If all of those were true, Apple would not bother shipping it.
I expect none of them will be true other than #4 and that is because there is nothing faster available for SoC applications.
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.
M1X worst case scenario
- Thread starter JMacHack
- Start date
- Sort by reaction score
If all of those were true, Apple would not bother shipping it.
I expect none of them will be true other than #4 and that is because there is nothing faster available for SoC applications.
Except it is not a one-year old SoC. It is a new SoC based on a year-old generation, but with significantly more CPU and GPU cores to significantly increase performance.
Intel does it with their "Extreme" gaming CPUs and Xeon workstation/server CPUs and you see people shedding out thousands - even tens of thousands - on these "old" CPUs not complaining "it's a year old!" because they educated themselves on the details and didn't focus solely on the generation.
We see the same within GPU generations, as well, and people buy them, as well, for the same reasons.
Well, Apple has made dubious decisions before. And I’m of the opinion that the released product will be better than my outline. But it’s a thought experiment of the worst case scenario.
Fair enough, but they have M1 as a baseline so if "M1X" offered very little benefit, I don't see them going forward with it, instead waiting for real benefits with "M2" (the next generation of Apple Silicon on Mac).
It’s likely that the M1X MacBooks were supposed to be released this summer. That would have put them 6-months behind the M1 which is completely reasonable and expected. It’s possible the pandemic caused massive delays, which set everything back by 6 months.
From a marketing standpoint, it would be odd to put an M2 into the new MacBook Pro unless in the marketing materials they have to specify core counts (which gets a bit tedious and awkward). It’s odd because what would they put into the new MacBook Air next year? Also an M2? Or an M3? Would they have to specify in marketing the core counts every time?
Its possible that the next chip is named something entirely different. Like P1 or M1 Pro or something. I’m not great at marketing but I do feel they need to establish some sort of hierarchy in the naming. Otherwise I see a flood of comments like “wait the new MacBook Air comes with an M3 chip? Is that better than the MacBook Pro M2 chip? What a rip off. I just spent $3000 on this dumb thing!”
I don’t think it was ever going to come out in the summer, not enough leaks about it.
I saw plenty, and Gurman who is the most reliable leaker, stated we could see them as early as “this summer” back in the spring.
I expect we will continue to see CPU and GPU core counts in the marketing as based on the rumor mill / leaks the "M1X" will come in 16 GPU and 32 GPU core variants and I would expect that to continue with future generations (along with the rumored new variants with 20/40 CPU cores and 64/128 GPU cores).
As noted, there was a fair bit of chatter about a possible WWDC reveal. General speculation was that the 14" and 16" MiniLED panels were too constrained in supply to support an actual launch at WWDC unless Apple wants to go with an extended (3-4 month) delay in shipping date.
Personally I think there would have been much more information on the device if it was only a few weeks from release.
Well, there is LPDDR5... and even if Apple keeps LPDDR4X, they can increase the bus width (something they will have to do anyway)...
Yes, and there was no other choice since Intel made the CPU, and Xeon had other features which professional users might want.
Another example is the 2019 Mac Pro which used a Skylake Xeon. The Skylake microarchitecture was released in 2015, so from a CPU microarchitectural standpoint it used a four-year-old design. That same Mac Pro is being sold today with the same CPU, which is now based on six-year-old microarchitecture. Intel did slight tweaks to that architecture but even up to Cooper Lake it was essentially identical to Skylake. Yet those CPUs had broader features that professional users wanted and were used in many workstations and servers.
Probably sometime next year the new Sapphire Rapids Xeon will be released, the first really major Xeon architectural improvement in seven years. So if anyone thinks M1X being based on one-year-old M1 microarchitecture is bad, try seven years.
The glacial rate of Intel's progress shows shows the benefit of Apple doing their own CPU design. M1X will likely have double the CPU multicore performance and up to four times the GPU performance. Then in 2022 there will probably be new Apple Silicon internal microarchitecture, maybe based on ARM9 and likely improved fabrication.
Regardless of which Apple Silicon CPU is used in the new 2022 Mac Pro, I doubt there will be a seven year delay in fundamentally updating it. I also doubt Apple/TSMC will stay on the same fabrication node for *six* years like Intel did from Skylake to Rocket Lake (14 nm).
M1 systems are getting a bit long in the tooth so M2 systems should be around the corner. Are you really gonna sink serious $$$ into an M1X product for it to be runover by the new M2 systems that appear a few weeks later?
If people are expecting the M1X machines to thrash the M1 at “general” tasks that aren’t optimised for multi-threading or GPU-based computing then they will likely be disappointed... Just as you would be with Intel if you paid top dollar for a Xeon or i9 extreme and expected Excel to run more smoothly. That’s not what the M?X processors are for.
(I’ve said before that Apple really need to dump the M?-suffix naming scheme and establish a separate branding for their “pro” chips)
For a lot of “pro” apps - media production, scientific computing etc. - which tend to be optimised for multi-core and/or GPU based computing, doubling the number of CPU and GPU cores will likely have a more dramatic impact on performance than moving to a slightly faster core design - in some cases, perfomance could almost double. For those cases where jobs actually need more than 16GB of RAM, more RAM could also have a more dramatic effect than slightly faster RAM. More external displays and maybe more Thunderbolt ports are also essential for some.
