How and when will Apple release the hardware features for Apple SOC Macs?

[the8thark]

We all know for iOS devices Apple doesn't add all the hardware features they have ready to go. They drip feed them to us bit by bit each year so we're forced to buy a new device each year or go without the latest and greatest hardware features. This isn't the case with Intel Macs as they've been mostly the same hardware wise for years. Better CPU/GPU and better memory are the main additions each generation. Every so often something new is added that's important but it's not every generation.

With Apple now controlling the silicon will Apple drip feed hardware features to Apple SOC Macs every generation, ie holding back features for successive generations. Or do you think we'll get a decent feature set in the first generation?

What's your opinion on this? Does it even matter to you?

 

[Pressure]

What hardware features do you feel they have been drip feeding so far?

 

[dmccloud]

If the first ARM-based Macs were to lack the same features the Intel-based Macs currently have, that would cause a great deal of trouble for Apple, both internally as well as externally from its userbase and investors. Additionally, new features take time to develop, and Apple like to wait until something is "ready" before releasing it. They were not the first company to release a smartphone, MP3 player, or tablet, yet they quickly took the lead in all three market segments because they wanted to be right rather than first.

 

[Waragainstsleep]

I think you'll find some of the extra features were added slowly as Apple became fed up with Intel not adding them but depending how good Apple Silicon for Macs currently is, it wouldn't surprise me if they held something back in case they run into a stumbling block down the line and need a way to get a boost.

 

[Erehy Dobon]

One thing for sure Apple cannot release an Apple Silicon Mac that is as good as/on par with last year's model. It needs to show substantial improvement over the previous generation.

They also have to do it in a manner that will keep price points in line with what the established price expectations are amongst their user base and they also need to preserve gross margins for shareholders.

Tim, Phil, Crag, Johny, et al. did not concoct this Apple Silicon fantasy over a pitcher of beers last summer.

Go back and watch the WWDC keynote video, particularly the first minute of Johny Sroudji's presentation. Performance-per-watt is the driver of this transition. Apple hammered this point again and again.

Apple's smaller MacBook Pros have a TDP of 25W or 28W (I forget), a thermal threshold that most Wintel notebook PC manufacturers skip. Apple Silicon is expected to get more performance at the same TDP than Intel or AMD silicon. Apple doesn't need to guess; they test this out in their labs. They get engineering samples of upcoming Intel silicon (and probably that of AMD as well).

My gut feeling is that Apple has had Arm-based prototypes running macOS in their labs for ten years, since Apple's first custom SoC for the iPhone, the A4.

That means for 9+ years they have walked out of their labs saying "That's great progress but it's not ready to be released."

There are undoubtedly plenty of other technologies being explored in their labs that aren't ready for release. Apple will release them when they feel it's ready. Look at AirPower. They actually pre-announced it then pulled it, something they rarely do.

Apple did not get to being a $1 trillion market cap company by throwing stuff at the wall and seeing what sticks.

 

[MikhailT]

One thing for sure Apple cannot release an Apple Silicon Mac that is as good as/on par with last year's model. It needs to show substantial improvement over the previous generation.

I don't think they need to at all but it depends on which Macs. For an example, if the first Apple Silicon Mac is going to be a MacBook Air, than they only need to show battery life improvement, not major performance improvement. Even getting 30-60m improvement will make folks happier even if the performance is the same. Heck, a quieter/cooler machine with same perf is still a better buy.

I don't think many users will care or know the difference as long as it is as good as the latest Macs with their current software if they're coming from old Macs that's already slower. If 100% of my apps run fine on it at the same performance under Rosetta and my current Macs is from 2015 or whatever and the new Macs is still faster, it'll still be a buy for me. Majority of people do not upgrade every year (lifespan is usually 3+ years). So, they only need to not regress from the current Macs with slight improvements toward all areas of the Macs, specifically the battery and heat.

It's all about the balance between battery, compute perf, heat, etc. If Apple strike it right as they continue to do so with iPad Pros, then they don't have to show any substantial improvements for the first year. Notice there was no outrage at Apple for not improving pref enough in `20 iPad Pros vs. `18 iPad Pros, it was the magic keyboard + trackpad that stole the show.

