Discussion Apple Silicon Team is better than ARM's own Cortex Team.

FlameTail

Member
Dec 15, 2021
53
8
36
Apple is one of the few companies that design their own cores based on ARM ISA.

Looking at the current Apple cores vs ARM Cortex cores, it is no secret that the Apple cores beat ARM in both Performance and Efficency.

Observing this situation, I conclude that Apple's Engineers are better than ARM'S own Cortex engineers. ( The sheer irony XD )

What to do you think ?
 

NTMBK

Diamond Member
Nov 14, 2011
9,691
3,524
136
They're solving different problems. ARM are designing a core that needs to be useful to a wide range of customers, potentially used on multiple different process nodes, and it needs to be small enough to be economically viable for mid to low end chips. Whereas Apple are designing for only the highest margin phones on the market, for a specific process node, and they are willing to throw huge swathes of die area at the problem. (Not just the core itself, but also the huge caches that go with it.)

It's not an apples to oranges comparison.

EDIT: Thanks @ashFTW :)
 
Last edited:

The Hardcard

Member
Oct 19, 2021
46
37
51
Apple is one of the few companies that design their own cores based on ARM ISA.

Looking at the current Apple cores vs ARM Cortex cores, it is no secret that the Apple cores beat ARM in both Performance and Efficency.

Observing this situation, I conclude that Apple's Engineers are better than ARM'S own Cortex engineers. ( The sheer irony XD )

What to do you think ?
The problem with the conclusion is that you aren’t making a full observation of the situation. The ovarall budget and how it’s being used, as well as design considerations.

Just two of potentially thousands of such considerations:

1 Part of Apple’s business model is to have far less concerns about backwards compatibility and industry standards than any other company. This is massively time consuming and frequently delays new IP. Some new IP block causes interference with needed timings for USB A connectors? ARM team needs to hold up until some group of engineers wades in and gets it worked out, while Apple’s team moves on because “hey, we don’t put USB A in anything, what do we care if it doesn’t work.”

2. Apple is partly financing TSMC’s node advancements in exchange for not only first dibs on manufacturing, but also their team gets to not only see a node‘s design rules as they are being made, also they can push for adjustments to have the node be more in line with their aspirations. ARM gets, “these are the design rules, make your desires fit them.”

If you really want irony, the former lead architect of the Cortex cores is now the head architect of Apple’s cores. Did he automagically get better?
 
Last edited:

Panino Manino

Senior member
Jan 28, 2017
604
729
136
I understand the argument in favor of Cortex Team, but still, aren't there customers that would kill to have something comparable to Apple? Why not develop a bigger core to those customers? They have multiple teams.
 

andermans

Member
Sep 11, 2020
151
153
76
Wouldn't it also be kind of natural? Like if a trillion dollar company starts designing their own cores instead of going with the off the shelf cores I'd bet it is not because they'd like up with a worse processor for their purpose. Being able to outspend ARM means Apple can attract & retain top talent and appropriately staff the design team. (Of course things can still fail or end up mediocre in the end ...)

This ability to outspend other players is something you hear about more frequently recently, e.g. Facebook now competing in the market for gamedevs with FAANG salaries.
 

JoeRambo

Golden Member
Jun 13, 2013
1,524
1,592
136
They're solving different problems. ARM are designing a core that needs to be useful to a wide range of customers, potentially used on multiple different process nodes, and it needs to be small enough to be economically viable for mid to low end chips. Whereas Apple are designing for only the highest margin phones on the market, for a specific process node, and they are willing to throw huge swathes of die area at the problem.
I don't think it is a valid "excuse" for mediocre and uninspired cores ARM was designing for a decade (and getting away with since there is no real competition). The thing that invalidates it is those "also ran" designers of custom ARM cores that were pursuing same highest margin phones market like Samsung or vendoring cores to everyone around like Qualcomm. Heck, some of those even had foundries of their own, can't really put too much credit to TSMC either.
So "custom" and "targeted" does not automatically result in success.

Just taking random iPhone review makes it obviuos how far ahead Apple was/is compared to ARM licensee crowd:

66% faster in INT? SIXTY SIX percent at similar clocks! The people who also happen to design ARMs cores can barely beat Apples efficiency core by more than 66%.

So at the end of the day Apple's team is better. We can speculate why is it so, but talent + organization enabling said talent to achieve things goes a long way, just like in every other field.
 

