Question Apple Silicon M series thread

[guidryp]

There is only one ARM MBP, the low end model at the smallest size. You don't think they want ARM Macs across the whole laptop line for back to school?

I don't think it factors at all. Students will primarily going for the small (I doubt any group actually carries around a laptop more) and inexpensive. So the 13" MPB and MBA would be the most popular MB's with students by far, and of those the less expensive MBA would be the most popular.

The real target audience for higher end models will be working professionals. It would have been nice to show them to developers at WWDC, but really there is no critical date for such machines. They will be ready, when they are ready.

 

[Eug]

Well, that's interesting. Not sure if this means anything but someone has noted the metadata keywords for the WWDC 2021 YouTube upload include "m1x" and "m1x MacBook Pro", and not "M2". I verified this myself and took a screen grab:

 

[DrMrLordX]

Hmm. M1x == A14x or . . . ?

 

[gdansk]

Well, that's interesting. Not sure if this means anything but someone has noted the metadata keywords for the WWDC 2021 YouTube upload include "m1x" and "m1x MacBook Pro", and not "M2". I verified this myself and took a screen grab:

View attachment 45530

That is set by YouTube's analytics of related keywords. On videos I uploaded it suggests all sorts of keywords I did not set.

 

[Eug]

BTW, I don't know if it has been posted yet, but a while back CPU Monkey posted its fake M1X benchmarks and specs, with probably the wrong number of cores, and which are the exact same specs as its fake M2 specs.

CPU benchmark

Our processor / CPU comparison helps you to compare two CPUs. We use benchmark results from Cinebench R20, Cinebench R23 and Geekbench 5 as well as the FP32 raw performance (GFLOPS) of the iGPU.

www.cpu-monkey.com

Apple M2 - Benchmark, Test and specs

Apple M2 benchmark results and review of this cpu with specs including the number of cores, threads, memory bandwidth, pcie lanes and power consumption. Benchmarks in Cinebench R23 and Geekbench 5

www.cpu-monkey.com

 

[nxre]

I've been thinking about it and I highly doubt we will see a M2/M2X based on the firestorm core sucessor.
Apple seems to have a very strict cycle of refreshes with a node shrinking with refined architecture followed by a new architecture. AX designs seem to be on a two year cycle that follows the refined architecture.
A10 - New process / Refined architecture > A10X
A11 - New architecture
A12 - New process / Refined Architecture > A12X
A13 - New architecture
A14 - New process / Refined architecture > A14X/M1
which leads me to think
A15 - New architecture (ARM V9?)
A16 - New process / Refined architecture > A16X/M3

This could mean we will only see improved single core performance in late 2022/early 2023, which could put them at a disadvantage against Intel/AMD with the Zen 3+ refresh and Alder Lake coming late 2021/early 2022. But it would also mean the single core performance improvement would be akin to a two gen improvement, which judging by apple's impecable execution could put it quite ahead of the pack when it does launch.

 

[guidryp]

I've been thinking about it and I highly doubt we will see a M2/M2X based on the firestorm core sucessor.
Apple seems to have a very strict cycle of refreshes with a node shrinking with refined architecture followed by a new architecture. AX designs seem to be on a two year cycle that follows the refined architecture.

We have already seen AX (iPad Chip ) replaced by M1.

So I wouldn't get too wrapped up in marketing names. They are just a convenience, easily replaced with another convenience. Same for any patterns that you see. Patterns emerge and disappear as circumstances dictate.

The main thing that seems likely is that there will be 8 Performance core SoC later this year for high end Laptops and mid range desktops, and it should be pretty damn great in that role even if it doesn't have a new core.

It really doesn't matter if Apple has a slight core disadvantage. Macs have never sold on having the absolute highest core performance.

 

[Doug S]

I've been thinking about it and I highly doubt we will see a M2/M2X based on the firestorm core sucessor.
Apple seems to have a very strict cycle of refreshes with a node shrinking with refined architecture followed by a new architecture. AX designs seem to be on a two year cycle that follows the refined architecture.
A10 - New process / Refined architecture > A10X
A11 - New architecture
A12 - New process / Refined Architecture > A12X
A13 - New architecture
A14 - New process / Refined architecture > A14X/M1
which leads me to think
A15 - New architecture (ARM V9?)
A16 - New process / Refined architecture > A16X/M3

This could mean we will only see improved single core performance in late 2022/early 2023, which could put them at a disadvantage against Intel/AMD with the Zen 3+ refresh and Alder Lake coming late 2021/early 2022. But it would also mean the single core performance improvement would be akin to a two gen improvement, which judging by apple's impecable execution could put it quite ahead of the pack when it does launch.

What are you basing your claims about which ones are "refined" architecture and which ones are new architecture on? It looks to me like you're trying to fit Apple's SoCs into some sort of Intel tick tock strategy. Where's the evidence for this?

