x PhoneArena is hiring! Reviewer in the USA

Exynos 7420 gets benchmarked

Posted: , by Alan F.

Tags:

The Exynos 7420 is benchmarked on Geekbench

The Exynos 7420 is benchmarked on Geekbench


With some versions of the Samsung Galaxy S6 expected to feature the Exynos 7420 under the hood, the chip was seemingly the subject of a benchmark test at the Geekbench site. The single-core score was 1520 and multi-core tally was 5478. The test scores were most likely the result of a Samsung Galaxy S6 prototype that went through the benchmark test.

You might have noticed the 1.5GHz clock speed on the listing. This is probably the result of the slower Cortex A-53 being in use. The big.LITTLE chip uses the latter cores for slower, less intensive tasks while the Cortex A-57 handles the more heavy-duty jobs.

One interesting note. According to the benchmark test, the Samsung Galaxy S6 protoype was equipped with 3GB of RAM despite speculation that the model would come with 4GB of the sweet stuff. Of course, if this is a prototype, there still is a chance that Sammy could squeeze 4GB of RAM into the final version, although the AnTuTu benchmark test also revealed 3GB of RAM inside the Galaxy S6. The version of the Galaxy S6 that went through that benchmark test was powered by the Snapdragon 810 CPU.

source: Geekbench via SamMobile

61 Comments
  • Options
    Close




posted on 26 Jan 2015, 18:41 7

1. Berzerk000 (Posts: 4273; Member since: 26 Jun 2011)


For comparison, the iPad Air 2 with the tri-core A8X gets 1800 in single core and 4500 in multi-core.

posted on 26 Jan 2015, 18:42 3

2. rd_nest (Posts: 1591; Member since: 06 Jun 2010)


Don't forget the TDP!!

Though I posted this in another article sometime back..

posted on 26 Jan 2015, 18:55

11. Neros (Posts: 1016; Member since: 19 Dec 2014)


What is Tdp?

posted on 26 Jan 2015, 19:08 1

13. Berzerk000 (Posts: 4273; Member since: 26 Jun 2011)


Thermal Design Power. It's the amount of heat the CPU/GPU can generate while still being "safe." If the CPU/GPU comes close to this threshold, it will throttle itself (reduce its speed) in order to keep itself from over heating.

posted on 26 Jan 2015, 19:09 1

14. tedkord (Posts: 10231; Member since: 17 Jun 2009)


Thermal design power. It's about the maximum heat generated by the CPU.

posted on 26 Jan 2015, 19:45

19. EC112987 (Posts: 523; Member since: 10 Nov 2014)


Snap 810 TDP blows up in hand lol....just poking fun

posted on 26 Jan 2015, 19:31 4

18. vincelongman (Posts: 4102; Member since: 10 Feb 2013)


Thermal Design Power
Like others have said, its the heat generated by the SoC/CPU/GPU
The in phones will have TDPs of about 2-3W
While tablets will have TDPs of about 4-5W

Which makes this more impressive
Since its single core is only 300 slower than the A8X and Denver K1 (both about 1800)
And the multi core is much higher than the A8X (4500) and Denver K1 (3000)

But some may say its not surprising since the 7420 is rumored to be 14 nm, while the older A8X is 20 nm and the K1 is 28 nm

Its going to be interesting to see the 7420 vs A9X vs X1

posted on 26 Jan 2015, 20:17

21. Salazzi (Posts: 537; Member since: 17 Feb 2014)


and don't forget the A8 and A8X are only dual-core. Single core the A8 and A8X beat the 7420 and if they had more cores, they'd also win in the multi-core score as well. Regardless though, this is a good chip and I hope it's used in the S6. Most android devices get ~900 single core score.

posted on 26 Jan 2015, 20:35 2

23. Gp_Customs (banned) (Posts: 269; Member since: 14 Jan 2015)


Android uses the Linux kernel and other specialized software to create the Android Operating System. It uses several processes and each process has at least one thread. Multi-threading on a single (and mutiple) processor hardware platform is accomplished by context switching. This gives the illusion of running more than one thread per processor at a time.

