Apple to replace Intel and move to ARM - *** Confirmed as "Apple Silicon" ***

LiE

LiE

Caporegime
Joined
2 Aug 2005
Posts
25,641
Location
Milton Keynes
Lots of assumptions in that video. Rumours are that A14 is going to be a big leap forward compared to A13.

Totally, but based on previous history of performance gains while being a little conservative. So not completely unrealistic that Apple could deliver these results.
 
Soldato
Joined
6 Oct 2009
Posts
3,998
Location
London

Some very interesting stuff in this video. Apple has the potential to bring some serious power with their new chips.

This screenshot is telling. Obviously it’s based on previous performance and extrapolating a bit, but damn. The score at that TDP. This isn’t even the Mac chip but an iPad chip. Imagine with a higher TDP and active cooling.

aSl15rZ.png

There's a big question mark on scaling though. On single-core performance A13 is already basically as fast as the fastest desktop Intel and AMD chips (at much lower TDP). A14 is expected to be even better.

We've seen Apple scale this to up to 4 performance cores. I expect their strategy is to just add more cores and maybe higher frequency as they go into larger power budgets.

A13 is 98.48mm2 on TSMC's 7nm. There definitely is room to make this bigger.

Apple so far has downplayed their performance, only focused on performance per watt. I think they're preparing for a moment to just blow everyone's minds.
 
Soldato
Joined
6 Oct 2009
Posts
3,998
Location
London
Totally, but based on previous history of performance gains while being a little conservative. So not completely unrealistic that Apple could deliver these results.

Yeah. Rumours say A13 to A14 is going to be like A10 to A11. That's some serious performance increase per core. 40% single core improvement if I recall.
 
Man of Honour
Joined
20 Sep 2006
Posts
34,021
Apple so far has downplayed their performance, only focused on performance per watt. I think they're preparing for a moment to just blow everyone's minds.
Agreed. I can’t wait to see what you can get in a MacBook Pro in a few years time.
 
Soldato
Joined
13 Apr 2009
Posts
6,183
Location
UK
So with Geekbench Pro 5 now running natively, the scores on the DTK Mac mini with A12Z chip are beating the 10th gen 13" MBP with upgraded i7 processor. That's pretty bloody good, especially if we're going to see A14 chip gen which could be even better.

This, combined with the fact it could be a lot better on battery life, and the new rumours about FaceID coming to the Mac means I might just hold out and take the risk on the first-ten Apple Silicon hardware when I buy a new laptop later this year (release timeline dependent).
 
Associate
Joined
7 Apr 2017
Posts
144
Location
London-Amstadam
There's going to be cost savings for apple. Do you think it's possible they may use the move to improve marketshare, perhaps with a lower model, a chepear iBook type of of Macbook? It's probably not likely in the first year or so, but I wonder.
 
Last edited:
Soldato
Joined
6 Oct 2009
Posts
3,998
Location
London
There's going to be cost savings for apple. Do you think it's possible they may use the move to improve marketshare, perhaps with a lowermodel, iBook type of of Macbook? It's probably not likely in the first year or so, but I wonder.

I definitely think Apple will go on to make a sub-$1000 Macbook product that's actually good. It won't be a huge deal, but similar to the £329 iPad or £399 iPhone SE, there can be a £500 Macbook. They could go on to have a very low margin on it, the goal would be to bring people into the Apple ecosystem, who will then are more likely to buy more expensive Apple products.
 
Associate
Joined
7 Apr 2017
Posts
144
Location
London-Amstadam
I definitely think Apple will go on to make a sub-$1000 Macbook product that's actually good. It won't be a huge deal, but similar to the £329 iPad or £399 iPhone SE, there can be a £500 Macbook. They could go on to have a very low margin on it, the goal would be to bring people into the Apple ecosystem, who will then are more likely to buy more expensive Apple products.
Even say off the top of my head £699. Magicboy makes valid arguments about the race to the bottom, margins for most PC laptops are tiny, and a lot of them are pretty terrible at the lowend as a result, and I also appreciate they want to push iPads, but there's still a use sceneario for laptops, and a case to be made for using the transition not just to make bigger profits, but grow the userbase as a long term investment. Chromebooks are a lot cheaper, but they largely took away much of the dominance iPads had in the US educational space, which may pay dividends when those tots one day grow up to purchaise a product for college.
 
