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
I still have, and use, the 2017 iMac. (I also still have some I couldn’t let go of, including a 11” MacBook Air, 12” PowerBook G4, G4 Cube, PowerBook Wallstreet and PowerMac 9600.)
Base price with 16GB of RAM and 512GB of SSD will likely be no less than $1799 and no more than $1999.
I expect RAM and storage prices to match what they charged for the Intel MBPs, so $400 to go from 16GB to 32GB and $200 to go from 512GB to 1TB of storage.
O
So then I guess I can sell my 16/256 for about 10k or similar.
perfect
ok, so then around 2400 usd for 32/512 that’s kind of good deal.
So then I guess I can sell my 16/256 for about 10k or similar.
perfect
I’m going to go against my gut reaction and say A15, and bet that the next Mac chip was supposed to be released at WWDC but got pushed back. This would explain why Apple was a bit more tight-lipped on A15 at this years iphone event. They don’t want to announce all of the secrets and benefits (some of which make way more sense for Mac anyways). There’s really no reason why Apple has to start with the iPhone chips first -> Mac chips second. They very well could have started design on the Mac chip first, using the new 5nm process, and then scaled down to A15.
I really hope this is the reason, but it could also be the case that the performance increase compared to the A14 isn’t what we’re used to, and Apple knows it.
That seems correct. I never bothered much with the GPU compute part of the test suite, since my main interest in the GPUs of these devices lie in producing actual 3D graphics as opposed to running compute kernels, so tests that don’t involve rasterisation or realistic memory footprints are of limited use. For me.
We’ll simply have to wait a little more before the devices reach a wider audience to get the usual GfxBench and 3DMark test data. Maybe some enterprising graphics programmer could whip up some interesting specialised subtests? (*cough*) ?
all words are made up
- That's a made-up word. - All words are made up.
Avengers: Infinity War (2018) clip with quote - That's a made-up word. - All words are made up. Yarn is the best search for video clips by quote. Find the exact moment in a TV show, movie, or music video you want to share. Easily move forward or backward to get to the perfect clip.
getyarn.io
Last edited:
Oh I was more referring to other chip specifics like RAM, cache, etc. but from benchmarks so far A15 still looks to be an improvement from the A14 - especially in graphics performance. Also seems battery life was a huge focus too which would be pleasant news for 16” MacBook Pro users too.
Early benchmarks show A15 with about 10% better performance in single-core and 20% in multi-core over A14.
That seems worth mentioning, would be great if they’re saving that for the Mac announcement.
I can recall examples of iPads being released with the new chip before iPhones, but I cannot think of any example of an X chip ever being released before, or even within months of, the corresponding A-series. Has it ever happened?
I’m guessing the X chips do not require significant engineering efforts beyond the A/M development, and if that’s the case there should be no significant hurdle to do so, unless there are production constraints?
As the M1 can be considered an “X chip” in the context of this question, they have done it at least once by releasing the M1 2 months after the A14. But the iPad Pro didn’t get the M1 until next spring, similar to previous release cycles.
If Apple is going to shuffle timing of things around, the Mac transition is a pretty good reason to. Especially since laptops and desktops have their own ideal release windows during the year, and I don’t think Apple will throw that out the window unless they are forced to.
It seem reasonable that the MBPs are late due to many reasons. I would have expected them at this WWDC. That suggest a A14 process will be used. Would it be possible to use the updated process 5nm+ using and the A14 architecture without much work? I know too little to judge. Then the MBP chip would be an A14+. As long as it gives us performance of the current top of the line MBP or better with half the power usage - it will be fine. However, it will not do that the M2 in etc spring will "beat" the MBP SoC in performance. Tricky balance!
It would seem that if the MBP was originally slated to be announced in WWDC, then it could have used the A14 as the base.
However, the A15 for the iPhone 13 had already started mass production at TSMC before WWDC. This suggests that the A15 design was probably completed by early this year or even earlier given how long it takes to test designs, test yields, and then finally start mass production. In fact, it wouldn't surprise me if SoC designs are finished one year earlier than the mass production start date with how much testing has to be done and how SoCs cannot be delayed for an important device like the iPhone.
In addition, Apple would have known about potential delays to the MBP many months in advance. Thus, they could have planned to skip A14 for MBP all along.
Finally, there was no guaranteed that Apple planned to scale the M1 up. To scale up, Apple needs to glue multiple dies together, like chiplets(AMD) and tiles(Intel). It would have been very ambitious to do this on the first generation of Mac chips. Thus, the M1 could have simply been an iPad Pro SoC repurposed for low-end Macs. And that the second generation is truly designed for Mac scale.
Remember that Apple planned for a complete ARM transition within 2 years. We're already one year in. This means the upcoming MBP SoC probably needs to scale to very high core counts for the Mac Pro. Hence, another reason to think that the upcoming MBP SoCs are a new generation with scale designed from the start.
However, the A15 for the iPhone 13 had already started mass production at TSMC before WWDC. This suggests that the A15 design was probably completed by early this year or even earlier given how long it takes to test designs, test yields, and then finally start mass production. In fact, it wouldn't surprise me if SoC designs are finished one year earlier than the mass production start date with how much testing has to be done and how SoCs cannot be delayed for an important device like the iPhone.