That's why I think the first generation m14 (A14 for Macs) is going to be minimal (5nm should still bring 10-15% IPC improvement by default compared to 7nm A12) compared to current A12Z in Mac minis/iPad Pros and lower TDP than expected (not 20w, 10W at most) but I am happy to be wrong here. I think the second generation Mac SoC is going to be more impactful as they shift more and more Macs to their Apple Silicon.

 

[Erehy Dobon]

Without a doubt, Apple has tested multiple combinations of CPU + GPU + other silicon in various lab machines.

Since we know nothing about the actual devices to be released, let's think about how they would have handled this in the lab. They would likely be doing some tests using existing case designs, meaning the thermal limits are constrained at known thresholds.

Let's say Notebook Enclosure A has a TDP ceiling of 25W. Apple would combine silicon to reach that level and vary the amount of CPU versus GPU capability. Or they could put in 22W and run a cooler machine with more battery life.

If Apple Silicon SoC uses 25W, it's drawing the same amount of current as the Intel part running at 25W. Again, this returns to the performance-per-watt focus. Apple would then have a much higher performance unit.

I have a suspicion that the first generation of Apple Silicon Macs will be using the same enclosures as the previous Intel model. No one here knows what Apple will release this fall.

What will be more interesting is when Apple starts releasing Apple Silicon Macs in enclosures specifically designed for Apple Silicon.

Until now, Apple's custom SoCs have largely been deployed in mobile devices with heavy power constraints. It doesn't make sense to extrapolate the Apple Silicon Mac product line roadmap based on mobile device specifications.

Someday, Apple will likely put Apple Silicon in a high-end MacBook Pro and it'll draw far more power than even the top iPad Pro will. With the extra real estate in the computers, Apple has far more options in terms of design choices especially when discrete custom silicon like the T2 Security Chip is integrated in the SoC.

My guess is that we will not see the full breadth of this transition until 2023-24 when both the silicon and the enclosures will have undergone several iterations of design in tandem.

 

[jinnyman]

I'm curious about their hardware roadmap. Apple should atleast inform us what lays ahead in chips for confidence builder. For example, I'd like to know how far they plan to go. Will Apple try to go beyond their competitors? Or will they settle for consumer level chips? I'm more worried about Apple's commitment in workstations.

 

[Erehy Dobon]

Don't hold your breath.

Apple does not provide product roadmaps.

Wishing so will not make it so.

 

[johannnn]

Apple should atleast inform us what lays ahead in chips for confidence builder.

Are you new to the Apple world?

 

[jinnyman]

So we wait for another 6 years or more waiting for Arm Mac Pro that may never come out?
unless Apple wants to give up on that, they will say something like "plan to introduce MP level machine".

This has happened before with MP 7,1. No I'm not new to the Apple world.

 

[Erehy Dobon]

Did you watch the WWDC keynote?

Considering they did most of the Apple Silicon demos on an Apple Silicon-powered Mac Pro, I would say there's a decent chance that they will release an Apple Silicon Mac Pro in the future. They highlighted a couple of pro-level applications -- Adobe Photoshop and Autodesk Maya -- as well which would lead one to believe that there was a minimal level of commitment to the high end user.

Maybe not this fall but Apple quoted a two-year transition period. My guess is that Mac Pro will be the last Mac product line to move to Apple Silicon.

 

[jinnyman]

Did you watch the WWDC keynote?

Considering they did most of the Apple Silicon demos on an Apple Silicon-powered Mac Pro, I would say there's a decent chance that they will release an Apple Silicon Mac Pro in the future.

Maybe not this fall but Apple quoted a two-year transition period. My guess is that Mac Pro will be the last Mac product line to move to Apple Silicon.

I watched WWDC Keynote. I remember they only ran A12Z equipped mac mini like one of those DTK unit. Correct me if I'm wrong, but I didn't see Mac Pro with Arm.

They also didn't mention about Mac Pro, and I just hope their goal of "complete transition in 2 years time" will include Mac Pro like workstations.

 

[Erehy Dobon]

I thought I remember seeing a Mac Pro. Maybe I should go and watch the WWDC keynote video again myself!

However, my belief is that Apple has a variety of prototype Apple Silicon systems ranging all the way from something like the discontinued Retina MacBook all the way to the Mac Pro running in their labs. Without a doubt, they are running a variety of configurations and most will never see the light of day.