Soldato
Joined
3 Jun 2005
Posts
3,065
Location
The South
So with Geekbench Pro 5 now running natively, the scores on the DTK Mac mini with A12Z chip are beating the 10th gen 13" MBP with upgraded i7 processor.

Genuinely not goading but is that the case?

Having a quick look on the Geekbench website, and purely from the first page of results (at the time of writing this) - highest single-core for the A12Z (VirtualApple)¹ is 846², where as the I7-1068NG7³ is 1385⁴; multi-core for the A12Z is 2975⁵ compared to the I7-1068NG7 at 4967⁶.

Obviously it's Geekbench and there's a multitude of reasons, including versioning, to why it isn't completely accurate and you can cut the results however which way but, the above suggests the MBP 2020 2.3GHz i7 (I7-1068NG7) is outperforming the Apple Silicon (happily 'eat my hat' if i've fudged it somewhere with the above).

Similarly the base spec Mac Mini (i3-8100B) is, marginally, "beating" the DTK Mac Mini if looking purely at Geekbench results.

Sources -
1 https://browser.geekbench.com/v5/cpu/search?q=VirtualApple
2 https://browser.geekbench.com/v5/cpu/2923376
3 https://browser.geekbench.com/v5/cpu/search?q=I7-1068NG7
4 https://browser.geekbench.com/v5/cpu/3049171
5 https://browser.geekbench.com/v5/cpu/2874828
6 https://browser.geekbench.com/v5/cpu/3055721


edit - Just to add, i think Geekbench's only show a fraction of the picture and i'd rather we be looking at application benchmarks for 'true' performance between platforms.
And i'm genuinely cautious about the performance of Apple Silicon, i believe (opinion) it'll take a few generations to 'get there' but who knows until official and final hardware is released.
 
Soldato
Joined
6 Oct 2009
Posts
3,998
Location
London
Genuinely not goading but is that the case?

Having a quick look on the Geekbench website, and purely from the first page of results (at the time of writing this) - highest single-core for the A12Z (VirtualApple)¹ is 846², where as the I7-1068NG7³ is 1385⁴; multi-core for the A12Z is 2975⁵ compared to the I7-1068NG7 at 4967⁶.

Obviously it's Geekbench and there's a multitude of reasons, including versioning, to why it isn't completely accurate and you can cut the results however which way but, the above suggests the MBP 2020 2.3GHz i7 (I7-1068NG7) is outperforming the Apple Silicon (happily 'eat my hat' if i've fudged it somewhere with the above).

Similarly the base spec Mac Mini (i3-8100B) is, marginally, "beating" the DTK Mac Mini if looking purely at Geekbench results.

Sources -
1 https://browser.geekbench.com/v5/cpu/search?q=VirtualApple
2 https://browser.geekbench.com/v5/cpu/2923376
3 https://browser.geekbench.com/v5/cpu/search?q=I7-1068NG7
4 https://browser.geekbench.com/v5/cpu/3049171
5 https://browser.geekbench.com/v5/cpu/2874828
6 https://browser.geekbench.com/v5/cpu/3055721


edit - Just to add, i think Geekbench's only show a fraction of the picture and i'd rather we be looking at application benchmarks for 'true' performance between platforms.
And i'm genuinely cautious about the performance of Apple Silicon, i believe (opinion) it'll take a few generations to 'get there' but who knows until official and final hardware is released.

You're looking at the emulated Rosetta benchmarks (those VirtualApple ones are emulated, not native). If you look at description they're x86. Don't eat your hat though!

If you look at iOS version of A12Z, it's basically 1100 singe core, 4700 multi-core. And ARM native also gives similar results. So only marginally slower than i7-1068NG7³. Caveat that this is using the iOS app, but on the Mac (ARM macs run iPhone apps). Geekbench however is supposed to be universal, but isn't perfect. There's no native ARM version of Geekbench for MacOS yet. A13 gets about 1350 single core. Apple also gets a process advantage, given that A14-based chips will be based on TSMC's 5nm process while Intel doesn't have that advantage.

Obviously we should wait for official hardware running real-world benchmarks, like how long does it take to render a frame in Blender, how long does it take to convert video formats, etc etc.

To me, based on limited information, it seems like Apple Silicon cores and Intel/AMD cores are at the same level right now. But Apple does it in a much lower power envelope. Apple is ahead on IPC, Intel/AMD are ahead on frequency. If Apple manages to significantly push ahead with A14 (rumours suggest an increase similar to A10 to A11, that was 40%), then Apple will be ahead of Intel.
 