Justinus

Platinum Member
Oct 10, 2005
2,848
941
136
Wouldn't it also be kind of natural? Like if a trillion dollar company starts designing their own cores instead of going with the off the shelf cores I'd bet it is not because they'd like up with a worse processor for their purpose. Being able to outspend ARM means Apple can attract & retain top talent and appropriately staff the design team. (Of course things can still fail or end up mediocre in the end ...)

This ability to outspend other players is something you hear about more frequently recently, e.g. Facebook now competing in the market for gamedevs with FAANG salaries.
Not to mention that with Apple's unlimited money, they literally went around the industry making offers that the top industry talent couldn't refuse. They hand-build their silicon team with pure cash and they have zero obligations to backwards compatibility nor to developers. They can do literally whatever they want, and have unlimited money to do so. It would be a disgrace if they couldn't make something better than companies like ARM that don't have any of these luxuries.

What a weird thread topic, and pitched in a weird way.
 

Mopetar

Diamond Member
Jan 31, 2011
6,674
3,718
136
This isn't new information or any kind of surprising hot take. Custom designs are harder than you might think. Companies such as Qualcomm and Nvidia tried there hand at it in the past and gave up. It's not just a matter of being better but being better enough to justify the cost.
 

Doug S

Golden Member
Feb 8, 2020
1,110
1,655
106
It isn't just people. Apple acquired a couple companies which included not only people but their IP to start things up, and Intrinsity in particular had some technology no one else did at the time that aids timing closure and allow for higher efficiency at a given clock rate. Who know what other software they've developed over the nearly 15 years since to help with their designs and gain an advantage over others? Going wider than others, using bigger queues/buffers/etc. than others only helps if you don't drown yourself with all the extra power that requires.

Intel may have some even more impressive stuff since they've been in it a lot longer. For many years they developed nearly their entire design tool suite internally, versus everyone else using third party tools - so it is possible some of their "special sauce" may be lost as they move to industry standard tools. ARM no doubt has developed some good stuff too over the years though funding may have been lacking for a lot of their history. Compare with companies like Samsung or Qualcomm who haven't had consistent investment in developing their own cores, though the DSP like cores used in cellular baseband is probably similar enough - though they certainly didn't have much to show for their initial attempt at custom ARM cores.
 
  • Like
Reactions: Tlh97 and Ajay

dmens

Platinum Member
Mar 18, 2005
2,271
917
136
Intel may have some even more impressive stuff since they've been in it a lot longer. For many years they developed nearly their entire design tool suite internally, versus everyone else using third party tools - so it is possible some of their "special sauce" may be lost as they move to industry standard tools.
Intel and their stale old in-house tool stack is one of the reasons they are getting crushed by everyone else.
 

StinkyPinky

Diamond Member
Jul 6, 2002
6,661
609
126
I don't think this is surprising at all. Apple is far larger than ARM and attracts a lot of talent. On top of that ARM needs to design for all sorts of weird stuff whereas Apple only has to design around their own roadmap.
 
  • Like
Reactions: Tlh97 and Spikke

oak8292

Member
Sep 14, 2016
65
44
91
2. Apple is partly financing TSMC’s node advancements in exchange for not only first dibs on manufacturing, but also their team gets to not only see a node‘s design rules as they are being made, also they can push for adjustments to have the node be more in line with their aspirations. ARM gets, “these are the design rules, make your desires fit them.”

If you really want irony, the former lead architect of the Cortex cores is now the head architect of Apple’s cores. Did he automagically get better?
I have been under the impression that ARM is in the room with TSMC while the node is being developed. TSMC often uses an ARM core to assess improvements in power and performance. ARM typically has a new core designed for the new node and ARM licenses Physical IP very early in the life cycle of the node. On TSMC N5 node the Huawei Kirin 9000 with ARM A77 and A55 cores was produced prior to September 2020 and prior to or at least at the same time as Apples N5 based processor. I don't know for a fact that Huawei bought Physical IP from ARM for this processor but it must have been taped out at the same time as the Apple processor. Anandtech with data on Kirin 9000.


Daniel Nenni suggests that Apple gets a custom set of design rules and this may be true based on the sheer volume of wafers that they purchase and their contract may have different cut points on yield. With complaints from Nvidia on the 28 nm node we learned that there is a shared cost between TSMC and the customer on initial node ramp to a threshold yield. If Apple wants to take more risks with the design rules they may compensate with a different threshold yield. Lower volume customers will want more design guard bands and less risk to get to yield more quickly.