Single core is no good on Android, but good for iOS.

posted on 26 Jan 2015, 23:35 3

36. Niva. (Posts: 418; Member since: 05 Jan 2015)


Context switching is actually commonly referred to what happens in cache which is managed by one processing core, meaning if the core is working on a task and a higher priority task comes in. If cache is limited then the cache used by the current task gets overwritten by data needed for the higher priority task, once work is complete on the higher priority task the cache can be freed up and loaded with whatever's required for the next highest priority task which isn't complete. The switching of tasks from one to a higher priority task is context switching.

The illusion of running more than one thread per processor doesn't have to be an illusion if the processor has multiple cores. You're actually referring to hyperthreading when a single core acts like multiple cores, but even that isn't necessarily an illusion as modern CPUs have actual hardware that supports hyperthreading.

Carry on now!

posted on 26 Jan 2015, 20:49 6

25. tedkord (Posts: 10231; Member since: 17 Jun 2009)


Isn't the A8X tri-core? Regardless, it doesn't matter. They don't have more cores. That's like saying if the SD400 ran at 5GHz it would be the fastest mobile CPU in the world. It doesn't, and it isn't.

posted on 26 Jan 2015, 20:53 4

26. sunspider (Posts: 4; Member since: 26 Jan 2015)


You know nothing at all. The apple A8 chip have 2 cores compare to the Snapdragon 805 (4 cores) and Exynos (8 cores) but the size (or the number of transistor ) of the Cyclone core of A8 is bigger than the others (nearly double the Snapdragon core) so its power consumption are also higher.

The thinking that a single core with higher benchmark score is better is not right, it also depends on the size and power consumption of the core. If apple use quad core of the A8, the benchmark score will be higher than other SoC but it will also consume more power.

posted on 26 Jan 2015, 21:10 2

27. Gp_Customs (banned) (Posts: 269; Member since: 14 Jan 2015)


The bigger the core, the more power.

posted on 26 Jan 2015, 21:30

29. Gp_Customs (banned) (Posts: 269; Member since: 14 Jan 2015)


If we had both chips running an identical OS with identical instructions, then we could compare power. You can't run a benchmark in one OS and compare it to a benchmark in another OS to compare pure speed because the benchmark is subject to the OS. Common knowledge is that Android isn't the most efficient OS compared to Apple's, therefore Apple doesn't need as much power to perform the same operations. Either that common thought is wrong, or you are in this case. If the Apple A8 was just as powerful as the Snapdragon 805, then on an Apple OS we would be seeing great differences in benchmark scores, YET the differences in benchmark scores are relatively small.

This means that either the Apple A8 isn't as fast as people are hyping it to be, or Apple's software is a lot slower than people make it out to be - either way, people are trying to hype something being better than it really is...

posted on 26 Jan 2015, 21:28

28. Gp_Customs (banned) (Posts: 269; Member since: 14 Jan 2015)


I thought it was fairly common knowledge that Apple uses less-specced hardware with more optimized software to achieve their results. This has always been true with their Apple computers and I'm not sure why they'd completely change their strategy when it comes to phones. Their phones are less powerful but more efficient, which contributes to why they have much better battery life.

posted on 26 Jan 2015, 21:31 3

30. tedkord (Posts: 10231; Member since: 17 Jun 2009)


IPhones haven't had better battery life for quote a few years now.

posted on 26 Jan 2015, 21:40 2

31. Dominique3392 (Posts: 10; Member since: 11 Dec 2013)


Does it matter which chip is faster? what matters is the end users experience and after using both lolipop and ios 8 i can tell you ios 8 is way more optimized and runs much better. even though android has great customization i still prefer ios. and optimizations arent everything. theyre sitll fast. the cpu may not be cutting edge but it can still hold its own to the a57 cpus due to optimizations whereas samsung and htc etc just paste in an 810 without further optimizations the gpu is what matters and thats why ios 8 performs so well.

