Whopping Exynos 8890 benchmarks tipped, Galaxy S7's chipset compared with Apple A9

Whopping Exynos 8890 benchmarks tipped, Galaxy S7's chipset compared with Apple A9
Rumored for development completion by year's end, the Galaxy S7 has been testing with both Samsung's own Exynos chipsets, and with Qualcomm's Snapdragon 820, that much we've been tipped already. Samsung is pairing its Exynos with LTE modems as well, and we unearthed two purported Galaxy S7 versions on Geekbench - one dubbed Project Lucky-LTE, and the other simply Project Lucky, both powered by Exynos 8890 (codenamed M1 or Mongoose).

How about an Exynos 8890 running at its purported 2.4 GHz maximums, though? Well, a renowned chipset leakster account over on Weibo has tipped the breathtaking results of such configuration - 2304 points on the single-threaded, and the whopping 8038 points on the multi-threaded Geekbench cross-platform test. To put these results in perspective, the Note5 with the latest Exynos 7420 is clocked at 2.1 GHz, and scored 1431 and 4717 points, respectively, when we ran Geekbench 3 on it.

When it comes to the newest Apple A9 chipset, however, the source says that Exynos 8890 keeps a similar distance from its performance, as the current Exynos 7420 keeps from Apple A8. They purportedly got a 1.75 better result in multi-core performance for Exynos 8890 vs Apple A9, and 1.1 times weaker result in single-core showing. 

To some extent this can be attested by this iPhone 6s benchmark chart on the same test below, though the results vary with each run. Let's not forget, however, that Apple's A9 is reportedly clocked at lower, 1.8 GHz frequency, so it will be much faster on a clock-per-clock basis, at least in a single-threaded performance.

Whopping Exynos 8890 benchmarks tipped, Galaxy S7's chipset compared with Apple A9

source: Weibo via G4Games & MobileSyrup

FEATURED VIDEO

105 Comments

1. Nathillien unregistered

That graph needs to be updated - There are so many Geekbench test of Note 5 scoring much better - from 4700 to 5000. Additionally single-core scores don't mean much, since all smartphones, apps and OS's are well optimized to work with multiple cores when necessary. Anandtech had recently a good research on that matter.

18. nctx77

Posts: 2540; Member since: Sep 03, 2013

I'm sure you can get higher scores on the A9 as well.

30. arch_angel

Posts: 1651; Member since: Feb 20, 2015

Actually The iPhone's Benchmarks Don't Vary As Much As Android Phones. For The iPhone 6/ 6 Plus Geekbench Has Their Highest Score Out Of 15k Test Results As 2,944 And Their Lowest Score As 2,926. Taking A Look At The Note 4's Results In Geekbench It's Highest score Out Of 175k Results Is 5,313 And The Lowest As 4,545. So Based On That The iPhone 6s Probably Want Get Much Better Scores Than It Already Has. IOS Devices Are Better Optimized To Get The Most Out Of Their Hardware Than Android Is. Used The Note 4 As My Android Example Because The Note 5 And S6 Don't Have Any Real Search Results In Geekbench Browser IDK Why.

33. Ordinary

Posts: 2454; Member since: Apr 23, 2015

No its cause people benchmark samsung with UPSM and Power Saving mode hence the lower scores.

44. arch_angel

Posts: 1651; Member since: Feb 20, 2015

You Can't Even Open Apps In UPSM And The Phones Have Software That Makes It Perform It's Best When Running A Benchmark Test, Doesn't Matter If Power Saving Is On Or Not.

53. Ordinary

Posts: 2454; Member since: Apr 23, 2015

Actually you can open apps even in UPSM, you can choose which one. Thats false. No matter what you are running benchmark/game it will always run cores at lower rate. Even when you flash a kernel and change CPU speed power saving mode will break it on its own. Tested many time on my S4 (power saving mode).

62. arch_angel

Posts: 1651; Member since: Feb 20, 2015

On My GS6 You Can Run One App In UPSM So You're Right About That. And I Don't Own A Note Four To Prove If Its Lowest Score From Geekbench Is Indeed From One With Power Saving Mode Used During Testing. I Can Say That My Phone's Results Are Vastly Different With It On.

68. Ordinary

Posts: 2454; Member since: Apr 23, 2015

Well maybe not UPSM but power saving is enabled for sure or it was running more benchmarks before and it throttled.

86. vincelongman

Posts: 5628; Member since: Feb 10, 2013

"For The iPhone 6/ 6 Plus Geekbench Has Their Highest Score Out Of 15k Test Results As 2,944 And Their Lowest Score As 2,926" Nope, just search for iphone 6 plus There plenty of scores lower than 2,926 http://browser.primatelabs.com/geekbench3/search?q=iphone+6+plus

100. arch_angel

Posts: 1651; Member since: Feb 20, 2015

The 6 Plus Also Has Scores That Say 4k+ But They Aren't True Because Geekbench Also Says That The A8 In The 6 Plus has 3 Cores. And Those Low Scores Are Most Likely The Result Of Back To Back Benchmarks. I Admit My Statement Wasn't Completely True And What I Should Have Said Is The Average iPhone 6 Geekbench Score Doesn't Vary By More Than 200 Multi Core Score But Android Phones Can Vary In Much Larger Amounts Like 700 Without Throttling Or Power saver Mode On. Cause Power Saver Drops The Average Score Of My S6 By More Than 1,000.