ARM is designing cores for a different purpose. Apple has a really high volume, 200 million+, of 'hero' phones and their design point is based on those volumes. ARM cores have huge volumes with many different customers. The two largest, Qualcomm and Mediatek have higher processor volumes than Apple but they have to compete on price at the low end of the market.

The ARM server cores designed under Softbank may help close some of the gap with Apple. However, I think it is still an open question as to whether the engineering costs on these cores is going to be recovered. Softbank hired a lot of engineers for ARM server cores and somewhat tanked the profitability of ARM. Engineering costs need volume and you can't overspend on engineering and stay profitable. ARM needs AWS, Alibaba and maybe some hero phone sales of the high end cores to recover the engineering.

ARM had;

"a loss of $400 million in fiscal 2020"


P.S. Apple can hide processor cost overruns in completed devices without much impact on margin. Qualcomm and Mediatek sell processors and margin compression would be pretty obvious.
 

Doug S

Golden Member
Feb 8, 2020
1,110
1,655
106
P.S. Apple can hide processor cost overruns in completed devices without much impact on margin. Qualcomm and Mediatek sell processors and margin compression would be pretty obvious.
While that's true, does for instance Qualcomm break out the profitability of their SoCs as a separate line in earnings reports, or is it in a bigger bucket that also includes all the non-SoC stuff like the modems they sell to Apple, chips for base stations, etc.? The bigger the bucket, the easier overruns are to hide.

Just like most of Apple's "margin" in A* SoCs is due to their ability to receive premium margins for iPhones, most of Qualcomm's margin in Android SoCs is their ability to receive premium margins for their SoC's cellular component. In particular in the US market (or in "world phones" for those who want to be able to operate in the US market) due to their exclusive support for 3G CDMA. That won't last much longer though, as 3G CDMA will be gone at the end of next year, which is probably one of the reasons why Mediatek is gaining traction - and Samsung will too if they ever get serious about using/licensing Exynos on a worldwide basis.
 

oak8292

Member
Sep 14, 2016
65
44
91
While that's true, does for instance Qualcomm break out the profitability of their SoCs as a separate line in earnings reports, or is it in a bigger bucket that also includes all the non-SoC stuff like the modems they sell to Apple, chips for base stations, etc.? The bigger the bucket, the easier overruns are to hide.
It is a fairly large bucket with all of the physical, QCT broken out from the licensing QTL.

For the last three years the margin in QCT for EBT has been;
2019 - 14.6% on 14.6 billion in revenue
2020 - 16.8%
2021 - 28.7% on 27 billion revenue

The revenues increased significantly in 2021 and this business is all about volume (Apple modems?).

Licensing IP has higher margins of 65-75% for the last three years.
 

Thala

Golden Member
Nov 12, 2014
1,307
609
136
66% faster in INT? SIXTY SIX percent at similar clocks! The people who also happen to design ARMs cores can barely beat Apples efficiency core by more than 66%.

So at the end of the day Apple's team is better. We can speculate why is it so, but talent + organization enabling said talent to achieve things goes a long way, just like in every other field.
God forbid, that a significantly larger core has higher IPC. So the differences in IPC is largely a design decision and has not much to do with talent. Only if you have high-margin products you can solve problems by spending area.
 
Last edited:

gdansk

Senior member
Feb 8, 2011
918
605
136
And the Neoverse team on the other hand?
Consider all the different CPU cores ARM designs: X, A7, A5, R, M, E, N, V and likely more of which I am unaware.
In ARM's business, licensing IP for every niche, what is better?
 
Last edited:

soresu

Golden Member
Dec 19, 2014
1,805
961
136
Consider all the different CPU cores ARM designs: X, A7, A5, R, M, E, N, V and likely more of which I am unaware.
Uncore aside for the most part N and V cores correspond to the big Cortex Axx/Axxx and Xn cores.

The most significant exception I've yet heard is V1 which closely matches X1, but also has support for extra instruction sets like SVE1 and LSE2.
 

FlameTail

Member
Dec 15, 2021
53
8
36
I think the problem is that if ARM makes a big core, no one would use it. Even the laughably 'big' Cortex X1 core consumes quite a lot of power and generates a ton of heat. An even bigger core would excarcebate those problems.

ARM cores have good Performance per Area but lag behind Apple when it comes to Performance per Watt
 

ASK THE COMMUNITY