Apple A10 Geekbench 4 Score

[Arachnotronic]

https://browser.primatelabs.com/v4/cpu/146939

3379 single-core, 5495 multi-core.

For reference, here's the top Core m 6Y75 (Skylake) result:

https://browser.primatelabs.com/v4/cpu/116705

3677 single-core, 6983 multi-core.

Apple is putting Core class performance into a slim smartphone. Amazing work.

 

[NTMBK]

I, for one, welcome our new ARM overlords.

 

[witeken]

Apple is putting Core class performance into a slim smartphone. Amazing work.

"Geekbench"

Isn't that a contested benchmark since it's such a synthetic one?

And we already know that the Apple core has high IPC since Cyclone (A7, '13), don't we?

So no surprises there. Real question is performance per watt (apples v apples comparison) and frequency. KBL-Y goes to 3.6GHz.

 

[Arachnotronic]

"Geekbench"

Isn't that a contested benchmark since it's such a synthetic one?

And we already know that the Apple core has high IPC since Cyclone (A7, '13), don't we?

So no surprises there. Real question is performance per watt (apples v apples comparison) and frequency. KBL-Y goes to 3.6GHz.

It's GB4, a much much better benchmark than GB3.

As far as perf/watt...A10 fits in a phone, should tell you all about its perf/watt

 

[jpiniero]

The result is likely fake, although the person that faked it did their homework. It's about 10% faster than the iPad Pro, which at 2.4 Ghz would be just about right.

 

[Arachnotronic]

The result is likely fake, although the person that faked it did their homework. It's about 10% faster than the iPad Pro, which at 2.4 Ghz would be just about right.

How would somebody fake this particular result?

Anyway, I don't see any signs that it's fake, everything checks out.

L1/L2 cache sizes look right, L3$ registering at 0KB is typical for the A-series SoCs. The naming of the logic board/platform "D101AP" is new and looks like a typical Apple board name. iPhone 7 is expected to have 2GB of RAM which is what this shows. And obviously the performance is in line with rumors.

 

[jpiniero]

- The OS is 10.1.
- The string is iPhone9,3, which doesn't really match. Apple has been using X,1 for the "regular" model and X,2 for the Plus. There were rumors of a third model called "Pro" but that seems to have been shot down.
- The Plus is expected to have 3 GB of memory. If this was the iPhone 7 Pro, it would presumably also have 3 GB.

 

[Face2Face]

Pretty beast. About 1000 points faster in single core vs. A9 and over 1000 in muti-core. Apple knows what they're doing when it comes to SoC design. I wish Intel or AMD could compete at the same level with X86 - performance per watt.

 

[Eug]

Yeah, looks fake. Also, the CPU speed is 396 MHz, although that type of error is common when the chip is not directly supported by Geekbench.

The Primate Labs programmer (a fellow Torontonian BTW) can often tell us whether or not such a score is real, but in the meantime I'd say the wait for a legit A10 performance leak continues.

 

[witeken]

Don't really understand actually why people are concerning themselves with all those rumors and benchmarks at this point.

We'll get the correct information within a few days.

Edit: Only 2 measly days, apparently.

 

[Eug]

Don't really understand actually why people are concerning themselves with all those rumors and benchmarks at this point.

We'll get the correct information within a few days.

Edit: Only 2 measly days, apparently.

We probably won't find out the Geekbench results on Wed. We will likely get those the following week... unless there is an early leak.

 

[mikk]

It's GB4, a much much better benchmark than GB3.

As far as perf/watt...A10 fits in a phone, should tell you all about its perf/watt

Are non Windows results still faster?

 

[teejee]

https://browser.primatelabs.com/v4/cpu/146939

3379 single-core, 5495 multi-core.

For reference, here's the top Core m 6Y75 (Skylake) result:

https://browser.primatelabs.com/v4/cpu/116705

3677 single-core, 6983 multi-core.

Apple is putting Core class performance into a slim smartphone. Amazing work.

I've heard that GB4 has longer execution time than GB3. So bursty CPU/SOCs like Core M are not doing as well in GB4.

 

[StinkyPinky]

Just the other day I was wondering how fast the A9 (or snapdragon 820) is compared to a higher end desktop CPU from 10 years ago. It's hard to find info on that.

No denying Apple make some fantastic cpu's.

 

[Arachnotronic]

Just the other day I was wondering how fast the A9 (or snapdragon 820) is compared to a higher end desktop CPU from 10 years ago. It's hard to find info on that.

