x PhoneArena is hiring! Reviewer in the USA
  • Hidden picShow menu
  • Home
  • News
  • Here are the first benchmarks of the Galaxy S8+ and its Exynos 8895 chipset

Here are the first benchmarks of the Galaxy S8+ and its Exynos 8895 chipset

Posted: , by Kaloyan C.

Tags :

Here are the first benchmarks of the Galaxy S8+ and its Exynos 8895 chipset

In completely unsurprising news, the first benchmarks of the international variant of the Galaxy S8+ may have just hit the web, courtesy of benchmarking tool Geekbench's database. Seriously, Samsung, get your leaks under control. Still, this is also possibly the first benchmark we have of the Exynos 8895 chipset, and the results are certainly promising.

First off, let's compare the overall scores between the two S8+ variants, the US-bound Snapdragon 835 one and the Exynos 8895-equipped international one. Samsung's chip scores better in both single-core and multi-core tests, with its scores of 1978 and 6375 being slightly higher than the 835's respective scores of 1929 and 6084. The difference is small, but it's there, though it's far from surprising, considering Exynos chips have consistently outperformed their Snapdragon counterparts.

Once we go into the specifics, however, the differences start getting a bit more apparent. Since the 835's low-power CPU cores run at a higher clock speed than the 8895's, the former's single-core performance is marginally better in a few areas, with the memory score being a particular outlier. However, Qualcomm's chip suffers from very high memory latency in both single- and multi-core tests, which is quite likely to negatively impact end-user performance. Apart from that, few things jump out as particularly interesting, especially when taking result fluctuations into account, as they are an unfortunate but unavoidable side effect of benchmarking.

Those metrics, however, are mostly irrelevant for the regular consumer, who won't really get a choice regarding which version of the device they'll get. So let's move onto the more interesting part, which is comparing it with its predecessor, the Galaxy S7 edge. When comparing the leaked results with a recent benchmark of the S7 edge, the Exynos 8895 vastly outclasses the older 8890... but only in its multi-core results. 

Users are unlikely to notice much difference in performance between the S8+ and its predecessor

Single-core results are a mixed bag, with the older chip scoring better in quite a few areas, though, admittedly, the differences are small. However, during regular usage of a device, single-core speeds are much more important, as most Android apps don't take advantage of multithreading, and also run on the slower, low-power CPU cores, never taking advantage of the better multi-core performance.

Still, it's worth noting Samsung's engineers have apparently worked hard in improving a few specific use cases, or at least is seems so from looking at the new benchmark results. These include face detection, speech recognition, and HDR processing, all three of which are getting ever more popular among consumers. The first one, face recognition, is also particularly interesting due to the recent rumors of Samsung finally including the option in its newest flagship pair.

As usual with leaks and benchmarks, take all of this information with a few grains of salt. Test results can be faked easily (there's a treasure trove of fake S8 benchmarks on the Geekbench site right now), but these appear to be the real deal, both due to the generally reputable source, and also the similarities with the older Snapdragon 835 benchmarks we used as a comparison. Of course, it remains to be seen how either of the S8+ variants perform in real life, but we expect nothing short of excellent, considering Samsung's previous track record.

  • Options

posted on 17 Mar 2017, 04:58 2

1. lpratas (Posts: 372; Member since: 09 Nov 2011)

Yeah, but at a frequency of 1.69GHz, i presume.

posted on 17 Mar 2017, 05:25 4

6. UglyFrank (Posts: 1877; Member since: 23 Jan 2014)

Does Geekbench have a habit of reporting the clock speed of the lower powered cluster?

posted on 17 Mar 2017, 06:08 1

19. Nopers (Posts: 395; Member since: 01 Jun 2015)

Yep, but only in leaks for some reason. The A10 benchmarks were leaked at 400mhz

posted on 17 Mar 2017, 06:30 10

24. vincelongman (Posts: 4950; Member since: 10 Feb 2013)