Soldato
Joined
3 Jun 2005
Posts
3,065
Location
The South
You're looking at the emulated Rosetta benchmarks (those VirtualApple ones are emulated, not native). If you look at description they're x86. Don't eat your hat though!

Well there you go; i assumed with @squerble mentioning "natively" that these result were ran (you know) natively on the Apple Silcon (especially as ARM variants do exist) and weren't emulated.
Either way, i should have looked into it more.

Out of curiosity, if the iOS app is ran under MacOS ARM/Apple Silicon, does it still get reported as an iOS(/iPad Pro) device? Essentially, can we filter the Geekbench results purely for the DTK Mac Mini that ran the iOS app (rather than emulated x86)?
 
Soldato
Joined
6 Oct 2009
Posts
3,998
Location
London
Well there you go; i assumed with @squerble mentioning "natively" that these result were ran (you know) natively on the Apple Silcon (especially as ARM variants do exist) and weren't emulated.
Either way, i should have looked into it more.

Out of curiosity, if the iOS app is ran under MacOS ARM/Apple Silicon, does it still get reported as an iOS(/iPad Pro) device? Essentially, can we filter the Geekbench results purely for the DTK Mac Mini that ran the iOS app (rather than emulated x86)?

I don't think it's possible yet. You can look at result pages to see the RAM amount (iPads have 6GB, DTK has 16GB) but you need to do it one by one among 2600+ benchmarks :D
 
Soldato
Joined
9 Mar 2003
Posts
14,229
People have been saying for years Apple will merge iOS and Mac OS. Given that MacOS can now run iOS apps natively, on the same hardware and use similar API’s etc. are we almost at the point of the two OS’s effectively the same under the hood?

They of course have a different UI layer on top, Mac OS ‘unlocked’ and has a lot more bells and whistles which are not needed on a mobile platform.

How long will it be until that UI layer starts coming closer together.
 
Soldato
Joined
6 Oct 2009
Posts
3,998
Location
London
People have been saying for years Apple will merge iOS and Mac OS. Given that MacOS can now run iOS apps natively, on the same hardware and use similar API’s etc. are we almost at the point of the two OS’s effectively the same under the hood?

They of course have a different UI layer on top, Mac OS ‘unlocked’ and has a lot more bells and whistles which are not needed on a mobile platform.

How long will it be until that UI layer starts coming closer together.

They've always been the same under the hood. Just different UI on top, different permissions, background tasks and memory management, etc etc...

Apple has been separating their operating systems for a while now. iOS, tvOS, watchOS, iPadOS. I don't expect them to merge iOS and MacOS, but a lot of the codebase has always been shared between the two.
 
Soldato
Joined
15 Feb 2003
Posts
10,051
Location
Europe
People have been saying for years Apple will merge iOS and Mac OS. Given that MacOS can now run iOS apps natively, on the same hardware and use similar API’s etc. are we almost at the point of the two OS’s effectively the same under the hood?

They of course have a different UI layer on top, Mac OS ‘unlocked’ and has a lot more bells and whistles which are not needed on a mobile platform.

How long will it be until that UI layer starts coming closer together.

Or how long until they start crippling MacOS and removing things on the lower end products. A bit like Windows Home vs Windows Pro.
 
Soldato
Joined
6 Oct 2009
Posts
3,998
Location
London
Or how long until they start crippling MacOS and removing things on the lower end products. A bit like Windows Home vs Windows Pro.

They used to have multiple versions of MacOS (normal and server), but they dropped that. I don't think it's like Apple to release two versions of the same OS anymore. Cheapest and most expensive iPhone or iPad run the same OS. It will be the same on Macs. If they cripple it, they'll cripple it for all Macs. Although we've been hearing that they're about to dumb MacOS down into iOS for more than 10 years now, we've seen zero evidence of that happening.

A Windows 10 "S mode" like version for education to compete with Chromebooks is a possibility, but I think they're pushing iPads to that market.
 
Soldato
Joined
18 Oct 2002
Posts
19,354
Location
South Manchester
Or how long until they start crippling MacOS and removing things on the lower end products. A bit like Windows Home vs Windows Pro.

Highly unlikely. Steve Jobs at a keynote years back took the mickey out of Microsoft for segmenting Windows into editions. Something along the lines of "We have one version, and it's got all the features enabled" ... and that was in the days of paid for OS upgrades.

EDIT - Found it : https://youtu.be/STuhwRwRqD4?t=3033
 
Back
Top Bottom