The first Apple A10x Geekbench4 score?

May 2, 2008
151
2
116
www.chip-architect.com
#1
http://browser.primatelabs.com/v4/cpu/3036382

We should see more soon.

If the frequency of 2.36 GHz is correct then the single thread
score of 3832 would significantly extend the already monumental
lead in IPC of the Apple designed cores/SOCs (edit) with much
the increase coming from the higher memory bandwidth though.
(See the link in the next post)
 
Last edited:

Nothingness

Golden Member
Jul 3, 2013
1,874
30
106
#2
Nov 4, 2012
156
0
81
#6
http://browser.primatelabs.com/v4/cpu/3038861

faster than skylake with equal clock?

is it possible?
Yeah, seems that way, they were already close with the previous iPad pro and this gets a 30% boost. Obviously you need to inspect the benchmark closely so things like memory, AES etc don't distort things too much but GB 4 is a much better benchmark than GB 3 was.

They don't clock as high as intel cores though so they can make them wider etc

A better comparison I think would be the new kaby lake Y processors.
Don't look at the frequency though, that chips is a 1.2Ghz with 3.2Ghz turbo and has approximately the same TDP as the iPad and both are fanless.

http://browser.geekbench.com/v4/cpu/compare/3036382?baseline=2922710

Edit because I finally figured out how to do the side-by-side comparison thingy.
 
Last edited:
Mar 11, 2000
22,585
0
106
#7
I figured the 3 cores of the A8X were going to be a one-off, but here we go again with A10X.

As for the comparison with Kaby Lake Y Core i5, it's not really the performance so much as the functionality. I still can't use the iPad or iPad Pro as an effective business productivity tool. The OS and software aren't quite there yet. Thus, I will be buying a laptop with Kaby Lake Y.

Things are different for creative content creation, but I'm not an artist.
 

Valantar

Golden Member
Aug 26, 2014
1,723
1
136
#8
That score pretty much exactly matches my Ryzen 1600X (w/ 2133MT/s DDR4). Which runs at 3.7GHz all-core boost and 4.1GHz single-core boost. Well, actually, it beats it by 125 points. Which tells me one of three things: either this benchmark is exceptionally well tuned/optimized for Apple's architecture, it's really, really badly optimized for Windows/X86/AMD64, or it simply plays to the strengths of Apple's mobile chips in some very, very specific ways. Likely a mix of all three, to be honest. I have no doubt Apple's architecture is really, really good, but it's not that good.
 
Feb 5, 2006
32,949
155
126
#9
Apple is better than Intel at the tablet or below power envelope.
 

FIVR

Diamond Member
Jun 1, 2016
3,753
11
106
#10
Apple has had a superior architecture to intel since A8X. The fact that skylake is slower at the same clock should be news to nobody, but of course this is Anandtech. People have a hard time believing any benchmark that doesn't show intel in a good light.


And it's not because "Intel chips clock higher and it's a design decision". Apple is a superior chip designer with superior chips. For christ sake, they put this thing in a 6mm wide iPad. Do you have any idea what 6100-U would clock in that form factor? LOL!
 

SAAA

Senior member
May 14, 2014
398
0
56
#11
That score pretty much exactly matches my Ryzen 1600X (w/ 2133MT/s DDR4). Which runs at 3.7GHz all-core boost and 4.1GHz single-core boost. Well, actually, it beats it by 125 points. Which tells me one of three things: either this benchmark is exceptionally well tuned/optimized for Apple's architecture, it's really, really badly optimized for Windows/X86/AMD64, or it simply plays to the strengths of Apple's mobile chips in some very, very specific ways. Likely a mix of all three, to be honest. I have no doubt Apple's architecture is really, really good, but it's not that good.
Great point.To be honest I was waiting this year for Ryzen to show up on geekbench exactly for this... in the end I wasn't surprised.
The change in scores over piledriver reflected the gains nicely, matching other benches, what's funny is that A10 and A9 still runs circles around both latest and greatest x86 architectures.

Some have mentioned Keller (sorry for citing only him again, hundreds of people who have designed processors over all the years) as some figure who magically enhanced Apple's chips to this level, but it seems weird when another core developed by him too isn't up to par.

So yes, either Geekbench really loves A-series arch or the much higher IPC must hit rock hard limits on clocks at a point. Not that it matters much when at 2.4GHz as you say it matches ~4GHz desktop chips... the only vague advantage Intel has is they can reach 5GHz, so absolute single thread performance is still their win.
 

dark zero

Platinum Member
Jun 2, 2015
2,499
4
61
#12
I guess the time to see ARM based MACs are near here... Oh boy, Intel would be royally screwed.
 
Mar 11, 2000
22,585
0
106
#13
I wonder how enormous A10X is. A10 Fusion is already 125 mm2.
 

Nothingness

Golden Member
Jul 3, 2013
1,874
30
106
#14
And it's not because "Intel chips clock higher and it's a design decision". Apple is a superior chip designer with superior chips.!
You're half clueless. But see the good side, you're half correct :D

Some have mentioned Keller (sorry for citing only him again, hundreds of people who have designed processors over all the years) as some figure who magically enhanced Apple's chips to this level, but it seems weird when another core developed by him too isn't up to par.
No higher-end chip micro-architecture is decided by a single engineer. Also all current micro-architectures are converging to very similar designs. The trade-offs are around frequency and power, which depend on the targeted market.

the only vague advantage Intel has is they can reach 5GHz, so absolute single thread performance is still their win.
Indeed.

