Samsung Again Caught Inflating Benchmarking Scores, Phil Schiller Calls 'Shenanigans'

[Zerilos]

If I understand right, that's not actually the case. The phone actually specifically checks for benchmark apps:



So they're purposely overlocking the CPU just for the benchmark apps. The question remains whether such overclocking ever happens with any non-benchmark apps, or if this feature is exclusively there to fake benchmarks.

I'm guessing the CPU can indeed reach that performance, but for battery life and overheating reasons, it's limited to never do so. And then an exception is made for benchmarks, I guess.

According to the Ars Technica article, it appears to only affect certain benchmarking apps.

 

[Ries]

Instead of innovating, they're copying and trying to cheat the benchmark. Meanwhile, Apple creates an M7, A7 and a whole lot of new things which are thought through and beyond known limits...

64-bit is from ARM, M7 is a Cortex-M3 core built by NXP (and they selected a bad sensor too http://www.insideactivitytracking.c...ate-accelerometer-and-gyroscope-measurements/ ) and the finger print sensor is properly also from some other firm.

 

[John.B]

4. You just throw a tantrum when an Apple employee points to an article that points to the fact that Samsung has again been caught again intentionally deceiving reviewers (and thus potential consumers). Oh…I forgot…he also said shenanigans.

Fixed.

 

[dec.]

This thread turned out as I predicted. Never change MR Users... never change.

I agree completely, the Apple fanboys and Samsung apologists make it a bit too easy, don't they

 

[Innuendo]

Apple thrives on any opportunity to get noticed. A clever strategy that helps them overcome their shortcomings.

Do you have anything to say that is actually relevant to the discussion, or are you limited to non-sequitur comments and embarrassingly obvious, thoughtless Apple-bashing?

 

[giantfan1224]

Flash ram in Apple computers and chips in i-devices?

The key word is "device". I don't consider my iPhone a Samsung device in any way, shape or form. I don't care how many Samsung stamped components are inside.

 

[Col Ronson]

id love to see the samsung army rationalize this one.

probably something along the lines of "apple does it too" lol yeah sure. our phones are good enough we dont need to inflate scores.

 

[Zerilos]

64-bit is from ARM, M7 is a Cortex-M3 core built by NXP (and they selected a bad sensor too http://www.insideactivitytracking.c...ate-accelerometer-and-gyroscope-measurements/ ) and the finger print sensor is properly also from some other firm.

According to anandtech.com "The CPU cores are, once again, a custom design by Apple. These aren’t Cortex A57 derivatives (still too early for that), but rather some evolution of Apple’s own Swift architecture".

The fingerprint sensor was designed by Authentech for Apple. Apple own's Authentec.

 

[Rogifan]

64-bit is from ARM, M7 is a Cortex-M3 core built by NXP (and they selected a bad sensor too http://www.insideactivitytracking.c...ate-accelerometer-and-gyroscope-measurements/ ) and the finger print sensor is properly also from some other firm.

A7 is an Apple designed chip, not ARM.

 

[TheHateMachine]

The trouble is that the device DOES NOT release more power for a demanding task. It only ups the power for SPECIFIC benchmark apps. In daily use you'll never actually get to use that power on anything you do (except running specific benchmarks).

The main point is that Samsung is speeding up the benchmarks, but the user experience is equivalent to devices with lower benchmarks. In other words, they're lying.

You know for a fact the device in question never pushes more voltage to throttle up when the load increases? Please provide your in depth tests of cpu clock speeds and voltages when engaging the phone in demanding tasks.

 

[Carlanga]

Apple executive Phil Schiller -- senior vice president of Worldwide Marketing and the most prolific tweeter amongst Apple's senior staff -- linked to the Ars article in a tweet, saying only "shenanigans".

Soooooo..... the executive Of Apple publically slanders it's biggest supplier of components, wow, School boy tit for tat is the Apple way these day's then? They are fast becoming even more hypocritical and truly unprofessional and immature than ever now.

I do hope one day samsung would just tell Apple to F Off! Or perhaps Apple feels it MUST slander the competition now seeing as iOS7 is an utter disaster and making people literally sick! Can't remember the last Galaxy smart phone that made people physically sick?

Oh and yeah I work for Samsung and get paid tons of moolah for posting stuff on here

Soooooo... I think you need to find a dictionary and search the word slander.

 

[Zerilos]

You know for a fact the device in question never pushes more voltage to throttle up when the load increases? Please provide your in depth tests of cpu clock speeds and voltages when engaging the phone in demanding tasks.

According to the Ars Technica article, this act of deception only effected certain benchmarking apps with the intention of providing misleading results. You can disagree with their expertise (which I assume far exceeds yours), but don't get mad when people here assume that Ars T knows what they're talking about.

 

[kobalap]

Does this seem more of a feature only to me?
When a particularly demanding task, like a benchmark, is executed, the CPU releases more power. I call this "efficiency".
And these numbers came out from the CPU's calculating power, not from nowhere.. how is this fake?

Well, let's put it this way.

Let's say that Samsung was offering a t.v. that supposedly provides 1080p. But the way Samsung is claiming 1080p resolution is through some measuring device for which the t.v. is set up to peak at 1080p.

However, for real high def programs - t.v. shows, movies on blu ray, live sports, etc, let's say that the t.v. can only provide 900p of resolution.

Would you feel like Samsung was providing honest specs by rating their t.v. as 1080p?