posted on 26 Jan 2015, 23:46

38. Niva. (Posts: 418; Member since: 05 Jan 2015)


What are you talking about? In the phone market Apple has historically been the dominant player in terms of hardware and access to the latest tech. Only in the last few years have non-Apple devices really caught up and in some cases surpassed Apple in terms of raw power. Efficiency is a product of a number of things, not just the CPU, and it's important to note the task for which efficiency is being measured. You mention in an earlier post that "the bigger the core the more the power" but what are we comparing? Newer hardware is produced with much finer circuitry due to the constant manufacturing node shrinking. A newer chip is typically smaller, runs at higher frequency, has more circuits and uses less power... yet at the same time it's actually more powerful than the bigger chip from the previous generation in terms of number crunching. On the same manufacturing node a chip with more circuitry (actually bigger) may use more power at peak utilization, but it may also complete a task much faster than a smaller less sophisticated chip and go to idle much sooner basically mostly shutting itself down, this results in the bigger and more sophisticated chip actually being more efficient.

posted on 26 Jan 2015, 23:20

35. alex3run (Posts: 710; Member since: 18 May 2014)


If Apple just uses more cores it wil be faster than Exynos 7420 but... hotter. Apple cores are very big even compared to Cortex A57 cores.

posted on 27 Jan 2015, 00:53 2

43. Macready (Posts: 831; Member since: 08 Dec 2014)


No, the A8X has 3 cores.

posted on 27 Jan 2015, 07:25

52. j2001m (Posts: 2788; Member since: 28 Apr 2014)


Hm, the a8x is. 3 cores

posted on 27 Jan 2015, 01:14 1

46. LifeSucks (Posts: 54; Member since: 15 Jan 2015)


Correct me if I am wrong but according to what I have read the so called "14 nm" transistors are actually 20 nm transistors that are FinFET instead of MOSFET. So the process isn't smaller than 20 nm but it's a lot more power efficient, especially for these ultra low power SoCs.

posted on 27 Jan 2015, 01:29 1

47. alex3run (Posts: 710; Member since: 18 May 2014)


I saw an article in which TSMC 16 Nm FinFet and FinFet Plus were compared to Samsung 14 Nm and Intel 14 Nm. Sorry, I don't remember where to find this article, but the conclusion was following: TSMC 16 Nm was actually 20 Nm node with 3D transistors; TSMC 16 Nm FinFet Plus and Samsung 14 Nm are noticeably thinner (they are almost equal), but Intel 14 Nm is the thinnest process node.

I don't think Samsung uses TSMC factories to produce its SoC-s.

posted on 27 Jan 2015, 01:07

44. hondaslzman (Posts: 78; Member since: 29 Dec 2014)


Total dog poop

posted on 16 Feb 2015, 18:49

63. KRONeage (Posts: 142; Member since: 17 Apr 2011)


Don't YOU forget this is the first chip to ever come off Samsung's extremely efficient 14nm 3D Stacked chip fabrication process! ......then you can start comparing TDP differences between the 20nm chips and these 14nm chips! ;-P

posted on 26 Jan 2015, 18:54

9. j2001m (Posts: 2788; Member since: 28 Apr 2014)


Yes but all 3 cores run at the Same speed and runs iOS, so means nothing

posted on 26 Jan 2015, 19:02 2

12. Berzerk000 (Posts: 4273; Member since: 26 Jun 2011)


The purpose of cross platform benchmarks like Geekbench is to see the raw power of the CPU/GPU. If they didn't have some indication of how SoCs compare to each other, they wouldn't exist. Even though it is a different platform, it's still pretty accurate.

posted on 26 Jan 2015, 19:15

16. joey_sfb (Posts: 5330; Member since: 29 Mar 2012)


iOS device are never short of CPU/GPU power. 2GB Ram would be more helpful for games and task switching.

posted on 26 Jan 2015, 20:44

24. Gp_Customs (banned) (Posts: 269; Member since: 14 Jan 2015)


Every “task” that’s run on a thread can be broken down into multiple instructions. These instructions don’t have to happen all at once. So a single-core device can switch to a thread “1” finish an instruction 1A, then switch to thread “2” finish an instruction 2A, switch back to 1 finish 1B, 1C, 1D, switch to 2, finish 2B, 2C and so on…

This switching between threads happens so fast that it appears, even on a single-core device, that all the threads are making progress at exactly the same time. It’s an illusion caused by speed, .