I still wonder if Apple will try to really get to Intel levels. It's unlikely their micro-architecture can reach as high frequencies as Intel, so they'd have to sacrifice IPC. But nonetheless their current chips are incredibly efficient.
 

ZGR

Golden Member
Oct 26, 2012
1,786
34
106
#15
I thought comparing the overall geekbench scores are very inaccurate when between x86 and ARM?
 

FIVR

Diamond Member
Jun 1, 2016
3,753
11
106
#16
I thought comparing the overall geekbench scores are very inaccurate when between x86 and ARM?
Yes, it is common knowledge that Geekbench4 is biased in favor of x86. A10X would be preform much better in a totally unbiased benchmark, but where can you find those? Almost every benchmark is designed on x86, for x86... and most are funded by intel. It's unfortunate, but we can infer that A10X is actually a much better design than skylake from these scores regardless.
 

IntelUser2000

Elite Member
Oct 14, 2003
5,750
73
126
#17
No higher-end chip micro-architecture is decided by a single engineer. Also all current micro-architectures are converging to very similar designs. The trade-offs are around frequency and power, which depend on the targeted market.
That means team organization and execution starts to really matter. Which based on leaks for Intel and the influence each company has I am betting Apple is far better than Intel. The "halo" surrounding Apple means top engineers move there + Intel keeps shuffling senior and experienced employees which are negative for product development in the longer run. It's not just about having smart engineers with load of degrees.

Back when Moore's Law scaling was healthy, it didn't matter. New process was outright better than older ones. You could tell chips on the new process by how they are grouped on benchmark charts. Not so anymore.
 

itsmydamnation

Golden Member
Feb 6, 2011
1,764
77
136
#18
Has anyone done and GB4 benchmark scaling with memory latency and or throughput. I having a feeling that 4000mhz memory in the apple devices helps massively.
 
Mar 10, 2006
11,709
102
126
#19
Has anyone done and GB4 benchmark scaling with memory latency and or throughput. I having a feeling that 4000mhz memory in the apple devices helps massively.
I'll mess around with it this weekend on my KBL system. I'll test DDR4-2133 -> DDR4-3200 to see if it makes an appreciable difference.
 

itsmydamnation

Golden Member
Feb 6, 2011
1,764
77
136
#20
I'll mess around with it this weekend on my KBL system. I'll test DDR4-2133 -> DDR4-3200 to see if it makes an appreciable difference.
Make sure to test multiple cas values at each memory speed so we can really get a handle on what matters
 

Nothingness

Golden Member
Jul 3, 2013
1,874
30
106
#21
Yes, it is common knowledge that Geekbench4 is biased in favor of x86. A10X would be preform much better in a totally unbiased benchmark, but where can you find those? Almost every benchmark is designed on x86, for x86... and most are funded by intel. It's unfortunate, but we can infer that A10X is actually a much better design than skylake from these scores regardless.
Funny, x86 fans believe Geekbench favors Apple and ARM in general.

I have seen the source code of Geekbench v4 and nothing seems to favor either x86 or ARM: there are specific code paths for GEMM and FFT for AVX(2) and for NEON.

So what about stopping spreading nonsense?

EDIT: public information about Geekbench: http://geekbench.com/doc/geekbench4-cpu-workloads.pdf
 
Last edited:

JoeRambo

Senior member
Jun 13, 2013
638
42
136
#22
Those LLVM and HTML DOM manipulation scores are eye watering. I have no idea what steroids is Apple using, but consensus is that those "GCC and pointer chase" workloads are really hard to cheat at.
 

Valantar

Golden Member
Aug 26, 2014
1,723
1
136
#23
Yes, it is common knowledge that Geekbench4 is biased in favor of x86. A10X would be preform much better in a totally unbiased benchmark, but where can you find those? Almost every benchmark is designed on x86, for x86... and most are funded by intel. It's unfortunate, but we can infer that A10X is actually a much better design than skylake from these scores regardless.
Sorry, but that makes little to no sense. A cross-platform benchmark is by default limited to the features available to the least feature-rich platform (unless similar workloads are performed in drastically different ways across platforms, in which case the actual comparability of the benchmarks is highly debatable. As such, X86/AMD64 (as the most feature-rich architecture on the market) is "doomed" to not be fully utilized in a cross-platform benchmark. Then again, the omitted capabilities might be irrelevant for most end-user workloads, or legacy capabilities not really in use any longer. This stuff is complicated.

Oh, and in terms of GB being X86-biased, that makes no sense. Why then do Apple chips consistently compare better to other architectures here than in most other benchmarks? Are all the other benchmarks even more biased? Where does it end?
 

Herr Kutz

Platinum Member
Jun 14, 2009
2,365
49
106
#24
I guess the time to see ARM based MACs are near here... Oh boy, Intel would be royally screwed.
I think that's a bit of an embellishment. While it's definitely not a good thing for Intel, macs are a small percentage of Intel's processor sales in the grand scheme of things and I don't remember them suffering when Apple used PowerPCs.
 
Mar 11, 2000
22,585
0
106
#25
Apple has a little under 7% of worldwide market share for PC shipments. They're roughly the size of Acer or Asus when it comes to computer shipments. Currently they're in the top 5 worldwide. Here are the top 6 for the last quarter:

Lenovo
HP
Dell
Asus
Apple
Acer

Lenovo and HP are both around 20%. Dell is around 15%, and the other three are around 7%.
 


ASK THE COMMUNITY

TRENDING THREADS