News: Samsung fakes the Note 3’s performance too




News: Samsung fakes the Note 3’s performance too. You may remember a little while ago we reported on the discovery that Samsung had set up the international variant of the Galaxy S4 to perform better when a benchmark test was detected. Well, it seems they haven’t learned anything from the experience and are at it again, this time with theNote 3. Basically it’s like saying your new car can do 300mph, but only if it’s going down a specific hill in France and Michael Schumacher is behind the wheel. On a Sunday. In a wedding dress. Sure, it’s technically possible, but it’s just not gonna happen for most of us.

galaxy-note-3-test-6

                  Don’t believe everything Samsung tells you. 
While our reporting of the ”cheating” last time raised some mixed comments – the process is, after all, common-place in desktop PCs – there is a fundamental element of deception at play. By making a device only perform at the highest level when someone is looking, so to speak, and never during normal usage, you are essentially inflating the actual performance of the device. While many readers seemed to have no problem with this, because it simply shows what the device is capable of, it deludes the average buyer because that performance is unattainable.

samsung-galaxy-note-3-ars-technica-benchmarks-03

Multi-core benchmark boosting improves performance by 20%./ © Ars Technica

The devices in question, the S4 and Note 3, are coded to detect when a benchmark application is running, though not all benchmark tests surprisingly, and if a benchmark application is active they ramp up the processor to have all cores firing at the highest possible level, even when the device is idling. This warps the benchmark by making it look as though the device is super-powered all the time, when in reality the device only uses multiple cores for complex processing, and never uses all eight cores. At most it will use four, and when idling only one; except when the benchmark is watching. We’ve reported previously about true octa-core performance, which ironically, will be coming to the Note 3 soon enough.

samsung-galaxy-note-3-ars-technica-benchmarks-04

While the results are more marked on multi-core tests, the same is visible with just one.  / © Ars Technica

While the device actually is capable of this performance in practice, it is only capable when a benchmark test is running. No user will ever get that level of performance out of the device, which to me at least, is false advertising, especially when other manufacturers use real scores based on actual performance. At the time of the S4 revelation, Samsung stated on their Korean website: ”[We] did not use a specific tool on purpose to achieve higher benchmark scores”, and “under normal conditions, the Galaxy S4 operates up to 533MHz at its best performance.” This is a downright lie, as the code uncovered by AnandTech specifically stated ”Benchmark Booster” and the S4 never operates at 533 MHz except under specific benchmark conditions.

The Note 3 has been doing the same sneaky test fixing. Ars Technica suspected the same scam and ran various benchmark tests, but with the clever addition of renaming them so the Note 3 wouldn’t put itself into overdrive for the duration of the assessment. The results, not surprisingly, are that the Note 3 posted crazy high scores, much higher than the LG G2 which has the exact same processor, when the benchmark test was detected and then when the same benchmark test was renamed, performed as we would expect it to, much like the G2. Which is exactly how every single owner of the Note 3 will experience it.

samsung-galaxy-note-3-ars-technica-benchmarks-02

Ars Technica renamed Geekbench to Stealthbench and got very different results. / © Ars Technica

Perhaps you don’t have a problem knowing the performance that is possible from your device even if you can never actually access it, but for me personally, I’d prefer to know what my actual usage of the device will yield, not what is possible in a perfect world under perfect, and useless conditions that will never actually benefit me. This is the first time the benchmark boosting has appeared on a US model, with the S4 in question being the international model. We always advise readers to take benchmark tests with a pinch of salt, but some still see them as an objective measure of which device is more powerful. So if you haven’t learned your lesson yet, now is the time. Hopefully Samsung will learn in time, too.