Yep, Geekbench still hasn't been updated to report Android SoCs with multiple CPU clusters

These benchmark seems to be legit
The Processor IDs are right and scores are what we should expect

The 835's scores are very close with the Kirin 960 (which points to A73s+A53s)
Hopefully this means Qualcomm's CPU hasn't messed up this time

posted on 17 Mar 2017, 07:37

46. androiduser (Posts: 217; Member since: 18 Jun 2014)

Imagine how powerful the kirin 970 will be lol. Makes me want to wait for the huawei mate 10

posted on 17 Mar 2017, 07:59 1

52. vincelongman (Posts: 4950; Member since: 10 Feb 2013)

Yea, you should wait for the 970 or 835
Both should be more efficient than the 960

The 960 isn't much of an upgrade from the 950, it's more power hungry, especially the GPU which is half as efficient as the 820's
Due to TSMC's cheaper 16FFC being designed for lower frequencies compared to their high end 16FF+ used in the 950 and probably slightly rushed implementation


posted on 17 Mar 2017, 08:47

53. androiduser (Posts: 217; Member since: 18 Jun 2014)

Yeah true which i didn't consider getting huawei mate 9 or p10 plus, definitely waiting for the more efficient kirin 970 u

posted on 17 Mar 2017, 08:47

54. androiduser (Posts: 217; Member since: 18 Jun 2014)

Yeah true which i didn't consider getting huawei mate 9 or p10 plus, definitely waiting for the more efficient kirin 970 u

posted on 17 Mar 2017, 10:13 4

67. sgodsell (Posts: 4518; Member since: 16 Mar 2013)

Geekbench should be taken with a grain of salt. It's one of the worst benchmarking tools, period.

posted on 17 Mar 2017, 10:23

70. kiko007 (Posts: 4978; Member since: 17 Feb 2016)


posted on 17 Mar 2017, 17:59

123. vincelongman (Posts: 4950; Member since: 10 Feb 2013)

I wouldn't say Geekbench 4 is one of the worst any more (Geekbench 3 was tho)

But all benchmarks should be taken as a grain of salt

posted on 17 Mar 2017, 05:11 3

2. Bankz (Posts: 1333; Member since: 08 Apr 2016)


posted on 17 Mar 2017, 05:17 6

3. Plasticsh1t (Posts: 1763; Member since: 01 Sep 2014)

Too slow.

posted on 17 Mar 2017, 05:54 9

15. kumar1234 (unregistered)

Another failure from Samsung. Only 1978 in single score. LOL. S-sheep who were hoping for better score than SD will sure be disappointed. This along with record low small battery for large screen phone, same megapixel yet insane asking price a low class korean phone S8 will be a disappointed device. Can't wait for its announce date.

posted on 17 Mar 2017, 06:23 23

23. DRS1977 (Posts: 478; Member since: 27 May 2015)

You can't afford the phone anyway, why are you so excited for it's announce date? Samsung has more money than you will ever have in a million lifetimes.

posted on 17 Mar 2017, 07:15 1

35. trojan_horse (Posts: 3894; Member since: 06 May 2016)

And this benchmark was definitely run at a lower frequency, probably at 1.7GHz... So this benchmark doesn't quite add up.

posted on 17 Mar 2017, 07:25 14

38. trojan_horse (Posts: 3894; Member since: 06 May 2016)

Like seriously, PA???

"However, during regular usage of a device, single-core speeds are much more important, as most Android appsdon't take advantage of multithreading"

Android isn't iOS, FFS!

Multi-core is what matters most on Android and apk apps definitely make use of multiple cores, like stock apls - Google's Chrome browser, and even video playback.

Not to mention games, for that matter.

posted on 17 Mar 2017, 09:07 1

57. TechieXP1969 (Posts: 13501; Member since: 25 Sep 2013)

Really? I have a question. Can you justify as to wwhy single scores matter?