Just based on timing alone I think it is a slam dunk that whatever they call the upcoming chip with more cores that will go into higher end versions of existing products, as well as products that haven't gone ARM yet like the 16" MBP, it will contain "Firestorm successor" cores from the A15. Why would they wait until mass production of A15 SoCs has begun to release something containing A14 cores? Don't you think having a newer core (whether or not it qualifies as a "new" or "refined" architecture to you) makes sense in what will become the "high end" offering for Macs that currently use the M1?

If you have a choice between a Mac Mini with an M1, or paying more for the high end version, wouldn't you expect to get more than improved multicore performance but also get at least a little bit of single thread improvement? I mean, I suppose it could simply be clocked higher due to N5P's process refinement, but I don't expect them to pass up the opportunity to give it IPC gains from a new core also given that the timing makes that an easy decision to make.

Anyway Apple has no reason to care how they stack up against any of AMD's products, that's something for benchmark obsessed Anandtech readers to worry about. Just like people aren't switching from Android to iPhone because Apple's phones have faster CPUs, no one is going to switch from Windows/x86 to Mac/ARM even if Macs beat both Intel and AMD CPUs. Apple's primary goal is to insure every ARM Mac is faster than the x86 Mac. I'm sure their secondary goal is to beat every Intel alternative, just to stop any potential claims of "they are giving up performance by switching they could have used this Xeon xxx or i9 yyy". Since there is no credible claim they ever considered using AMD CPUs in a Mac, I don't think they give a damn how those compare. That's Intel's problem to worry about, not Apple's.

 

[nxre]

What are you basing your claims about which ones are "refined" architecture and which ones are new architecture on? It looks to me like you're trying to fit Apple's SoCs into some sort of Intel tick tock strategy. Where's the evidence for this?

I quote anandtech on this - Over the years, Apple’s design cadence seems to have settled down around major bi-generation microarchitecture updates starting with the A7 chipset, with the A9, A11, A13 all showcasing major increases of their design complexity and microarchitectural width and depth(source).

It makes sense for Apple to use the new core, but Apple is also no stranger to situations like this. Like releasing an iPad Pro in 2020 with a chip that had lower single core performance than a 2019 iPhone. It's weird, but not impossible and I wouldnt be surprised if iPhone 13 ended up having a newer high performance core than an upcoming Macbook Pro 16. They could problably offset the difference by increasing clockspeeds and improving the memory subsystem of Firestorm core however.
The rumours that M1X macbooks were meant to be showed in WWDC but got delayed due to mini-led supply issues also lead me to believe in this theory, given Apple has never introduced a new architecture on a non iPhone device(until, of course, 2020 with the iPad Air, but this was mainly an iPhone supply issue that caused them to delay the iPhone launch rather than a conscious choice to debut a new uarch in an iPad).

 

[Gideon]

Hmm, it does look M1 is doing some fancy coprocessor speedup for display drivers, interesting:

https://twitter.com/i/web/status/1415360411260493826

 

[Eug]

Apple event scheduled for Monday, October, 18.

M1X is coming folks!

 

[guidryp]

I would think at this point they would call it M2, but regardless of naming it will be very interesting to see what they have. Hope it ends up in the Mac Mini as well. I could see that as my first Apple product.

 

[jpiniero]

I would think at this point they would call it M2, but regardless of naming it will be very interesting to see what they have. Hope it ends up in the Mac Mini as well. I could see that as my first Apple product.

Decent chance the Mini gets bumped but don't be shocked if the MBP 16" processor is an $$$ upgrade on the higher end Mini.

 

[guidryp]

Decent chance the Mini gets bumped but don't be shocked if the MBP 16" processor is an $$$ upgrade on the higher end Mini.

I definitely expect it to be a very expensive upgrade. I am hoping they have a 6 performance core segment(down from 8 core expected) version that is more moderate price increase.

Probably a decent chance they will hold off on putting them in a Mini for a while, as they prioritize the more important Macbooks.

 

[Eug]

I would think at this point they would call it M2, but regardless of naming it will be very interesting to see what they have. Hope it ends up in the Mac Mini as well. I could see that as my first Apple product.

It will most likely be M1X.

My guess is M2 will come next year and will be SLOWER than M1X for multi-core CPU and for graphics. M2 would be the successor to M1. M2X would be the successor to M1X.

It's like the old days of the iPad and iPad Pro, where the A. series chips were for the iPad and the A.X chips were for the iPad Pro. eg. A12X is faster than not only A12, but also faster than A13. A13 has faster single-core, but A12X has more cores so it is faster than A13 for multi-core. In fact, for multi-core, A12X is even faster than A14. Plus, A12X has a much faster GPU than the A13 GPU.

I definitely expect it to be a very expensive upgrade. I am hoping they have a 6 performance core segment(down from 8 core expected) version that is more moderate price increase.

That is extremely unlikely. I'm guessing the 14" MacBook Pro will start around US$1799 or so, but that's with 16 GB RAM and 512 GB storage. Or at most US$1899. However, that is just a guess.

