First Apple Silicon Benchmarks

Yes


  • Total voters
    33

erek

Supreme [H]ardness
Joined
Dec 19, 2005
Messages
5,560
Opinion?

"Compared to modern PCs with native Intel-type chipsets, that’s not all that impressive, but that’s to be expected since it’s emulated. But compared to Microsoft’s Surface Pro X, which has the fastest available Qualcomm-based ARM chipset and can run Geekbench natively—not emulated—it’s amazing: Surface Pro X only averages 764 on the single-core test and 2983 in multi-core.

Right. The emulated performance of the Apple silicon is as good or better than the native performance of the SQ-1-based Surface Pro X. This suggests that the performance of native code on Apple silicon will be quite impressive, and will leave Surface Pro X and WOA in the dust."


https://www.thurrott.com/apple/237225/first-apple-silicon-benchmarks-destroy-surface-pro-x#
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,300
Shouldn't be a surprise at all.

Apple has been making the most capable ARM based desigs for some time now, making the likes of Qualcomm and others in the market look silly.

Now, this particular benchmark is not useful due to the emulation, but the benchmarks which will be really interesting are the native Apple ARM vs. native x86 benchmarks. I think this is what we are all waiting for.
 
Last edited:

erek

Supreme [H]ardness
Joined
Dec 19, 2005
Messages
5,560
These are benchmarks of Rosetta, not the raw performance, so it should be even better. And I’m certain that the chips going into their future Macs will stomp all over this.
Hmm
 
  • Like
Reactions: filip
like this

schmide

Limp Gawd
Joined
Jul 22, 2008
Messages
258
Not sure what the poll is?

Come to think of it, I don't know what Geekbench does? There is no list of operations it performs.

As much as this could tell us something, it doesn't. It could really shine in the compute side which is native and puke on processor side. Apple GPU is far above Adreno.
 

defaultluser

[H]F Junkie
Joined
Jan 14, 2006
Messages
13,530
Come on guys. It's simple.

Yes or no? :D
Yes, I mean, no!

*Flies Away*

I fully support this transition from a performance standpoint, as Apple has been shipping pointlessly-overpowered phones for years. They've e also been castrating what exactly you could do with all that horsepower by giving you half the ram of an Android phone, and 1/4 the ram of a real laptop.

I'm laughing at the fact the folks somehow think you can magically port iOS applications on an operating system without touch support, and not run onto masses of issues. Just because it was compiled for ARM doesn't automatically make it easy to port mukltitouch/motion controls to keyboard and mouse!
 

T4rd

Fully [H]
Joined
Apr 8, 2009
Messages
17,522
Are we voting Erek off the island?

Because that's why I voted yes in this craptacular poll!
The poll is clearly for "Opinion?", which is generally the first/only word Erek can muster for commentary in most of his news articles. Though it seems he typo'd the poll title with "Yes", so my opinion is that "yes" is "no".

Anyways, spanking Qualcomm has been Apple's favorite past-time since the dawn of the iPhone, so it's not surprising to see it continue its ARM domination elsewhere.
 

IdiotInCharge

[H]F Junkie
Joined
Jun 13, 2003
Messages
14,304

IdiotInCharge

[H]F Junkie
Joined
Jun 13, 2003
Messages
14,304
i'm surprised about apple if they're truly about to cement some serious performance on ARM
Getting ARM performance for an appliance isn't hard; ARM just runs the OS and device drivers for application-specific logic. On x86 this mostly isn't done because that logic typically resides in other cores (like GPUs), and desktop workloads are typically different than mobile workloads.

So the real question is how well Apple builds up all the other pieces; the custom logic, and the APIs to access it. In terms of running through branching code, in all likelihood Apple will still be very much behind, but since they're focused on building 'computing appliances' rather than general-purpose computers, that might also be all well and good.

Or not.

For Apple to succeed, they have to correctly enumerate the current computing paradigm and also prepare for the next, because they're going to have to build custom logic for that too. x86 will likely remain at the top in terms of compute flexibility.
 