But iOS does not function like that.

posted on 27 Jan 2015, 12:10

58. Iodine (Posts: 1303; Member since: 19 Jun 2014)


Yeah, because you know iOS devices have a 6-way superscalar design in them, which can actually process 6 instructions at once.

posted on 26 Jan 2015, 18:55

10. engineer-1701d (Posts: 3259; Member since: 13 Mar 2014)


who the hell thinks this is the gs6 samsung would not risk the future of there mobile line on 3gb ram and a 1.5gh cpu with that bad of a single core score.
this will be like the gs3 they put 2gb ram in the phone for the usa model and pushed all limits with the specs and it paid off the 3rd phone since the gs3 will be a spec beast with 4gb ram since the chinese are doing it in the note pro. and others.

posted on 26 Jan 2015, 19:15 3

15. Berzerk000 (Posts: 4273; Member since: 26 Jun 2011)


That is definitely not a bad single core score. It's better than the Exynos 5433 by ~300 points (+25%), and the multi-core score is astronomical, the best Android has ever seen by ~1500 points, +40% compared to the Exynos 5433.

posted on 27 Jan 2015, 04:55

51. teenaxta (Posts: 42; Member since: 10 Sep 2014)


well considering the amazing single core score of this chipset, multicore score is quite disappointing

posted on 27 Jan 2015, 07:34

53. j2001m (Posts: 2788; Member since: 28 Apr 2014)


Nope as 4 of the cores have less power

posted on 26 Jan 2015, 18:42

3. hung2900 (Posts: 949; Member since: 02 Mar 2012)


So 7420 is basically the same as 5433 but performs much better?

posted on 26 Jan 2015, 18:47 4

5. rd_nest (Posts: 1591; Member since: 06 Jun 2010)


Nope, not same. First, it's a different class of memory. 2nd, 5433 didn't even have AArch64. It was like a car which has the muscles of a A57, but didn't have a driver than can run it more than a A15.

posted on 27 Jan 2015, 07:36

54. j2001m (Posts: 2788; Member since: 28 Apr 2014)


Yes to the mem, but I think Samsung also damaged it so to cut it back to the 805

posted on 28 Jan 2015, 11:42

60. thegeneral7010 (Posts: 319; Member since: 10 Dec 2014)


so isn't this gonna change when android L reaches note 4 and 64bit capabilities activiated

posted on 26 Jan 2015, 18:44 3

4. rd_nest (Posts: 1591; Member since: 06 Jun 2010)


And 3GB RAM is more than enough for anything.. Yea, anything! if it doesn't work with 3GB, I doubt any additional RAM is going to make any difference.

posted on 26 Jan 2015, 18:50 2

6. Berzerk000 (Posts: 4273; Member since: 26 Jun 2011)


Samsung said they were going for a Nexus-like experience with the next version of Touchwiz, so 3 GB is definitely enough as long as they aren't lying.

posted on 26 Jan 2015, 19:25 1

17. tedkord (Posts: 10231; Member since: 17 Jun 2009)


Individually, sure. But more ram means better multitasking.

posted on 26 Jan 2015, 20:22

22. Chuck007 (Posts: 1035; Member since: 02 Mar 2014)


^Pretty much this. My Z3 didn't have any problems with it's 3GB RAM with Android Kitkat. Yes there was optimization problems with GTA Vice City but that's just lazy programming on Rockstar's part.

posted on 26 Jan 2015, 18:53 1

7. Neros (Posts: 1016; Member since: 19 Dec 2014)


S6 edge might get 4gigs of ram, but not the base model.

posted on 26 Jan 2015, 18:54 2

8. Mube007 (Posts: 235; Member since: 24 Sep 2013)


S6 designing will decide the future of s series

posted on 26 Jan 2015, 20:15

20. AfterShock (Posts: 3698; Member since: 02 Nov 2012)


I'd like to see Exynos be marketed to HTC an possible LG at some point too.
Wishful thinking maybe.

posted on 26 Jan 2015, 21:52 2

32. joey_sfb (Posts: 5330; Member since: 29 Mar 2012)


HTC and LG does not use Exynos by choice. They are afraid of Samsung having control over their component supply chain.

