Samsung denies cheating on benchmark tests
0. phoneArena 03 Oct 2013, 13:06 posted on
The evidence seems pretty hard to argue, but Samsung wants to try anyway. Multiple sources have found that Samsung (as well as most other Android manufacturers) artificially boost CPU frequencies when the system detects a benchmarking test running. Samsung is denying that its devices are trying to cheat the tests, and is actually putting the blame on the tests themselves...
This is a discussion for a news. To read the whole news, click here
1. AfterShock (Posts: 2564; Member since: 02 Nov 2012)
Just like former world class atheletes., I didn't dope!
49. Commentator (Posts: 2331; Member since: 16 Aug 2011)
Ryan Braun would fit in just fine as their spokesman.
55. LG_is_the_best (unregistered)
Cheaters never lose, and losers never cheat
71. JC557 (Posts: 956; Member since: 07 Dec 2011)
Take a look at the Anandtech article on this. It's pretty evident that all OEMs have some cheats up their sleeves so Samsung shouldn't be so quick to deny it.
74. yudi.nemesis (Posts: 1127; Member since: 14 Mar 2013)
At the Anandtech article “ONLY” Samsung, Asus, LG, and HTC is proven of cheating in android benchmarks.
76. joey_sfb (Posts: 2737; Member since: 29 Mar 2012)
WELCOME! to the Samdung Hate Group Gathering.
Have a Samdung today!! Everyday!
83. JC557 (Posts: 956; Member since: 07 Dec 2011)
That leaves only Motorola and Nexus branded phones... though one could argue that it's just Google that's not cheating as Moto belonogs to them now.
No Sonys on the chart so it remains to be seen, but pretty much all major OEMs "optimize" a bit while some go too far.
86. JC557 (Posts: 956; Member since: 07 Dec 2011)
About the Note 3, I'm curious if the increase in thermal headroom has anything to do with better benchmark scores. The CPU and GPU might be able to run a bit less restricted.
90. yudi.nemesis (Posts: 1127; Member since: 14 Mar 2013)
*Motorola, Nexus, and NVIDIA.
*“ALMOST” all Android OEMs.
80. blazee (Posts: 254; Member since: 02 Jan 2012)
This just further solidifies my choice in only getting nexus devices. Google seems to be trying to create a great user experience rather than try to deceive and cheat to sell their devices. Sad that there might be time wasted before rolling out a system update because of optimization and tweaking solely to achieve higher benchmark scores by some manufacturers.
85. JC557 (Posts: 956; Member since: 07 Dec 2011)
There's an alarming amount of Samsung apologists... seems like they're on their way to becoming what they accuse Apple users of.
Every OEM needs to be scrutinized since consumers and not bah bahs should control the market trends.
2. Ninetysix (Posts: 1476; Member since: 08 Oct 2012)
BS again. If that was the case, the gpu/cpu speeds should be the same EVEN if you rename the application.
3. datsyuk (Posts: 156; Member since: 11 Jan 2012)
In other words they only "maximize" the CPU/GPU on benchmark tests.
41. PhoneArenaUser (Posts: 5471; Member since: 05 Aug 2011)
Agree, Samsung didn't fake benchmark results, only "boosts" the CPU/GPU on benchmark tests. So on the one hand it is not cheating because results are real. But on the other hand it still tricks user to think that CPU/GPU always runs with high performance but it is not true, CPU/GPU runs full power only with benchmark apps.
50. Shatter (Posts: 1989; Member since: 29 May 2013)
This is like Intels turbo boost. It is not cheating it just boost performance when needed if the chip is at good temperatures. Samsung could shut everyone up if they added to the specs sheet with "SBoost" up to 2.5ghz in applications that need it or whatever.
60. PhoneArenaUser (Posts: 5471; Member since: 05 Aug 2011)
"Samsung could shut everyone up if they added to the specs sheet with "SBoost" up to 2.5ghz in applications that need it or whatever."
Actually Samsung already made such list, but the problem is that on this list is only benchmark applications and that's why now we see such scandal about cheating.
73. tedkord (Posts: 4497; Member since: 17 Jun 2009)
That's not what happens. They don't overclock, they just ramp the CPU/GPU to max for the test, rather than letting the app load determine the speed. It's not technically a cheat, a cheat would be giving a false number that it didn't achieve. It's more of a best case scenario.
But it's misleading, so I have no problem with the word cheat.
52. HansGoneInsane (Posts: 371; Member since: 09 Aug 2013)
But you said yourself that Samsung tricks his customers by creating the illusion that they could get the same performance when not running benchmarks. The word "trick" is a synonym for the word "cheat". In other words: Samsung cheats on his customers by purposely creating wrong impressions of its products.
61. Shatter (Posts: 1989; Member since: 29 May 2013)
They trick their customers with ads and everybody fakes benchmarks because idiots believe phones are only as good as their benchmarks show. The only benchmarks should be real world performance ones like how much FPS a game gets or how fast a zip files compresses or how quick it can encode a video. Artificial benchmarks like antutu give you a false idea of how powerful a device is. Some roms have the Samsung cpu auto overclock crap deleted and it makes benchmark scores drop but the device will be much more smooth/efficient with all the optimization and such. There are also roms that target benchmarks to give you higher scores so you can feel better about yourself if your one of those people.
66. PhoneArenaUser (Posts: 5471; Member since: 05 Aug 2011)
Samsung didn't fake benchmark results, CPU/GPU really achieves such results. The problem is that user by default can't achieve such CPU/GPU power while using other games and applications. CPU/GPU runs full powe only with benchmark applications which are specified on the list.