First off, the operating system is multi-threaded, which means NO OS uses a single core to run.

Apps are multi-threaded too and even if you happen to use one that isn't, which now is pretty much impossible without using a dinosaur phone, they use multi-cores too.

So really the single core score doesn't matter. The only reason its matter to you is because the iPhone scores higher in a single core. Yet my principle which is a fact remains the same. IOS is a multi-threaded OS and so are the apps.

S-Sheep? About as dumb as saying iSheep. But iSheep actually looks better on paper. S-Sheep as a made up word looks complete stupid. But hey what can you expect, most iPhone users are completely tech-stupid to begin with.

Small battery for a large phone. The iPhone really has a small battery for alarge phone. IN fact the iPhone battery is 40%+ smaller. You are a fool bruh!

I'm not gonna even go any further.

I hate trolls. You all ruin tech talk.

posted on 17 Mar 2017, 16:48

122. Trex95 (Posts: 1068; Member since: 03 Mar 2013)

Don't think so. I own the S7 edge 128 gb Exyonse 8890 aka black pearl single core 1870 no diffrence than 32GB version! I thought it's like iPhone 7/ 7 plus bigger storage faster, but it doesn't seems that iPhone still the winner iOS still remain the most optimized smart phone OS to date.

posted on 17 Mar 2017, 05:25 17

4. dazed1 (Posts: 516; Member since: 28 Jul 2015)

"However, during regular usage of a device, single-core speeds are much more important, as most Android apps don't take advantage of multithreading, and also run on the slower, low-power CPU cores, never taking advantage of the better multi-core performance"

Bunch of iTard BS.

posted on 17 Mar 2017, 05:38 3

10. gehrig (Posts: 321; Member since: 17 Apr 2014)

You spelled "the truth" wrong.

posted on 17 Mar 2017, 05:57 3

16. kumar1234 (unregistered)

You are F-ing moron. In mobile app single app performance is much more important than multicore. Go read anandtech, idiot. They'll say the same thing. Don't get butthurt just because your god's chip-set score is low and PA is telling the truth.

posted on 17 Mar 2017, 06:34 6

25. vincelongman (Posts: 4950; Member since: 10 Feb 2013)

While single core is more important, its not hugely more important

posted on 17 Mar 2017, 06:35 6

26. Supraman21 (Posts: 371; Member since: 09 Jun 2010)


I figured it was common knowledge that Android apps and Android itself utilizes all cores.

posted on 17 Mar 2017, 07:00 9

30. Macready (Posts: 1402; Member since: 08 Dec 2014)

Sounds like you should take your own medicine, since Anandtech clearly proved you and the writer of this article wrong, see the links above (and many more articles on the matter).

I cant believe this " most Android apps don't take advantage of multithreading" myth is still repeated. 6 year ago my S2 already used multiple (2) cores for most apps. S-I-X years being in hybernation?

Now if the write head said "most apps don't maximize the full potential of multithreading between 8 cores", then he would have a point. But they are still used, even if code isn't optimized for them.

posted on 17 Mar 2017, 07:14 1

34. Macready (Posts: 1402; Member since: 08 Dec 2014)

"Now if the writer had said" *

posted on 17 Mar 2017, 08:59 6

55. tedkord (Posts: 13235; Member since: 17 Jun 2009)

Not only this, but these trolls like to act as though the app you're using is the only process the phone is performing at the time. There are dozens of processes running besides the foreground app, and with Android you've also got the multitasking, so you can have other apps running simultaneously.

But, because the iPhone can't do true multitasking, and is better with single core, it's all that care about.

posted on 17 Mar 2017, 09:32

61. TechieXP1969 (Posts: 13501; Member since: 25 Sep 2013)

What matter is simple. If you took todays mobile applications and loaded them on an older Android with an OS high enough to run them, and that phone had a "single-core" teh app will still work.

But dpening on what that app is, it could be very slow.

Let's take a game.