IdiotInCharge

[H]F Junkie
Joined
Jun 13, 2003
Messages
14,304
yah except that is highly dependent on the compiler
And the workload, and the branch predictor... most ARM implementations succeed in terms of running predictable workloads efficiently; what that means for Apple's ARM products is that they'll run whatever Apple tuned the ecosystem for efficiently, and anything that they allow you to run outside of their tuning will likely run much slower / less efficiently.
 

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
8,745
The poll is clearly for "Opinion?", which is generally the first/only word Erek can muster for commentary in most of his news articles. Though it seems he typo'd the poll title with "Yes", so my opinion is that "yes" is "no".

Anyways, spanking Qualcomm has been Apple's favorite past-time since the dawn of the iPhone, so it's not surprising to see it continue its ARM domination elsewhere.
He wants to know if you are "excited?"

Yes or No

I'll wait to see if I'm excited or not.
 
  • Like
Reactions: T4rd
like this

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
4,631
Getting ARM performance for an appliance isn't hard; ARM just runs the OS and device drivers for application-specific logic. On x86 this mostly isn't done because that logic typically resides in other cores (like GPUs), and desktop workloads are typically different than mobile workloads.

So the real question is how well Apple builds up all the other pieces; the custom logic, and the APIs to access it. In terms of running through branching code, in all likelihood Apple will still be very much behind, but since they're focused on building 'computing appliances' rather than general-purpose computers, that might also be all well and good.

Or not.

For Apple to succeed, they have to correctly enumerate the current computing paradigm and also prepare for the next, because they're going to have to build custom logic for that too. x86 will likely remain at the top in terms of compute flexibility.
Oh man its going to hurt us PC guys when the first desktop Arm macs hit. We are not ready for the pain and agony of knowing our kick ass x86 stuff isn't the best performance option. lol

Apple hired Mike Filippo last year to design the chips that are going into their new machines. My guess their desktop mac chips are going to make Intel and AMD... and all us PC boys cry. There is a slim chance they just clock bump some ipad chips and we all laugh.... na there isn't even a a chance of that.

Filippo has been at work at Apple for over a year... and it seems obvious now he was hired to create a performance desktop ARM CPU. He is well known for his high performance ARM AND x86 designs, the Zeus core he designed while at ARM will just start hitting HPC machines later this year. Most of us are still happy to be running cortex 76 phones. But he also was the lead on 24 core 96 thread Intel chips over a decade ago. Apple didn't hire him away from ARM to clock bump ipad chips.

Whatever Apple is cooking its going to be game changing. Make no mistake these "leaked" benches showing rosette 2 performance on ipad chips... are nice and all. But the CPU they are readying for their Macs will be a completely different animal. I imagine we will see the slim version first in macbooks.... but the full on big cores they put in mac pros in a year or two down the line are going to be much more interesting. (and with the reduced cost I expect some monster multi chip mac pros that will have some massive compute performance)
 

Mchart

2[H]4U
Joined
Aug 7, 2004
Messages
3,866
Considering the performance of these chips already given their low TDP I think it’s going to be a serious contender given that they can put larger and higher TDP parts in desktops compared to a phone with no active cooling and finite power.

IMO Apple is making the switch because they’ve likely already run the numbers and know they can put something out that will be at least equal with whatever Intel could give them at the same TDP.

As for this benchmark, it’s pretty irrelevant. It’s running a current gen iPad SOC that is downclocked this is not a full desktop part.
 

Sycraft

Supreme [H]ardness
Joined
Nov 9, 2006
Messages
4,668
I assume you all have seen this:
https://forums.anandtech.com/thread...owerful-cpu-cores-ipc-ppc-comparison.2580622/

Even if emulation takes a 20% chunk of performance from native, it will be faster clock for clock than x86/64
Well something to be careful of before you get too excited: This is all Geekbench stuff. It is a somewhat questionably useful benchmark. You see it all the time on smartphones, but never anywhere else and that should tell you something. I mean do any of you here use Geekbench to evaluate your CPU purchases for your desktop? Linus Torvalds really doesn't like it and he's the kind of guy who might know a bit about this sort of thing (cross platform development/benchmarking).