Again Apple does not share product roadmaps so people need to have some faith. If you are a repeat Apple customer you already do this to some extent.

People are also free to walk away from Apple. There are some fine AMD CPUs and Nvidia GPUs right now, one can easily build a Windows PC graphics workstation comparable to the best appointed Mac Pro at a very competitive performance-per-dollar value.

There is nothing -- apart from dollars or fear of excessive system administration load -- that prevents anyone from enjoying using both Windows and Mac hardware in their lives. For me that meant Windows at work and Mac at home for over twenty years. Today the lines are blurred since I work for myself and I have both Windows and Mac computers at home.

 

[Waragainstsleep]

So we wait for another 6 years or more waiting for Arm Mac Pro that may never come out?

Whole line will transition within two years. This includes the Mac Pro.
The WWDC video featuring the hardware lab had stacks of Mac Pros around the room. Who's to say they were Intel Mac Pros....

 

[MyopicPaideia]

How: At an event specifically for new Macs
When: In October, after the iPhone 12 release event

 

[Kostask]

I'm curious about their hardware roadmap. Apple should atleast inform us what lays ahead in chips for confidence builder. For example, I'd like to know how far they plan to go. Will Apple try to go beyond their competitors? Or will they settle for consumer level chips? I'm more worried about Apple's commitment in workstations.

Intel used to put out road maps, they even had a name fpr them. the "Tick-Tock" cycle. Since 2016, there has been no "Tick", and "Tock" is completely dead, since ther fab process really hasn't improved much. This is what pushed Apple to move away from Intel.

the "roadmaps" are substantially useless unless you are selling merchant/commercially available chips. They give the potential customers/builders using your chips an idea of where you are going. However, Apple will NOT be selling chips to anybody but themselves. And internally to Apple, the chip designers are already talking to the OS developers and system design engineers, and have been for many years; I would guess for as long as they have had MacOS running on Apple Silicon chips. The system design engineers have also been part of it, they just haven't bothered to design anything around the Apple Silicon chips until the OS developers said that the performance was acceptable.

In any case, issuing a road map would be the same as revealing future models of the Apple products that use those chips. That is never going to happen.

 

[the8thark]

One thing for sure Apple cannot release an Apple Silicon Mac that is as good as/on par with last year's model. It needs to show substantial improvement over the previous generation.

Add in the rest of your comments here also. I agree 100%. Well said.

My point I was making was, lets say this new Apple SoC Mac is the true generational leap as it really needs to be. Will Apple hold back a hardware feature or two just to have something new to add to the second generation of Apple SoC Macs? Or is this a case of, this has to be the way, as the features are literally added into Macs as soon as they are ready for the prime time?

In any case, issuing a road map would be the same as revealing future models of the Apple products that use those chips. That is never going to happen.

Agreed. The "complete transition to Apple SoC within two years" is as close as we'll ever get to any kind of longer term roadmap from them.

Also as an aside, since you can't buy Apple SoC on it's own, it's the overall Mac experience that matters more then the SoC in isolation. I feel for the Apple SoC Macs, the benchmark nubmers will not be as important as for other brands of PC.

 

[Waragainstsleep]

I feel for the Apple SoC Macs, the benchmark nubmers will not be as important as for other brands of PC.

If they score really highly, the benchmarks will matter. If they don't, Apple will find ways to shift the narrative to real world performance, battery life, design and other benefits.

 

[MikhailT]

My point I was making was, lets say this new Apple SoC Mac is the true generational leap as it really needs to be. Will Apple hold back a hardware feature or two just to have something new to add to the second generation of Apple SoC Macs? Or is this a case of, this has to be the way, as the features are literally added into Macs as soon as they are ready for the prime time?

We won't know for a while but I seriously doubt it. The main reason is the hardware cost of fab'ing a custom SoC, they're in hundreds of millions of dollars on the most advanced process node and these designs are worked on for years. Iterations to the same design would be cheaper than to do major changes every single time.

The upcoming A14 chip? That was done 2-3 years, it takes a long time to validate a very expensive design and not to mention, they have to wait for TSMC to finish the 5nm fab on time.

In other words, it may be easier for Apple to hold back stuff like cameras and external designs to have something for another year, it is a small cost and can be changed at the fab/supplier/packager but for a custom SoC? Too prohibitive of a cost and risky.