No denying Apple make some fantastic cpu's.

Core 2 Quad Q9550: https://browser.primatelabs.com/v4/cpu/169767

Apple A9: https://browser.primatelabs.com/v4/cpu/172831

A9 is faster than a C2Q in single thread and close in multi-thread.

 

[dark zero]

Core 2 Quad Q9550: https://browser.primatelabs.com/v4/cpu/169767

Apple A9: https://browser.primatelabs.com/v4/cpu/172831

A9 is faster than a C2Q in single thread and close in multi-thread.

Better to compare to Pentium G3258
https://browser.primatelabs.com/v4/cpu/171563

Still far in ST and MT.

 

[StinkyPinky]

Thanks. Very interesting - my exynos Galaxy S7 gets 1879/5621 in geekbench which puts it on the same level as the high end quad core Core 2 cpu's. That's impressive, amazing how far tech has come in such a short time. It's actually faster than my old Q6600.

 

[Eug]

I'm not sure we can really compare cross platform like this in any meaningful way. What I found earlier versions of Geekbench most useful for was comparing chips within the same platform. eg. iOS vs iOS.

Is Geekbench really better as a cross platform benchmark? And even if it is better, is "better" good enough as a true cross platform test? Call me a skeptic.

 

[StrangerGuy]

Just the other day I was wondering how fast the A9 (or snapdragon 820) is compared to a higher end desktop CPU from 10 years ago. It's hard to find info on that.

No denying Apple make some fantastic cpu's.

Yeah, their CPU division so far actually delivers the hype on such a short product cycle too no less, unlike some other "A" company.

 

[coercitiv]

Apple is putting Core class performance into a slim smartphone. Amazing work.

So, based on the ~40% memory bandwidth improvement over the previous iPhone, what kind of LPDDR4 do you reckon the A10 is using?

 

[StrangerGuy]

So, based on the ~40% memory bandwidth improvement over the previous iPhone, what kind of LPDDR4 do you reckon the A10 is using?

http://www.extremetech.com/computin...unces-new-12gb-lpddr4-dram-clocked-at-4266mhz

Probably that @ 4266 MHz versus 3200 on A9. Plus some other prefetching and caching tweaks.

 

[IntelUser2000]

http://www.extremetech.com/computin...unces-new-12gb-lpddr4-dram-clocked-at-4266mhz

Probably that @ 4266 MHz versus 3200 on A9. Plus some other prefetching and caching tweaks.

Now that's a cutting edge chip. Up until very recently, Intel had that title. It may not seem significant, but you lose technological lead before you lose the main thing. Which is what's happening. On the iPad it'll have 68GB/s system memory bandwidth without resorting to new memory technologies like WIO(does it even exist anymore?).

Too bad looking at iPad results that you can't really take advantage of it. It would have been an awesome x86 SoC.

Intel Process + Apple CPU + Nvidia GPU on x86.

Side note: For all the hurrah given to having competition, competition is exactly the reason hindering such things from happening. Competition does curb greed though, so you can't have a unified humanity developing towards a single goal either(because having so much power gets to the top peoples' heads). And to be fair, computers are unnecessary luxury when you think about the big picture.

 

[coercitiv]

Intel Process + Apple CPU + Nvidia GPU on x86.

Side note: For all the hurrah given to having competition, competition is exactly the reason hindering such things from happening.

Apple CPUs are the product of competitive practices, so unless you have a very elegant solution to the grandfather paradox...

 

[IntelUser2000]

Apple CPUs are the product of competitive practices, so unless you have a very elegant solution to the grandfather paradox...

Not talking about time travel at all. Not that I agree its possible anyways but off topic.

We could do it even in the future. Of course barring world-changing circumstances that unify everyone that's not happening. And its unlikely people will rally around making fastest computers.

Besides, we wouldn't realize it if it worked as ideally as I said, and people would still be complaining.


I wonder how far Apple can push this. Or will they stop since every low hanging fruit is exhausted. They just caught up incredibly rapidly. Really the sad part is that super fast Tablet will still be without software support.

 

[beginner99]

I'm looking for a new phone right now and I wasn't too impressed with the performance increases for simple everyday browsing use compared to 4 yard old one (Snapdragon s4 pro). Webpages do load faster no question but it wasn't that impressive. Often the network speed is the more limiting factor which you have 0 influence on.