Samsung caught cheating again - this time in Note 3!

nikufellow

In the zone
The Samsung Galaxy S4 was caught ramping up its GPU clock speed when running a benchmark, so naturally the Galaxy
Note 3 was under suspicion too. Thanks to some excellent detective work by Ars Technica , now we have a confirmation.
The Samsung Galaxy Note 3 N9005 (the Snapdragon 800) version was tested against the LG G2 (same chipset) and it
came out well ahead, even though the internals are identical.
Using a CPU monitor, Ars found out that certain benchmarks are detected and the CPU is put on full throttle – all four
cores run at 2.26GHz all the time. When a non-benchmark app is run, when the app is idling three cores are off and the
fourth one is at 300MHz.
To get around the issue a special version of Geekbench 3 was created – since benchmark detection depends on the app’s
package name, Ars renamed the package and the CPU showed its regular “three cores off, other one at 300MHz” idle
behavior. The list of benchmarks that are treated specially also includes, GFXBench, Benchmark Pi, Linpack, Quadrant and
other popular titles.
And while the CPU monitor doesn’t catch it, there’s more going on than can be explained by disabling the idle mode of the
processor. Geekbench 3 scores are 20% higher with the artificial boost, while Linpack (also a renamed version) showed a
whopping 50% difference. Without the boost, the benchmarks are around the LG G2 scores, where you would expect them
to be. In all fairness the Note 3 does have some advantage remaining, only it’s not as big as initial tests suggest.
Source code for the Note 3 also refers to “LCD frame rate adjustment,” so the GPU might be getting boosted too.
While it’s not 100% clear what the boosting involves, by the looks of it Samsung targets benchmarks specifically – if
Samsung had included some games, we might take this as a special optimization, but with just benchmarks it seems like
trying to cheat.



Source : Samsung Galaxy Note 3 artificially boosts chipsets for select benchmarks by 20-50% - GSMArena Blog
 

amjath

Human Spambot
Using a CPU monitor, Ars found out that certain benchmarks are detected and the CPU is put on full throttle – all four
cores run at 2.26GHz all the time. When a non-benchmark app is run, when the app is idling three cores are off and the
fourth one is at 300MHz.

When you run benchmarks, CPU should run in full throttle right??? When u run simple apps it runs slower, so Can anyone explain me what is wrong here
 
When you run benchmarks, CPU should run in full throttle right??? When u run simple apps it runs slower, so Can anyone explain me what is wrong here

What's wrong is that people decide between products by referring to benchmarks. If note 3 appears super fast in benchmarks but is actually slower in other apps, then whats the point of benchmarks altogether?
 

Desmond

Destroy Erase Improve
Staff member
Admin
Exactly. They have been cheating at benchmarks for a long time now.

This sums this up very easily :

*cdn.gsmarena.com/pics/13/10/note3-benchmar-boost/gsmarena_001.jpg
 

rijinpk1

Aspiring Novelist
When you run benchmarks, CPU should run in full throttle right??? When u run simple apps it runs slower, so Can anyone explain me what is wrong here

LG G2 and note 3 has same cpu,nut note 3 performs far superior than G2 in bechmarks just because of this.
 
Last edited:

amjath

Human Spambot
LG G2 and note 3 has same cpu,nut note 3 performs far superior than G2 just because of this.

*cdn.gsmarena.com/pics/13/10/note3-benchmar-boost/gsmarena_002.jpg

Without benchmark booster, note 3 beats G2.

Now tell what this benchmarks do, I just dont get it. Benchmark for PC does so much things, what benchmark tool does for phones
 

rijinpk1

Aspiring Novelist
Nah. They cheating. Look at the screenshots above.

i meant to say that note 3 performs better on bench marks despite the fact that g2 and note 3 has the same cpu. sorry for not completing my previous comment. will edit now

*cdn.gsmarena.com/pics/13/10/note3-benchmar-boost/gsmarena_002.jpg

Without benchmark booster, note 3 beats G2.

Now tell what this benchmarks do, I just dont get it. Benchmark for PC does so much things, what benchmark tool does for phones

i meant to say that note 3 performs better on bench marks despite the fact that g2 and note 3 has the same cpu.
benchmarks for mobiles does so much things. It has its own value. But doing things like these by companies like samsung will certainly tend to say "bechmarks are gimmic for mobile phones".
 

The Sorcerer

oh wow...Xenforo!!!
Not really into smartphones, but how many of these tests are synthetic benchmarks? If pretty much how all of them are, I don't see how they're relevant in the first place? All this hype is no different when 3DMark Vantage earlier versions used to show benchmarks boosted with Physx on as opposed to physx off. A lot of people used the words like 'cheater' and all that over Nvidia, but isn't that the benchmark (Futuremark did make sure that it didn't interfere with future updates or setting the physx as off) software maker's job to make sure it doesn't get affected- or have a release which bypasses that boost, considering that it can be turned off.

If the boost does have real world performance effect, its not much different from Turbo or motherboard's features which bumps clock speed according to the workload, is it?
 
And moreover, its not the processing performed by the application that boosts the processor, but the fact thet the application is identified as a benchmarking software.
 
OP
nikufellow

nikufellow

In the zone
Facts can't be denied no matter how irrelevant the benchmarking process itself may seem it does heavily influence buying decision these days, even reviewing sites /blogs use these tools to rate devices.
 

Aerrow

Journeyman
So does this boost helps Note 3 to beat the Apple 5S?

There's simply no competition there. Android wins handsdown. :D

Realistically, iit's not ideal to compare two platforms which are completely different on the basis of such benchmarks.
 

amjath

Human Spambot
IMO benchmarks for phones are BS. We know snapdragon 800 is faster period. If u want to compare, compare 800 to a7 or Exynos 5420.
 

Aerrow

Journeyman
IMO benchmarks for phones are BS. We know snapdragon 800 is faster period. If u want to compare, compare 800 to a7 or Exynos 5420.


Exactly!

Just outta curiosity, is anybody gonna purchase n3 s800 version? Ebay or import?

I had an eye for that one and would have upgraded if that was the device for Indian market.
 

ZTR

Cyborg Agent
Exactly!

Just outta curiosity, is anybody gonna purchase n3 s800 version? Ebay or import?

I had an eye for that one and would have upgraded if that was the device for Indian market.

It might not work here though as Samsung has started to region lock its devices :cry:
Samsung Is Region-Locking Phones, But Not for the Reason You Think
 

srkmish

Ambassador of Buzz
Lol. Every manufacturer is doing this apparently.

AnandTech | They're (Almost) All Dirty: The State of Cheating in Android Benchmarks
 
Top Bottom