I do think Apple won't reveal all of the hardware in software at the time, meaning they may enable hardware feature via firmware for future updates, like what they did with wireless and NFC in the past. Much easier to disable hardware features in firmware than remove them from the design to re-add in the future (where a single mistake can be very expensive).

 

[leman]

My point I was making was, lets say this new Apple SoC Mac is the true generational leap as it really needs to be. Will Apple hold back a hardware feature or two just to have something new to add to the second generation of Apple SoC Macs? Or is this a case of, this has to be the way, as the features are literally added into Macs as soon as they are ready for the prime time?

You still haven’t made clear what kind of features you are talking about. Just performance or something more substantial?

My guess is that if Apple Silicon indeed performs that much better - Apple is unlikely to sell full speed configurations as first gen products. They only need to outperform everyone else, there us little gain in getting too far ahead.

 

[JMacHack]

Hardware features? Dunno, but I don't see things like Bootcamp going away forever. I'd imagine within the next five(?) years we'll have WinARM Bootcamp.

In terms of hardware features, I think upon release we'll at least have feature parity with the Intel offerings.
Maybe with Apple Silicon GPUs we'll see the return of Target Display Mode, or at the very least Sidecar for MacOS.

My guess is that if Apple Silicon indeed performs that much better - Apple is unlikely to sell full speed configurations as first gen products. They only need to outperform everyone else, there us little gain in getting too far ahead.

Keeping my expectations realistic, I'd expect a 20%-50% gain over current Intel models, at least for 1st gen products like the Air and 13-inch MBP. That'd offset Rosetta's performance hit and allow Apple to flex on Intel while providing better battery life. Apple's probably not gonna go whole-hog with the first models anyway.

I'd expect their "Pro" models to start kicking around Intel pretty badly though.

 

[JohnnyGo]

Apple has a couple of advantages:

1) it has an asymmetrical design with high performance coupled with low power cores, that gives then flexibility to build different SOCs for different Macs (lighter notebooks will use more low power cores to extend battery life while pro laptops may lean towards more high power cores whilst iMacs can run with little or no low power cores)

2) Apple has lots of experience with multi core SOCs as iPad SOCs have historically 2x cores vs its iPhone counterparts, that gives Apple the ability to add 8-12-16 cores to its SOC and differentiate its offering within the same design generation

3) TSMC investments in 5nm allows for higher density / lower power consumption and decreases manufacturing costs (more SOCs per wafer) which in turns allow for higher IPCs and binning possibilities (a 16 core design can achieve higher yields if you can use a 12 core model given 4 defective cores, or just sell an 8 core design by turning off the other cores

4) Apple has build image and video special units, storage controllers, security, machine learning, wireless communication, etc around its CPU and GPU cores, this offloads part of the instructions out of the CPU but still in the faster bandwidth within a SOC (no need for PCIe lanes) at the same time conserving energy

5) All its Macs have internal fans which are not present due to size constraints in iPhones and iPads, that allows Apple to run more power through its SOCs than the current sub 10 watts to 15-20 watts thus allowing for even greater performance

The first Apple Silicon will simply blow past the A12z and whatever equivalent intel processor in the previous version of the same Mac.

Some people will have to see it to believe it. I’m in the camp of reading the tea leafs (as laid out above) and eagerly awaiting the future to materialize!

 

[Hexley]

We all know for iOS devices Apple doesn't add all the hardware features they have ready to go. They drip feed them to us bit by bit each year so we're forced to buy a new device each year or go without the latest and greatest hardware features. This isn't the case with Intel Macs as they've been mostly the same hardware wise for years. Better CPU/GPU and better memory are the main additions each generation. Every so often something new is added that's important but it's not every generation.

With Apple now controlling the silicon will Apple drip feed hardware features to Apple SOC Macs every generation, ie holding back features for successive generations. Or do you think we'll get a decent feature set in the first generation?

What's your opinion on this? Does it even matter to you?

That's a bit cynical don't you think?

Typical upgrade/replacement cycle of a smartphone has lengthened from 2 years to 3 years while computers from 3 years to 5-6 years.

 

[thenewperson]

That's a bit cynical don't you think?

Typical upgrade/replacement cycle of a smartphone has lengthened from 2 years to 3 years while computers from 3 years to 5-6 years.

Necro-ing old threads is annoying 😕