Now that the iPhone 13 with an A15 chip has been announced, do you think the Macbook Pro slated for later this year will be based on the one year old A14 or the current A15?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Will the upcoming Macbook Pros use A14 or A15 cores?
- Thread starter senttoschool
- Start date
- Sort by reaction score
Since mini LED likely delayed the introduction of MacBook Pro along with the rumors of it being called M1X, it's likely the cores are A14 based.
Gurman seems to suggest M1X is the actual marketing name.
In general, you also don't take a leading edge design (A15) and expand on it with more cores and controllers, due to the risk involved.
I think it’s going to be based on the same technology as A15, but also with more hardware differences. Basically, I expect it to be a different chip, better tailored for (mobile) desktop use.
There is no hard rule that states that. There are many examples in the industry where new designs came to high-end products first. The truth is that we simply don’t know how Apple will handle these things because they have exclusively designed for mobile, and M1 is not much more but a beefed up mobile SoC.
There is no hard rule that states that. There are many examples in the industry where new designs came to high-end products first. The truth is that we simply don’t know how Apple will handle these things because they have exclusively designed for mobile, and M1 is not much more but a beefed up mobile SoC.
I believe it will be based on the A14, regardless of it's name (though I fully believe it will be called "M1X").
Hard to say.
It is interesting to note that the A15 is available in 4-core (iPhone 13) and 5-core (iPhone 13 Pro + iPad mini 6) GPU configurations.
This means that the A15 is a 5-core GPU design (the 4-core variant is the same but one of the cores is disabled to increase yield).
We also know from the rumors that the M2 SOC going in the 2022 MacBook Air is going to have a 10-core GPU...
Do you see a pattern?
The M1 was an A14 with double the count of high performance cores and double the count of GPU cores.
The M2 will be an A15 with double the count of high performance cores (still 4) and double the count of GPU cores (10).
So where does the M1X fit in?
The rumors say the M1X will come in 16-core and 32-core GPU configurations. If I had to see a pattern here, I'd say that 16 and 32 are multiples of 4, so.... I'm thinking the M1X will be an evolution of the A14 design rather than the A15 design.
There's nothing preventing Apple from building some kind of hybrid, but if you want to see a pattern here, it's definitely this one.
It is interesting to note that the A15 is available in 4-core (iPhone 13) and 5-core (iPhone 13 Pro + iPad mini 6) GPU configurations.
This means that the A15 is a 5-core GPU design (the 4-core variant is the same but one of the cores is disabled to increase yield).
We also know from the rumors that the M2 SOC going in the 2022 MacBook Air is going to have a 10-core GPU...
Do you see a pattern?
The M1 was an A14 with double the count of high performance cores and double the count of GPU cores.
The M2 will be an A15 with double the count of high performance cores (still 4) and double the count of GPU cores (10).
So where does the M1X fit in?
The rumors say the M1X will come in 16-core and 32-core GPU configurations. If I had to see a pattern here, I'd say that 16 and 32 are multiples of 4, so.... I'm thinking the M1X will be an evolution of the A14 design rather than the A15 design.
There's nothing preventing Apple from building some kind of hybrid, but if you want to see a pattern here, it's definitely this one.
Yup.
Not that it necessarily means much in terms of CPU performance. The numbers given by Apple for the iPad mini suggested that CPU and GPU performance was roughly equal to the A14. The overall device had improved power efficiency however, possibly partly due to the SoC.
They did mention that the last level cache had doubled in size, but then that was always likely to change with the bigger laptop chip. Unfortunately there was no hint in the presentation that the SoC used LPDDR5.
Yes, I was thinking the same thing. This reinforces the idea of M-series being an A-series doubled, a straightforward adaptation of a mobile platform to suit desktop needs. This is the same technology we’ve seen with older iPad chips. M1 even uses the same packaging technology as A12X etc. - that chip could have also been called A14X with nobody complaining.
That’s why I have been thinking for a while now that the pro-level Apple Silicon will be something different. To serve the MBP and the Mac Pro range simultaneously Apple will need scalable technology beyond what the A/M series offer. Rumours (and patents) point towards a chiplet-based design where multiple chips can be combined to build more powerful systems on a package. If Apple is indeed designing such a platform, there is little point in keeping it tied to A/M series. After all, this technology will not be designed to ship in low-power device, and it has to offer completely different level of scalability.
So my idea is still the same as one I was describing months ago: one low-power microarchitecture (with two concrete chip implementations in form of A and M series), and one professional microarchitecture, in form of chiplets. Both architectures will obviously be related by a common technological basis, but they will also feature significant microarchitectural differences.
It is pretty easy for the memory controller to support both LPDDR4 and LPDDR5. Tiger Lake does support both for example. On the other hand, the next Mac ASi SoC is going to be a different design than the A15 so maybe they'll just substitute a LPDDR5 controller for the LPDDR4 in the A15.