Going by the rumours, the 2021(ish) high-end Macs are getting the M1X and the 2022 Air is getting the M2 - that means the M2 will be optimised for ultraportables and tablets which probably means that size and power consumption will be the priority - not a huge increase in the number of cores. It might also make sense to keep the restrictions on displays, max RAM and I/O - which aren’t really a big issue on an Air or iPad. Maybe there will be a M2X a year or so down the line, or maybe the “X” will skip a generation (since it will sell in lower numbers and Pro customers wont want to be the ones to debug the new core design).
That could very well happen. It is possible that an M2 could come out with faster single core performance, but still limited by RAM, ports, etc, while the pro version has more (older) cores and RAM and external display capability.
Because, M2 is obviously 1 better than an M1X.
Sigh. I rest my case for why Apple need to find a better naming scheme than the one being used in rumour.
...and I bet you an internet that the M1X will thrash the M2 on multi-core benchmarks and many video encoding/rendering tasks.
1. Perhaps up to 3.6GHz on the P-cores.
2. Maybe an increase of one or two. It’s extremely improbable to be four or whatever some users are wishing.
3. I am assuming the addition of 32GB and 64GB options.
4. Yes. I don’t see DDR5 support happening before the next gen. (i.e., “M2”).
5. Two more P-cores.
6. Probably. I’d expect four, maybe eight more (i.e., 12 and 16-core GPU)
What’s the badge/achievmeent for having the closest (or exact) prediction?
?
If only. Too bad, I think, it’s a pipe dream at any point now that people are seemingly okay with the bump, double bump, triple bump, whatever.
I am expecting the prosumer chips (whatever the name) to be faster in single-core. To be considered a success, Apple needs to outperform Alder Lake as well. Whether Apple will clock the prosumer chips higher, or whether they will use a new microarchitecture to achieve this remains to be seen, but I doubt they will use Firestorm@3.2 in these machines.
I mentioned something similar a while ago as well. P-series sounds nice.
There has been a significant amount of information if you believe the leaks:
- A new SoC with 10 CPU cores and a GPU with 16 and 32 core options
- At least 32GB RAM support and possibly 64GB support (though more recent leaks point to 32GB max)
- MiniLED displays
- 1080p FaceTime camera
- Three USB4+TB ports, MagSafe and HDMI (and SD Card?)
- Removal of the Touchbar and a return to physical Escape and Function Keys
- A chassis design more reminiscent of the iPhone 12 and iPad Pro
The "M2" should outperform the "M1X" in single-core operations because it will use a newer Performance Core from the A15. And it will have longer battery life because it will have newer generation CPU and GPU cores and a fair bit less of each. I would also expect it to remain limited to 16GB maximum of RAM and two USB4+TB ports.
But "M1X" will still handily outperform "M2" on multi-core operations and graphics operations. And it will still have excellent battery life. And there will be more RAM, more ports, probably larger SSD storage capacities, and other differentiators.
As for myself, I will likely buy the 14" or 16" "M1X" MacBook Pro even though they are overkill for what I need, but I do want to future-proof myself. But I might wait to see what the "M2" MacBook Air brings to the table because once it adds MiniLED (as rumored), it arguably nicely fits my actual needs.
Like the A13 runs over the A12X? Oh wait it doesn't. Single gen jumps won't make a small chip more powerful than last years large chip.
I believe it will, just not at the levels you appear to be associating. Apple will probably assign a different prefix for the Mac Pro SoC.
AMD: Ryzen = consumer/prosumer, Threadripper = workstation/server
Intel: Core = consumer/prosumer, Core-X = workstation, Xeon = workstation/server
Apple will make sure that they demonstrate a noticeable improvement in performance, otherwise everyone will claim that the M1 was an anomaly and there is no future in their chip strategy. Everything will be fine and the next chips will be a lot better.
Its crazy to claim that a one year old chip is long in the tooth.
Which is fair. I think it’s much easier to differentiate M1, M1X, M1P (i don’t know just making up names) while having 2 variants of each chip, than to just have one name for the entire generation, like M1 meaning 8,10,16,32 cores and a array of gpu options.
Either way, whatever they do they’ll have to somehow differentiate the levels in marketing so it’ll be interesting to see how they do this so there’s no confusion among your average consumer. Then again processor naming has always been confusing to the average consumer.
I'm calling it
(Replace # with generation number)
M#X -> 14/16" MBP + most [27-32]" iMacs.
M#Z -> Highest end [27-32]" iMac (Pro). Possibly optional upgrade for second generation 16" MacBook Pro. Will also surface in a G4 Cube style mini Mac Pro
W# -> Mac Pro chip family.
W#X -> Upgraded Mac Pro chip - effectively 2x W# in a tile/chiplet setup.
If Apple eventually want to go even further with extreme CPUs (which I doubt) I would foresee
Z# and
Z#X, possibly even
ZX#
This is a long term nomenclature prediction. I do not necessarily predict this will hold for the first generations of listed products. For example, the first Mac Pro may use an M#Z and only go to a W# after a generation or two with no W#X until even later.
Furthermore, eventually, after 79 generations of Apple Silicon chips, Apple will wrap around in history, and develop an x86 compatible CPU that'll be called the ZX80
*badum-tisch*