In addition, Apple would have known about potential delays to the MBP many months in advance. Thus, they could have planned to skip A14 for MBP all along.
Finally, there was no guaranteed that Apple planned to scale the M1 up. To scale up, Apple needs to glue multiple dies together, like chiplets(AMD) and tiles(Intel). It would have been very ambitious to do this on the first generation of Mac chips. Thus, the M1 could have simply been an iPad Pro SoC repurposed for low-end Macs. And that the second generation is truly designed for Mac scale.
Remember that Apple planned for a complete ARM transition within 2 years. We're already one year in. This means the upcoming MBP SoC probably needs to scale to very high core counts for the Mac Pro. Hence, another reason to think that the upcoming MBP SoCs are a new generation with scale designed from the start.
Last edited:
It seems a bit strange that Mark Gurman is reporting an 8/2 big.Little SoC with 16 or 32 GPU cores.
The 8/2 big.Little suggests it's a new architecture since the M1 was 4/4. But the 16/32 GPU cores fall more in line with the M1's GPU core scale because 16 and 32 are multiples of 8. Meanwhile, the A15 SoC features a 5-core GPU so we should expect 20/40 core GPUs for MBPs.
It could mean that Marc Gurman is wrong. It could also mean that Apple does not plan to keep mobile and Mac SoCs to scale.
The 8/2 big.Little suggests it's a new architecture since the M1 was 4/4. But the 16/32 GPU cores fall more in line with the M1's GPU core scale because 16 and 32 are multiples of 8. Meanwhile, the A15 SoC features a 5-core GPU so we should expect 20/40 core GPUs for MBPs.
It could mean that Marc Gurman is wrong. It could also mean that Apple does not plan to keep mobile and Mac SoCs to scale.
It would seem reasonable to introduce the high-end first (like Rolex or BMW do) and then let the features trickle down to the more affordable products.
I guess the reason not to would be that higher-performance chips require additional development, testing or manufacturing resources and it is much more practical to introduce the lower-clocked, fewer-cored and lesser-featured chips first?
Assuming that these rumors are correct, my guess is that it would mean that these chips are fundamentally different. It would make sense for Apple to design the CPU and GPU clusters in the A-series chips so that they can easily be doubled. The internal topology of the chips and the cache hierarchy is likely already set up with M-series in mind, so that both A and M chips can be built on the same platform.
However, the prosumer chips need to be more scalable. They need to support more on-chip clusters and, if rumors to be believed, intra-chip connections. This would likely require a different topology than what the M-series use. Which means a different platform. In regards to this, Apple was aggressively hiring interconnect engineers a while back, but M1 is still based on the technology they have developed for A1xX chips, so they definitely have been working on something newer and larger.
To sum it up, we probably can’t make any conclusions about the CPU and GPU technology in the upcoming chips just based on the cluster size rumors alone. M2 will certainly use A15 cores, but the prosumer chip could use anything at this point.
Keep in mind the M1 built directly on Apple’s experience building the AX chips. So much so it’s a bit difficult to see what makes it an “M” chip instead of an “AX” chip. There’s a lot of supporting architecture higher end chips need beyond just more cores that Apple had yet to design and/or integrate/test when they started work. So as you point out, it would be a lot easier to move what they could onto the M1 first, as it was based on work they already had plenty of experience with.
More what I meant here is that including the Mac changes when SoCs need to be available throughout the year, and so shifting deadlines on when things need to be taped out/etc makes sense. We could very well emerge from the transition with a new SoC release schedule compared to before.
I just want 32GB of ram and 1tb storage now. The cpu are not an issue for me today.
They’ve done this in the past too - tag things that aren’t released at events. It’s likely just SEO and Apple trying to show up wherever they can on YouTube. This likely doesn’t mean anything. Though I still do believe MacBooks were originally planned for WWDC at one point.
Well the "M2" will beat the "M1X" only in single-core and then only by around 10% based on early benchmarks of the A15 vs. A14. The "M1X" will still handily overtake the 20% increase in multi-core of the A15/"M2" thanks to having so many more CPU cores and will of course crush it in GPU thanks to having so many more GPU cores (2-4x).
I believe the iPad Pro received the M1 because it did everything an "A14X" would have done and the extra "Mac-only" silicon can just be ignored. And making more M1s drives down the unit price.
As for the MBP delays, it is just as possible Apple already had "M1X" SoCs in production to support a summer launch. They also could have been building systemboards and cases and everything else to support a summer launch on "M1X". So the delays in the MiniLED displays meant Apple just warehoused the "M1X" MacBook Pro parts until this Fall.
Apple Silicon is meant to be scaleable both to smaller, lower-power/performance SoCs and to larger, higher-power/performance SoCs depending on application. MacBook Pros and the higher-end iMacs all came with the most powerful Intel CPUs so it is not surprising Apple is making an SoC with more CPU performance cores and GPU cores for these more powerful systems.
I believe it is.
Last edited: