A Samsung Electronics' Galaxy S4 (front) and Apple's iPhone 5 smartphones are seen in this picture illustration taken in Seoul May 13, 2013. REUTERS/Kim Hong-Ji

Investigation by a tech website supposedly reveals lines of code used to manipulate benchmark tests

Samsung have denied recent allegations that they tricked benchmark tests designed to evaluate the performance of their phones.

An investigation by website AnandTech accused the company of altering the code in their flagship Galaxy S4 handset so that its hardware would run faster when running specific benchmark apps.

These apps are designed to test how fast processors and are other computer innards are, and their results are used to evaluate and sell hardware. In the case of the Galaxy S4 such benchmark tests are especially important as Samsung do not offer any official clock speeds for the phone’s processors.

The site used various benchmarking tools including GLBenchmark and Quadrant that recorded processor speeds of 533MHz – an 11 per cent performance boost on the usual 480MHz. When benchmarking software was loaded onto a US model of the Galaxy S4 the CPU rocketed to its highest performance level, regardless of the strain on the device.

Samsung have denied any “specific tools” used to achieve higher benchmark ratings, and sought to differentiate between apps which run full screen and obscure the status bar and those which do not.

They claim that the former applications “require the highest performance” and so demand different clock speeds, thus potentially explaining the different clock speeds.

However, this does not address the strings of code found by AnandTech that seem to indicate specific benchmarking apps being marked out to receive higher performance speeds. AnandTech themselves note that whilst Samsung's cheating could undermine objective tests of phones, the applications that were being tricked are themselves out of date.