It is like a cars engine with limiter which restrict engine's maximum rotational speed. Car really can go faster just need to remove limiter.
Anyway, I agree that such things sucks! :(
75. tedkord (Posts: 4497; Member since: 17 Jun 2009)
They don't only run full tilt then. Any good benchmark should taboo them to full anyway. If an app or game doesn't, it's because it doesn't need it.
7. Ninetysix (Posts: 1476; Member since: 08 Oct 2012)
Yes. Just wait a little longer for the Samsung herd to post. It'll come....trust me :)
edit: Post #6 is one
51. Shatter (Posts: 1989; Member since: 29 May 2013)
Is Turbo boost on a desktop chip cheating? NO. Samsung should list though in the official specs what it actually goes up to.
56. Ninetysix (Posts: 1476; Member since: 08 Oct 2012)
is Turbo boost only active on certain applications controlled by the manufacturer or is it always on/on demand? There is a difference.
59. Shatter (Posts: 1989; Member since: 29 May 2013)
It activates automatically just like Samsungs does when the CPU is nearing 100% use. The only smartphone apps that use 100% are benchmarks as far as I know. It does not target applications, if it did then go make a brand new benchmark and see if it targets it without a system update.
63. Ninetysix (Posts: 1476; Member since: 08 Oct 2012)
It's not the same in the case. Samsung only runs their "turbo boost" in their whitelist apps. Rename the benchmark app listed in those whitelist and it runs in normal mode.
69. ngo2dd (Posts: 781; Member since: 08 Jul 2011)
He is a troll or a shill one of the two. Turbo boost is very different. It will only boost if the program need the extra amount of power. All program can have access to turbo boost. Now the Samsung boost is only for the white list app.
92. RebelwithoutaClue (Posts: 593; Member since: 05 Apr 2013)
So why doesnt it boost when the benchmark app is running under a different name? Same situation, other name and it should boost then too. No need to make a brand new benchmark.
8. rawbow (Posts: 409; Member since: 30 Mar 2012)
Well, benchmarks does satisfy most of the droid lovers out there. Its only natural for these OEMs to target that.
9. amasog (Posts: 129; Member since: 22 Aug 2013)
of course Sammy will deny it.
what to expect from them? lol!!
11. Danishs (Posts: 147; Member since: 23 May 2013)
BS. The CPU raises it's maximum frequency while running the tests
12. nestea (unregistered)
emmm silly fandroid, yes it goes for maximum frequency but they increase lowest frequency also. Try to benchmark your android with 200-1500mhz and then with 1200-1500mhz….
77. tedkord (Posts: 4497; Member since: 17 Jun 2009)
No, it doesn't. It does ramp it up to it's max advertised speed for the duration of the bench, instead of letting the built in governor control it.
It's like using performance governor on a custom ROM - it does the same thing, keeps the CPU at max speed always (except when asleep)
82. Danishs (Posts: 147; Member since: 23 May 2013)
No the max CPU frequency gets raised according to anandtech
99. tedkord (Posts: 4497; Member since: 17 Jun 2009)
No, that's not what Anand stated. They clearly said that max supported CPU speed is never exceeded. They also stated that the effect is negligible, since these benchmarks should drive the frequencies to max.
13. darkkjedii (Posts: 10490; Member since: 05 Feb 2011)
Barrysung Bonds. S-Rod. Roger Sammens. Samsung Censeco.
79. darkkjedii (Posts: 10490; Member since: 05 Feb 2011)
Thanx Ted, appreciate the kind words bro. +1
21. verbalize (Posts: 81; Member since: 03 May 2013)
No I'm 2... Any person with common spence knows Samsung's practices are low and desperate. How do you think they got where they are? Lie, lie, lie
93. RebelwithoutaClue (Posts: 593; Member since: 05 Apr 2013)
Welcome to the world of big cooperations. They all lie, that's how you make the big bucks. Some get caught, others don't.
What is common spence?
28. verbalize (Posts: 81; Member since: 03 May 2013)
No just a person who educates themselves with company's who copies every idea from someone else. I see why you are a Samsung fan.. LOL
53. Pancholo (Posts: 377; Member since: 27 Feb 2012)
You should spend half of said education in grammar.
81. darkkjedii (Posts: 10490; Member since: 05 Feb 2011)
Why would you call them "a bunch of losers"? They're no different that any other company, build for low sell for high. The rep their product, and they should. It's fun to poke fun at them, just like it is to do it to apple, but in reality it's just big business.
94. RebelwithoutaClue (Posts: 593; Member since: 05 Apr 2013)
So very true.
And everytime we poke at them, naming them, their brand name is mentioned ;)
Sometimes I wonder if Apple and Samsung aren't working together in this. It's always Apple/Samsung and they do work together in some areas. You don't hear people talking sh#t about the HTC one max getting a fingerprint sensor and 'copying' Apple.
97. darkkjedii (Posts: 10490; Member since: 05 Feb 2011)
You noticed that too huh? Just like the Pepsi/coke wars, the wars bring focus to just those two, and makes the rest irrelevant. I've said before, these two are not the enemies we all think they are. +1
98. RebelwithoutaClue (Posts: 593; Member since: 05 Apr 2013)
I agree, like that big trial that was going on, where Samsung had to pay hunderds of millions to Apple. Never heard anything if they paid it or not. And it's usually Apple and Samsung going to court. Guess those two have the last laugh.
Btw, did you get married or something (your avatar changed)? If so, congrats on your marriage.