Samsung is also their greatest competitors so they don't want to feed their enemy.

posted on 26 Jan 2015, 23:59 1

39. tedkord (Posts: 10231; Member since: 17 Jun 2009)


Then why does Apple use them so much?

posted on 27 Jan 2015, 07:38

55. j2001m (Posts: 2788; Member since: 28 Apr 2014)


As Samsung the only one that can deal with the numbers

posted on 26 Jan 2015, 22:27 3

33. Freecableguy (Posts: 42; Member since: 21 Jul 2014)


This MT score simply blows other mobile SoCs out of the water. Even the ST score almost matches the overhyped Apple A8. Amazing job by Samsung again, I can understand why they ditched Snapdragon 810.

posted on 26 Jan 2015, 23:19

34. Tsepz_GP (Posts: 978; Member since: 12 Apr 2012)


4GB RAM is highly doubtful, the Note 5 will bring 4GB RAM, not the S6, IMO. Never has a Galaxy S brought more RAM than the Note released just before it. The Note 2 came with 2GB RAM then the S4 came with 2GB RAM, the Note 3 came with 3GB RAM then the S5 came with 2GB RAM.

posted on 26 Jan 2015, 23:39 2

37. Mechanized (Posts: 13; Member since: 31 May 2014)


The Snapdragon variant of the Galaxy S3 offered 2GB of RAM, while the first Note offered 1GB.

posted on 27 Jan 2015, 00:30

41. Suo.Eno (Posts: 480; Member since: 17 Feb 2013)


One step at a time maybe is the best way moving forward. Code just isn't there yet for proper L/aarch64 on 4GB RAM save for maybe x86-64 Intel builds. The latter if paced well should be a boost for say Zenfone 2 to start.

posted on 27 Jan 2015, 00:15

40. NihonDaisuki (unregistered)


ステマか?

posted on 27 Jan 2015, 00:52

42. realjjj (Posts: 361; Member since: 28 Jan 2014)


So likely 2.4 to 2.5GHz and maybe DDR4 since they gain some 25% over the 5433 and SD810 in single core.

posted on 27 Jan 2015, 01:09 1

45. hondaslzman (Posts: 78; Member since: 29 Dec 2014)


All Isheeps get out of here, you guys are too stupid to use Android

posted on 27 Jan 2015, 01:44 1

48. livyatan (Posts: 867; Member since: 19 Jun 2013)


As expected .. The multi core score is on par or higher than top of the line 14nm Intel Core M!! (Broadwell Y).

Looks like Samsung will rule, even the A9 in next iPhone won't score higher in multi thread (since it will probably be either a 3 core version of A8X shrunk to 14nm, either the upclocked dual core A8 enhancement)

posted on 27 Jan 2015, 03:03

49. wando77 (Posts: 808; Member since: 23 Aug 2012)


Aren't these amazing scores on the slower cores? (Says 1.5ghz)

posted on 27 Jan 2015, 09:07

56. shalinsaini29 (Posts: 5; Member since: 14 Nov 2014)


Well impressive scores from a exynos

posted on 28 Jan 2015, 11:40 2

59. thegeneral7010 (Posts: 319; Member since: 10 Dec 2014)


i wish if samsung produce s6 in only one variant (exynos variant) this will help it to develop many things in it more than cpu and gpu this will help samsung to insert its innovations in it samsung innovations in processors were ristricted before they couldnt make devlopments like 4k video streaming and 4G modems because they were restricted by they half other variant which was qualcomm, samsung wait to see wat qualcomm does in there processors then it try to copy that in exynos but now all of this is changed they make only 1 variant which is their homemade exynos so they can make develops than any other company cant make and iam sure that samsung can make it.

posted on 28 Jan 2015, 15:59

61. MagicHead23 (Posts: 1; Member since: 28 Jan 2015)


I agree, I really don't want to buy a Galaxy S6 without an Exynos 7420 anymore. It would be smart for Samsung to just stick with their in-house chips. This benchmark shows that the 7420 clearly outperforms the Snapdragon 810 that is seen in the LG. I hope Samsung isn't stupid enough to use the 810 the Qualcomm is supposedly "fixing" for the S6.

Want to comment? Please login or register.

Latest stories