102. vincelongman

Posts: 5628; Member since: Feb 10, 2013

I agree that the Apple SoCs tend to vary less Also on the Android side you have custom kernel, can cause even more changes (different governors, clock rates, ...) Just your number I don't see where your numbers came from AFAIK Geekbench don't release average scores, highest lowest scores or total number of scores

105. arch_angel

Posts: 1651; Member since: Feb 20, 2015

I Forgot About Those Aspects Of Android And Yes I Made A Mistake So My Numbers Are Messed Up LOL.

120. juxt417

Posts: 6; Member since: Feb 14, 2015

You have to search by the model number sm-g925r4 for the s6 edge you can google the rest.

125. mitchytan92

Posts: 92; Member since: Mar 02, 2015

I think the problem is thermal throttling on the Exynos. Apple result is very stable because their processor isn't so hot. Exynos on the other side is always keep throttle down the speed when it is hot. You can keep running Geekbench benchmark non stop to test it for yourself. My G925I results (Running non stop) 5206 5134 4899 4967 4934 4961 4902 4888 4821 4659 4511 4561 4670 4574 4565 You see that at first it is scoring very well. Bot over time as the processor gets heat up, the throttling kicks in and the speed goes down and down to the 4500 range. But still the Exynos is stronger than A9 but seeing how well the A9 perform in real life, I guess single core performance is more important than multi cores.

126. arch_angel

Posts: 1651; Member since: Feb 20, 2015

single core is more important for ios, multi is more important for android. that is way they differ so much when it comes to core count.

127. coldspring22

Posts: 349; Member since: Feb 28, 2015

For future performance improvements, Intel has shown that trying to improve single core performance continuously does not scale in the long run. Multicore approach is the way forward, otherwise, PC today would be running on 100Ghz Pentium 4.

134. Emacs244

Posts: 9; Member since: Oct 29, 2015

Throttling. Both qualcomm and samsung forced to overly push their CPU's clocks to compete with apple. Throttling shows how much better apple cpu dedign is.

89. engineer-1701d unregistered

if only samsung would look past there ego for once and dump or cut the ui in half or stock android m then this would not be a problem.

121. juxt417

Posts: 6; Member since: Feb 14, 2015

They have done this with the this years version of touchwiz. Everything you want and almost everything else has to be downloaded from the galaxy app store.

98. rubyonrails3

Posts: 369; Member since: Oct 01, 2014

Let me guess you are an Android user?

135. Emacs244

Posts: 9; Member since: Oct 29, 2015

I am afraid you have a little idea about what are talking about. Only a minority of algorithms can be effectively paralellized. I am currently writing from old Nexus 5 device powered with snapdragon 800 and the simple js (or even html) editor lags. I don't know the real reason behind it, but I am sure if all 4 cores were utilized it will run smoother. Unfortunatelly, blink web engine (powers the chrome) is single-threaded. 60% better performing core is generally better than two cores. Multi-core software development is a source of errors and better be avoided whenever possible.

2. medicci37

Posts: 1361; Member since: Nov 19, 2011

Hope Samsung doesn't decide to put a 820 in their flagships. If so performance won't be much better than what you get out of the S6

3. mdsheik

Posts: 52; Member since: Sep 03, 2015

I dont trust samsung until original review comes out. They ll cheat like volkswagon cheated in carbon emission test. Softwares can give whooping scores.

10. Hallyu

Posts: 790; Member since: Jul 21, 2015

Cheating? Are you saying that every tech sites cheated you too? since they also accept that the Exynos is superior to or at least on par with Snapdragon. And keep in mind that the SD 820 is based on Samsung's 14nm tech that proves which one has better tech.

11. Napalm_3nema

Posts: 2236; Member since: Jun 14, 2013

It's Samsung's manufacturing technology, not Samsung's design, that Qualcomm is using. Until we actually see this Mongoose SoC, Samsung has never designed their own SoC, they have just usedARM reference designs.

13. Hallyu

Posts: 790; Member since: Jul 21, 2015

And how many companies actually design its own cpu besides ARM? At least Samsung has an edge over its competitors by manufacturing its own chip which has the best performance and the leading manufacturing process.

31. Napalm_3nema

Posts: 2236; Member since: Jun 14, 2013

Qualcomm, Apple, and Nvidia, off the top of my head.

35. sipha

Posts: 439; Member since: May 12, 2012

you got 0 out 3!!!

41. Hallyu

Posts: 790; Member since: Jul 21, 2015

LoL. Aren't they using ARM architecture? What are you smoking?

74. Napalm_3nema

Posts: 2236; Member since: Jun 14, 2013

They are using their own custom architecture based off ARM designs., unlike Samsung, Huawei, and others, who use stock ARM designs. Be flippant all you like, but it is obvious that the dumbass in this conversation is you. https://en.wikipedia.org/wiki/ARM_architecture#Architectural_licence

Latest Stories

This copy is for your personal, non-commercial use only. You can order presentation-ready copies for distribution to your colleagues, clients or customers at https://www.parsintl.com/phonearena or use the Reprints & Permissions tool that appears at the bottom of each web page. Visit https://www.parsintl.com/ for samples and additional information.