Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
Android Cellphones Software Technology

Huawei Caught Cheating Performance Test For New Phones (techcrunch.com) 74

An anonymous reader quotes a report from TechCrunch: UL, the company behind the tablet and phone performance benchmark app 3DMark, has delisted new Huawei phones from its "Best Smartphone" leaderboard after AnandTech discovered the phone maker was boosting its performance to ace the app's test. The phones delisted were the P20, P20 Pro, Nova 3 and the Honor Play. "After testing the devices in our own lab and confirming that they breach our rules, we have decided to delist the affected models and remove them from our performance rankings," the company said in a statement.

For the Huawei case, the rules are actually a little fuzzy. Phones are permitted to adjust performance based on workload, which results in peaks or dips in performance for different apps, but they are not permitted to hard-code peaks in performance specifically for the benchmark app. Huawei reportedly claimed that the peak in performance seen during the run of the benchmark app was an intuitive jump determined by AI; however, when an unlabeled version of the benchmark test was run, the phones were unable to recognize it and, as a result, displayed lower performances. In other words, the phones aren't so smart after all.

This discussion has been archived. No new comments can be posted.

Huawei Caught Cheating Performance Test For New Phones

Comments Filter:
  • Cheating and spying are no way to go through life, son.

    • by Anonymous Coward

      Damn right. Let that be a lesson to Americans, because there is no other country that has cheated, rigged, lied, spied, and manipulated as much as America to get to where it is.
       

  • AI (Score:5, Insightful)

    by 110010001000 ( 697113 ) on Friday September 07, 2018 @09:11AM (#57268576) Homepage Journal
    "Huawei reportedly claimed that the peak in performance seen during the run of the benchmark app was an intuitive jump determined by AI"

    Whenever someone makes an AI claim you know they are lying.
  • by Anonymous Coward on Friday September 07, 2018 @09:28AM (#57268654)

    If you know what chipset is in a phone, then you know as much as you usefully need to know about the real world performance of that phone relative to other phones with the same chipset, i.e. pretty much the same. If you see a benchmark saying one such phone is outperforming its peers, then you assume that, if it's not down to actual cheating, then it's a quirk of the benchmark, because you know the same cpu/gpu at the same speed gives the same performance. Does anybody actually look at benchmarks when buying a phone?

    • What's more: using the same SoC, performance may still vary due to cpu / gpu / memory clocks. But higher clocks translate to higher power consumption. For a portable device, that means more battery drain. So even if you can run apps a few % faster, you'll only be able to do (roughly) the same amount of work before battery runs empty.

      In battery operated devices, you'll often see that components aren't clocked up to their highest possible spec. The above is one important reason for that. Local heating / st

      • by Anonymous Coward

        I assume they either have connected the phone to AC with the charger, or not, and have not changed this aspect when testing with the original 3DMark app and then with a replacement named differently.

        This is cheating, and thus I will avoid Huawei/Honor in the future when making purchase decisions. It is the same pattern as with Volkswagen (VW).

      • "For a portable device, that means more battery drain. So even if you can run apps a few % faster, you'll only be able to do (roughly) the same amount of work before battery runs empty"
        Actually no. In the olden days the supply voltage was fixed, leakage current was low resulting in a constant energy per cycle for a given operation. In modern times the supply voltage is carefully adjusted with clock speed. Run the part down at 1V and it might only run at 1 GHz, crank it up to 1.5V and it goes at 2 GHz, bu

    • by Desler ( 1608317 )

      Not true at all. The same SoC can have a wide difference in performance due to differing thermal headroom depending on the device.

  • Comment removed based on user account deletion
  • by Anonymous Coward

    So now we know where ex-Volkswagen engineers end up.

  • by Anonymous Coward

    Quake/Quack3 was 17 years ago and people still haven't learnt.

  • by guygo ( 894298 ) on Friday September 07, 2018 @11:37AM (#57269388)
    Who wouldve thought the company that gives approval rights for their designs to the Chinese government would lie to us? But for sure there's no possibilty of any backdoors. Trust us!
    • Yeah because American companies would never cheat benchmarks or add backdoors. Right? RIGHT?

    • Ironically, a backdoor controlled by the Chinese government is no threat to me. What could they possibly do? On the other hand, the NSA having backdoors into my communications is a real threat indeed.
  • But is this what GPU manufacturers have been doing for years?

    • Yes, it was wrong for them and it's wrong for Huawei now. Anandtech, HardOCP, Tom's, and others caught GPU manufacturers doing this and called them out just like now. Part of the reason for Anandtech's development of their own test suite was because they didn't completely trust that vendors weren't cheating the known industry benchmarks.

      So far as I know, the above sites are still looking for cheating and GPU manufacturers have stopped. They've at least stopped obviously cheating by looking for when a know

All seems condemned in the long run to approximate a state akin to Gaussian noise. -- James Martin

Working...