Either option could well be. We'll see. Given the graphics performance seems largely unchanged from the A14, I think it's a pretty safe bet that the announced devices do not use LPDDR5, but we'll probably get a teardown from iFixit within a couple of weeks that settles the matter, if we don't get memory tests from Anandtech first.
This question should be posed as "Firestorm/Icestorm"-based vs "Avalanche/Blizzard"-based.
I have a hard time thinking that Apple will release new pro chips based on Firestorm/Icestorm. Maybe they made a few for the first year as an experiment, but now they've got another year's of experience with TSMC.
I have a hard time thinking that Apple will release new pro chips based on Firestorm/Icestorm. Maybe they made a few for the first year as an experiment, but now they've got another year's of experience with TSMC.
And if "M1X" is on the Enhanced 5nm process, the Firestorm / Icestorm should be able to clock a bit higher or run a bit cooler at the same clock and use a bit less power than the M1 so there can still be advantages.
Same same. From the presentation I have the impression that the A15 CPU cores are more a fine tuning off a14 rather than a complete new design. So it may be A15 but not because there is much performance improvements same frequencies, but they might contain other improvements needed for SOCs with more cores. Maybe a bit more power efficiency and/or higher top frequencies.
Anandtech predicts that the A15 GPU performance is up about 28% over the A14.
https://www.anandtech.com/show/16934/apple-announces-iphone-13-series
Yes but that is from the extra graphics core, not architectural enhancements (unless you count the doubled LLC as such). And it doesn’t apply to all A15 chips, unfortunately, some only have four GPU ”cores” active.
It is a prediction though. Apple were careful not to make performance comparisons to the predecessor, and I think it’s a good idea to not overinterpret their rather nebulous claims. We’ll have better data soon enough. Also, it is not a given that what goes for the A15 will also be valid for the new Mac SoC.
Absolutely, and I think you have it figured out. We will be seeing an M1X as described above, although I'd be thrilled if we get an M2X MBP next month.
(Moving forward it would seem reasonable to introduce the MxX MBP before the M(x+1) MBA, to avoid a situation where the MBA has better single-core performance and/or other technical features. The M2X could have been finished for months, so I guess there is a glimmer of hope.)
Maybe. On the other hand it is more than 25% predicted by the 5th core and presumably at the same or less power as the A14 GPU. Seems like that could architectural though some of that is because of the enhanced TSMC N5P process.
By and large, and judging from very little data indeed, the main enhancements to the A15 seems to be boosted neural engine performance, added GPU core and doubled LLC cache (and new hardware codec support). Longer battery life may or may not have much to do with the SoC itself, and even then may be more about better management rather than lower operational power draw per se.
To me, this seems rather sensible when you use basically the same lithography and operate within the same thermal and power limits. They added a bit more than 25% transistors but they could not let them increase the power draw, rather it would be preferred if the power draw went down. And overall, it did.
What this means for the Mac chips is opaque, and the OPs speculation kind of makes sense. But it’s almost a year since the M1, and a simple scaling of the number of M1 functional units seems a little simplistic/unambitious. I will hold on to some hope for more. But I don’t think that we can infer all that much about that from the A15 even when we have real test data. Patience, I’ve heard, is a virtue. 😀
Not quite the same lithography. Apple is using the next TSMC tweaked 5nm process, N5P. TSMC has claimed that N5P gives either a 7% performance improvement or a 15% reduction in power over N5 using the same design rules.
The assumption (and it is a safe assumption, IMO) is that "M2" will be the A15 with more performance cores and more GPU cores plus the Mac-specific features an M-series SoC has.
And considering another safe assumption (IMO) is that "M2" will replace "M1" on the current M1 Macs (so 24" iMac, MacBook Air, 13.3" MacBook Pro [if it continues] and Mac mini) it will still provide a noticeable and meaningful performance increase over the M1 models.
It's a tried and true process not just for Apple (AX and AZ series), but the industry as a whole. So I see no reason for Apple to not follow the same with "M1X" vis-a-vis M1.
Last edited:
Earlier M2 leak mentioned 10 GPU cores which makes perfect sense given that A15 has 5 GPU cores.
Depends on the product though. Take AMD's Zen3 for example. The desktop Zen3 chip (Vermeer) contains 8 CPU cores and a large 32MB cache, and is designed to be scalable (multiple Vermer chips can be connected together to make larger CPUs). The mobile Zen3 chip (Cezanne) is an SoC that includes 8 CPU cores, a GPU and a smaller 16MB cache. They also have different memory controllers, Cezanne has lower peak CPU clock. Same CPU core, rather different products with different design goals.
M1 was a rather straightforward upscaling of A14. Sure, it contained some desktop-specific functionality, but the basic building blocks and the design principles are the same. I am not convinced that the same platform will be sufficient for Apple's prosumer chips however. If Apple wants scalable chip technology, simply "doubling" the M1 is not sufficient.