The benefits of muti-core are exactly the same as the benefits of multiple cylinders in a car.

FACT - No matter how many cylinders a car has, technically it can run on just one.

But like the CPU, you have to consider how much load it is going to be pulling.

If an app uses very lower power, then yes the OS can restrict said app to on core. Many app like web browsers and any app that deals mostly with text can use just one core.

Most OS' use all the cores when the app is first launched. Once idled down, the app can be reduced to a single thread.

But lets say that app is a game. Games have more load. Thus one core woudl not be enough.

The iPhone can survive with jsut 2 cores and a single score matters because the whole OS is different.

iOS has direct hardware access an regulates ist need of power. iOS doesn't need quadcores because the OS has been limited to specific hardware set.

Android phones all have very different hardware specs and one core woudl not provide enough resources to run the software needed to run so many services that support every driver and application that is running.

The article you posted, actually tells IOS users here how stupdi they are.

BEcause first iOS and Android handle apps and hardware different. Virtual Machine work different vs how iOS woudl work on identical hardware.

But here is one fact that seems to never be brought up. Any Android phone today, would run iOS better than how iOS runs on its own hardware, do to our phones having much higher specs.

Android as afact woudl crash and burn because iPhone hardware is simply to low.

The iPhone for example, could likely runs older versions of Android with no issue. But it could run anything past Android 5.x.

Proof? Find a phone with similar specs as todays iPhone and put the latest version of Android on it and see how slow it will be.

While PlasticDummy and KumarTHEDUMMER try to talk, the facts show Android devices are far more powerful vs an IOS based device and its just not the specs. Our devices do more, yet still run as fast or faster than the iPhone in the VAST MAJORITY OF TESTS.

The only tests they all car about is ones like this. Benchmarks have never told how a device will wotk based on how it will be uses.

Benchmarks were designed to push hardware to ist complete peak. Its dumb because no one ever does this in normal usage.

So while the ifans are always having their panties in a bunch over numbers, we are in control knowing Android devices have proven their snibbling BS is just that...BS!

posted on 17 Mar 2017, 09:53 1

63. toukale (Posts: 63; Member since: 10 Jun 2015)

@TechieXP1969 - Damn, someone is really upset, lol. Let's take some of your examples. You said since games uses multicores more often than others apps, it should show how powerful android is and yet all over youtube it shows that games run better and faster on ios. Then I am sure you will move the goalpost and tell me developers code their games better on ios. The only place android run faster is the animations, which Apple allows you to turn off if you don't want to deal with them. Try to have any android device decode a 4k files and see how much they struggle.

You guys just can't let it go really. When you design an SOC you can go different routes depends on what is more important to you and how much you want to spend on your silicon. It's just easier,simpler and cheaper to just throw more cores at the problem than to do what Apple is doing with their SOC. It's the same approach android oem's take with their ram managements. They would rather put more ram into the device than to spend more time to optimize the software. It's just easier and cheaper to do.

First They do not own android and the platform so it makes little sense to put all the energy into something that won't benefit them long term. It's very logical and actually makes sense they would go that route. Apple's SOC's size is big and silicon costs money, something Apple does not need to worry about but Qualcomm and all the other vendors do, Since they have to make money on selling those SOC's where Apple do not. At the end of the day it all comes down to money.

posted on 17 Mar 2017, 10:15

68. TechieXP1969 (Posts: 13501; Member since: 25 Sep 2013)

You and MacReady should be siblings.

Always a pleasure to read you guys posts.

These tech-ignorant children, really take the joy out of reading about tech.

They are so concern with a number, which don't really mean squat. Or when a writer who is a stupid fanboy liek them, post something only than can agree with and they are so stupid to agree, even whe their own facts prove them wrong.

How ignorant can you be to see a "STOP" sign and not stop?!

* Some comments have been hidden, because they don't meet the discussions rules.

Want to comment? Please login or register.

Latest stories