will run that fast if needed. It's a benchmark you want it at 100% you don't benchmark at 80%. But boosting it past where it would run naturally is just plain cheating.