Probably a decent chance they will hold off on putting them in a Mini for a while, as they prioritize the more important Macbooks.

Agreed. That will probably be the case. I'm more interested in an M1X Mac mini or a fall 2022 M2 Mac mini, but not for performance reasons. It's because I generally don't like buying gen 1 stuff. Indeed, the M1 Mac mini has had some weird monitor and Bluetooth compatibility bugs. Plus the M1 Mac mini has the old case. It's rumoured to get a new case design with M1X, which would surely be adopted with the M2. The M1X one will likely also have more ports than either the M1 or M2 models.

 

[Roland00Address]

Would it be easier marketing wise? ... since Apple wants people to know the name of the chips (but less than 5% of the buying public, only the people who care about speeds and feeds and comparison between Apple and non Apple)

To use M1 "Big" instead of M1X?

( I am just throwing stuff at the wall, not making a case of true advocacy of such a name. )

 

[Ajay]

Geez, took a look at current M1 Mac Mini's. Pretty expensive for 8GB RAM and 256GB NVME. Oh, and only upgradeable by Apple. Literally requires higher capacity chips to be soldered on to the the tiny motherboard . Have been thinking about learning more about Linux and maybe trying MacOS/IOS programming - would combine both into one package. It's just that $1099 for 16GB RAM and 512GB NAND for a bit of future proofing is, again, a hard nut to swallow being a PC enthusiast for ~25 years. [I've used UNIX and MacOS professionally, but that was > 15 years ago]

I think my current system cost me a total $1500 (sans monitor and 4TB HDD).

 

[Eug]

Would it be easier marketing wise?

...snip...

To use M1 "Big" instead of M1X?

( I am just throwing stuff at the wall, not making a case of true advocacy of such a name. )

Apple users were already very comfortable with the ..X nomenclature for the iPad Pros. It's not a big leap to start using it for their Macs now. But we shall see next week.

Have been thinking about learning more about Linux and maybe trying MacOS/IOS programming - would combine both into one package. It's just that $1099 for 16GB RAM and 512GB NAND for a bit of future proofing is, again, a hard nut to swallow being a PC enthusiast for ~25 years.

Linux on M1? Is that even viable?

 

[moinmoin]

Linux on M1? Is that even viable?

Seems to be getting there:

Blog - Asahi Linux

asahilinux.org

The big outstanding part is GPU acceleration which is obviously the hardest part, being a complete black box in multiple regards.

But some neat discoveries as well:

Apple is unique in putting emphasis in keeping hardware interfaces compatible across SoC generations – the UART hardware in the M1 dates back to the original iPhone! This means we are in a unique position to be able to try writing drivers that will not only work for the M1, but may work –unchanged– on future chips as well. This is a very exciting opportunity in the ARM64 world. We won’t know until Apple releases the M1X/M2, but if we succeed in making enough drivers forwards-compatible to boot Linux on newer chips, that will make things like booting older distro installers possible on newer hardware. That is something people take for granted on x86, but it’s usually impossible in the embedded world – and we hope we can change that on these machines.

 

[Ajay]

Linux on M1? Is that even viable?

Well, BSD under the hood in MacOS. Apparently with a lot of Linux interop.

 

[StinkyPinky]

So if the rumors are true and the base models start with 16GB ram and 512GB nvme, I may trade in my 16" MB Pro for the 14" new model. Apple are generous with their trade-ins so shouldn't cost much, and just having normal ports again would be worth the price of admission.

 

[moinmoin]

Well, BSD under the hood in MacOS. Apparently with a lot of Linux interop.

They are both Unix OSes which follow some basic standards. But the (mostly) FreeBSD based userland Darwin/macOS uses can deviate significantly from the GNU Linux userland.

 

[Ajay]

They are both Unix OSes which follow some basic standards. But the (mostly) FreeBSD based userland Darwin/macOS uses can deviate significantly from the GNU Linux userland.

BSD and variants as well as System V predate Linux. Linux wasn't very well received by a lot of Unix folks at the time, kind of funny as it has become totally dominant in the *NIX world.
When I first used it for a project in '99, I wasn't very happy with it. Coming from SunOS and Solaris (Slowaris), enough stuff was different that it drove me nuts at first.

 

[moinmoin]

BSD and variants as well as System V predate Linux. Linux wasn't very well received by a lot of Unix folks at the time, kind of funny as it has become totally dominant in the *NIX world.
When I first used it for a project in '99, I wasn't very happy with it. Coming from SunOS and Solaris (Slowaris), enough stuff was different that it drove me nuts at first.

Linux has become so dominant for the same reason first DOS then Windows has, being readily accessible to anybody. Anyway you seem perfectly aware of the differences, just don't expect Darwin to be a drop in replacement for Linux.

 

[guidryp]

Well, BSD under the hood in MacOS. Apparently with a lot of Linux interop.

That's really irrelevant for getting native Linux running optimally.

The stumbling block is creating good reverse engineered drivers for all the custom hardware, or you end up leaving a lot of functionality behind.