You just always want to be wary of performance claims when there is a single benchmark being used, particularly a single synthetic benchmark. Those are often cherry picked to make something look the best and are not that representative of real use. Geekbench in particular has been criticized for not accurately representing a user's experience with a phone.
 

IdiotInCharge

[H]F Junkie
Joined
Jun 13, 2003
Messages
14,304
My guess their desktop mac chips are going to make Intel and AMD... and all us PC boys cry.
I think that's mostly going to depend on the workload, and whether it's been... approved. I think for the box of approved that Apple users allow the company to put them in, it will fly; deviations will exemplify weaknesses.
 

aokman

[H]ard|Gawd
Joined
Jan 3, 2012
Messages
1,033
For anyone who didn’t see it, there was also an interview where they stated the dev kit will not be used as a basis for future Macs and also it shows what they can do when their silicon team isn’t even trying :ROFLMAO:

I am excited to see what the future holds!

 

Auer

[H]ard|Gawd
Joined
Nov 2, 2018
Messages
1,760
I think that's mostly going to depend on the workload, and whether it's been... approved. I think for the box of approved that Apple users allow the company to put them in, it will fly; deviations will exemplify weaknesses.
I expect some pretty kick ass photo/video editing workstations. Who cares if they don't run anything else :)
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,300
I assume you all have seen this:
https://forums.anandtech.com/thread...owerful-cpu-cores-ipc-ppc-comparison.2580622/

Even if emulation takes a 20% chunk of performance from native, it will be faster clock for clock than x86/64
Yeah, but per clock performance is only half the equation. Will they hit high enough clocks to take the lead in typical desktop applications? The A13 that leads the bench you posted has a max clock of 2.66 Ghz.

Generally when you design a CPU for high performance per clock, it is more difficult to hit high clocks, and vice versa. x86 designers have had a lot of time to optimize and maximize getting the most out of both of them.

I could be wrong, but honestly, I'm not buying that Apples ARM design will magically come out of the gate and spank existing x86 designs.

I am totally convinced that they will show the world that ARM is more than just a silly low power phone and embedded CPU and that it is a real contender, but I really kind of doubt that they will perform better than either AMD or Intel in typical desktop apps.

Time will tell though.
 
Last edited:

clockdogg

[H]ard|Gawd
Joined
Dec 12, 2007
Messages
1,049
why am i getting votd off the island? pendragon1 any comments on this?
Possibly because we have been told by the Head Insider from the Intelligensia that benchmarks don't matter. And yet you flaunt the industry leader's prime directive with a poll about... benchmarking. That sort of flaunting can only confuses the locals. ;-)
 

aokman

[H]ard|Gawd
Joined
Jan 3, 2012
Messages
1,033
Yeah, but per clock performance is only half the equation. Will they hit high enough clocks to take the lead in typical desktop applications? The A13 that leads the bench you posted has a max clock of 2.66 Ghz.

Generally when you design a CPU for high performance per clock, it is more difficult to hit high clocks, and vice versa. x86 designers have had a lot of time to optimize and maximize getting the most out of both of them.

I could be wrong, but honestly, I'm not buying that Apples ARM design will magically come out of the gate and spank existing x86 designs.

I am totally convinced that they will show the world that ARM is more than just a silly low power phone and embedded CPU and that it is a real contender, but I really kind of doubt that they will perform better than either AMD or Intel in typical desktop apps.

Time will tell though.
Yeaaaah na lol, Apple have been working on this for probably 5yrs behind the scenes... they wouldn’t be replacing Intel unless they could beat them. People will be in for a shock when they hit the market.
 

erek

Supreme [H]ardness
Joined
Dec 19, 2005
Messages
5,560
Possibly because we have been told by the Head Insider from the Intelligensia that benchmarks don't matter. And yet you flaunt the industry leader's prime directive with a poll about... benchmarking. That sort of flaunting can only confuses the locals. ;-)
am i really getting cancelled on here? be honest
 
Top