Samsung, HTC, LG and ASUS Caught Tempering Benchmark Apps to Appear Better

BenchMark Apps Tempering

Way back in July, Samsung was caught rigging benchmark scores for its Galaxy S4 flagship smartphone.

Three days ago, that happened again, this time in the case of the Note 3. That, coupled with it getting caught posting fake reviews on HTC’s user reviews page must’ve tarnished its image to a lot of extent but now, its looking like the other companies aren’t any good either.

According to a new and detailed report by AnandTech, companies including HTC, LG and ASUS are doing exactly the same with their flagships.

How is it done? You see, different benchmark apps (majorly for Android iOS) are available at Play Store to demonstrate how the investigated Android device will perform in varying conditions. In case of the Note 3 as soon as the benchmark was started, the phone’s 4 cores kicked in at a full speed of 2.3Ghz which gave it a 20% speed boost and helped it thrash the similarly-specced LG G2.

When those culprit lines of code were identified and were removed from the benchmark app, the result was considerably less overwhelming then before.

This new report, however, identifies even more phones including the LG G2 (yes, no one’s the underdog here), HTC One, One Mini, ASUS Padfone Infinity, Galaxy Tab 2 10.1 and the new Galaxy Note 10.1 2014 as the culprits.

AnTuTu and Vellamo were the apps which saw most tempering attempts, something expected given their acclaim.

The only companies which emerged as the innocent were Motorola, Apple and Sony that didn’t indulge themselves in such practises at all.

Truth betold, not many people base their judgement of a product on benchmarks solely but still, fooling users and critics alike into believing that a product is better than it is isn’t a good thing to do, by any means.

via Ars Technica and AnandTech


  • Kisi zamany main ye harkat, 1 pathan ne Sohrab Goth main 586 Laptop ko P4 bana kr 1 factory k owner ko Rs. 7500 main sale kya tha. Lagta hai wohi pathan, in companies main lag gya hai. Ha Ha Ha. Thank God, I have Sony Ericsson Live with Walkman.

  • ProPakistani, please verify before you publish. You’re turning into a tabloid. it’s not possible to fool benchmark apps in the way you mention, at least. I don’t believe this one bit. It’s just an attempt to fool people even more! Where were the ‘lines of code’ that were tweaked? In Android? That’s not possible because it would be impossible to write several dozens of different kernels for different phone models. Manufacturers would incur losses on the assembly line. In the benchmarking app itself? Then there is no unfair advantage because all phones would run the same code. Are you indicating that the phones somehow have processors written for those benchmarking apps? Ridiculous!

    • Well this information is correct. CPUs and GPUs don’t always run at their 100 percent. They vary their performance by the requirements of the process being run. E.g. games are allocated a limited amount of RAM and CPU. This helps in running mamy processes simultaneously. What these manufacturers are doing that they program their devices to recognize the names of the processes being used by these benchmarks. So when these processes are running, the performance of the device is cranked up to 100 percent. This is cheating because this doesn’t happen in real life scenarios, so the results in these benchmarks have no real impact in regular usage of the device. When the benchmarks are made to run with different process name, the results are pretty different.

    • Search on internet its viral every tech blog has published it and has been varified by experts, its not fake. .

    • In case you don’t know, we weren’t the ones who did this research. We’ve only covered this “news”.

    • ahaha “line of code” here is a programmer ;)
      btw each n every phone has a different kernel according to phone’s hardware requirements and drivers etc
      though you’r right its not called cheating its called selling your product with a better look, Like we flash custom tweaks n scripts to get some better performance, they are jux using it in stock….
      “Antutu benchmarks”
      see yourself
      does your 1.2 ghz processor shows 2.ghz in benchmarks No ? wel then no one is hacking!!

      • Dear xda_devtech
        What these companies do is that the kernel recognizes the app name (com.google.search etc) and overclock up the cpu to 110-120% as you know that all processors have some headroom in the maximum allowable frequency but is normally not used for safety reasons. (My old Desire S had stock freguency of 1.0Ghz but ran stable at 1.2 – 1.3Ghz and thus could potentially beat all 1.0Ghz based devices)

        In these cases the kernel has a list of app names for which to do this so for all other apps, the maximum frequency you can get is 100%.
        When the benchmarking software asks the processor frequency, it reports 1.0Ghz Dual Core (for example) but when the benchmarker asks the processor to do some work it actually overclocks the processor to 1.2Ghz so that the workload finishes early and the device gets better scores.

        The problem is that the kernel does not allow external apps to overclock the processor to 1.2Ghz nor does it when it is under load but none of the apps in the secret list are running.

        How the researchers figured out the cheating is that they changed the name of the app in the manifest (so com.google.search becomes com.google.newsearch) and the rest of the code remains the same. Now since it is basically the same app, it should post similar benchmark score, but since it is now not on that secret list, the processor is not overclocked and it posts much lower score.

        So if you are really a programmer and not a script kiddie, you would know that you do not need to reprogram the kernel for each phone, just a couple of lines in the application to see if the phone from that big name company is cheating…

  • why to test your phone with benchmark apps test it your self you will see the diffrence i m not defending any brand because i hate some of these brands but you will clearly feel the difrence are they fooling other apps also if not then why perfomance differs on every phone innocent heehheehee

  • Simple way to test any device is to use it upto limits.
    U will see whether device meets ur requirements…


  • close
    >