 

[apolloa]

1. If you think an Apple VP linking an article and saying "Shenanigans" is slander, then no you don't.
2. Yes I was talking about TV ads. This doesn't count? Why? They were certainly approved by Samsung execs.
3. see #2
4. You just throw a tantrum when an Apple employee points to an article that points to the fact that Samsung has again been caught intentionally deceiving reviewers (and thus potential consumers). Oh…I forgot…he also said shenanigans.

I haven't 'thrown a tantrum', did you read my post and realise I have actually NOT defended Samsung and now changed your wording then? seems like it. Also those TV ads only show the features the iPhone lacks really. Like NFC for one. But hey I didn't feel insulted by them, then again I've never queued for several hours to buy a phone
And Phil Schiller is not exactly an Apple employee in the same light as a store assistant! He is a well known Apple executive who should behave better rather than stoke the fire.

 

[E.Lizardo]

This is pretty hilarious actually. Why not just always run the CPU this way

Cuz the battery will be dead in 15 minutes.

 

[I am Andy]

It's not about numbers. It's about personal preference.

Both Samsung and Apple make outstanding phones. Both Android and iOS are excellent OSes. There are pluses and minuses to both hardware and software platforms.

Don't mix up your fantasy and reality. See what I did there?

Well, you can't talk about 'the' Android. There's a bunch of different versions out there.
Not comparable to the consistent landscape of iOS.
Whatever. I like iOS and I'm happy without malware.
So you buy android. And I'll buy the good stuff
I'm willing to pay for it.

 

[gnasher729]

The trouble is that the device DOES NOT release more power for a demanding task. It only ups the power for SPECIFIC benchmark apps. In daily use you'll never actually get to use that power on anything you do (except running specific benchmarks).

The main point is that Samsung is speeding up the benchmarks, but the user experience is equivalent to devices with lower benchmarks. In other words, they're lying.

Actually, it seems to speed up for every app that is called "GeekBench 3" (and some other names). Take "Angry Birds", rename it to "GeekBench 3" and it will run 20% faster while killing your battery.

 

[kobalap]

If you're going to make an accusation you better have evidence to back it up...

All you have to do is go buy a light bulb and you will see confusing and misleading information. For example, what is a 100 watt light bulb? How does a watt indicate how much brightness a light bulb emits?

Or go to Best Buy and look at speakers. What is a 500 watt speaker? Or, what is a 1000 watt amplifier?

Or go to a car lot. Which car that shows 30 miles per gallon can actually get 30 miles per gallon unless all excess weight is removed (like the spare tire) and al of the tires are over inflated?

 

[TWSS37]

I was not aware the unwritten rule of benching is "normal activity"

The reason I say this, when people in IT refer to benching, its all about getting the highest score possible, which in terms of benching systems (PCs) is to overclock them to the point of where they are just stable enough to run the benchmark, and not remain stable in everyday use. They even run events to see how far people can OC intel/amd chips etc. Choose OS versions and drivers that deliver the most performance, instead of stability.

Now samsung have been know to be dodgey but these benchmarks mean little to average users anyway, as do hardcore benchmarks

Precisely my point, the site that did this cited benchmarking to be under 'normal usage conditions', in which is quite subjective. At least if you attempt to hit 100% you're at the very limit of what can be achieved which cuts down on the subjectivity considerably.

 

[kobalap]

I have to worry about the products made by a company that is willing to lie about their performance and to do this so overtly. GM lied about its Chevy hybrid and Nissan lied about its Leaf and Samsung lies all the time. There is "specsmanship" and there is outright lying and there is a difference.

People who buy on specs alone pretty deserve it.

 

[XboxMySocks]

Battery life... It's the same concept as Intel's SpeedStep. Pretty sure when benchmarks are running, SpeedStep is pretty much disabled.

However, on Android, one can change the speed of the CPU on a rooted phone; so while stock phones may not be able to achieve these speeds, a rooted phone should be able to with some additional software.

Yes, as stated in another post I'm well aware of this. I was trying to come off as sarcastic and not successful. My apologies.

 

[kobalap]

Yes he's quite good at whoa is me, we're the victims.

It's just business, man up and quit whining Apple. If you're still sad look at the billions you have. With Apple greed knows no limits.

You do realize that it is you Samsung customers that Samsung is trying to cheat, right?

 

[typeadam]

This is a fun read as the clock nears 5pm. Thanks MR!

 

[apolloa]

Precisely my point, the site that did this cited benchmarking to be under 'normal usage conditions', in which is quite subjective. At least if you attempt to hit 100% you're at the very limit of what can be achieved which cuts down on the subjectivity considerably.

Exactly, benchmarking is designed to push the system to it's maximum and tax it in order for you to see a result of the maximum performance available.

 

[TWSS37]

Only the Samsung device wasn't speeding up when it detected a "when the need for power." I think we'd all agree there was nothing wrong with that.

It was speeding up when it detected that the application you were running was in a list of benchmarking applications that were coded into the device. So the benchmark reported performance 20% above that of any real-world applicaton.

But the code told it when it needed the power, which was when running the benchmark (and other core apps - why is this being ignored from the article?)

----------

That makes no sense. Every CPU can be over clocked. Yet, they are not generally sold that way because doing so hurts the CPU and can cause overheating, sporadic OS issues, and decreased battery life.

Benchmarking is not to test what the chip can do when pushed to its over clocked absolute max, but the performance of the shipping product. I guarentee if Apple did this it would be sued by people who relied on the test to make buying decisions.

I disagree with your assessment; when I OC my CPU/GPU overvolt RAM, etc. I'm running benchmarks to get the highest score with pushing the boundaries of my rig.