S6 has been claimed to be 7 nm. I don’t know about S7 but it has the same T8301 product number and same performance so I suspect it’s the same chip on the same process.
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.
3nm M2 Pro Chip for MacBook Pro Reportedly Entering Production Later This Year
- Thread starter MacRumors
- Start date
- Sort by reaction score
They'll be called whatever the Apple marketing department decide to call them. These are just labels, folks!
You'll have a long wait.
Intel Macs were basically PC clones with nicer cases and trackpads, the firmware was closely related to PC firmware and all the GPUs, storage controllers, interfaces etc. were designed for PCs and had Windows drivers readily available. BootCamp was little more than a point-and-drool tool for making the minor tweaks necessary to get a standard Windows distro to install on a Mac, setting up dual boot and installing drivers.
About the only thing Apple Silicon Macs have in common with other ARM systems is the ARM instruction set, pretty much everything else - graphics, storage, I/O - is done by Apple's proprietary system-on-a-chip and the firmware/bootloader is more like the iPad than any of the emerging standards for regular ARM systems. Microsoft would have to expend significant effort producing an Apple Silicon version of Windows-on-ARM complete with bootloader and custom drivers for Apple Silicon hardware, something that Apple have explicitly said that they're not going to support.
Virtualization works because Parallels etc. can intercept calls to already-supported Windows-on-ARM hardware and pass them to the drivers in MacOS.
Nah.
M1 Max MacBook Pro came out in October 2021.
M1 Max Mac Studio came out in March 2022.
It's pretty unlikely that the Studio will get bumped to M2 until it's at least a year old - probably more like 18 months.
Anyway, it's not like the regular M2 is proving to be a "must have" upgrade from the M1 - and remember that some of the "new" features the M2 gained over the regular M1 - like faster LPDDR5 RAM and the hardware ProRes codec - are already present in the M1 Pro/Max. Also, while the power consumption/thermal advantage of 3nm could be a big deal in a MacBook Pro it's not such a game changer in a Studio which is already looking to have somewhat over-specified cooling...
less than 1 quarter from chip fab to product launch?
You're forgetting logic board assembly, housing/structure of the MBP, quality control, LCD panel/lighting ports in the supply chain then product stock levels and shipping.
Q2/Q3 of 2023 at least is my guess.
I understand the sentiment, but it's unclear that this is true. I suspect most Watch energy goes to display and radios, and if those are improved each year, a CPU boost is of limited value.
Of course the way you get to ever lower energy is by improving every little small thing! So I wouldn't complain if we got a new core, but I suspect that's the reasoning.
The other interesting question is whether there is value in Apple adding a tiny core to complement the existing Thunder core, to perform the background tasks and updates that are most of the compute done by the watch, and that don't need to be that fast. Or is that handled adequately by just running the CPU at a substantially slower frequency?
Yet a third dimension of interest is whether more energy is actually burned by the DRAM than by the CPU? Perhaps Apple Watch is an ideal target for something like MRAM, acting as a pool of "large enough" random access storage that may be slower than DRAM but that doesn't need on-going refresh energy?
For many year, after Apple first migrated to Intel chips, they would do CPU spec-bumps more frequently. Annual was common and sometimes more than once a year as new chips became available. It was seen as a good thing and not a problem.
Now that these MacBooks are all on Apple Silicon, we can expect regular chip updates. The jumps will be incremental not revolutionay. These will be incremental updates that add 10-20% performance improvements. You existing MacBook will not be obsolete. Just using them side-by-side with the new model you will probably have to really try to tell a difference.
And the Max is just a Pro with an extra helping of GPU cores. Unless you are using those extra cores, it’s just filler. All of the other functions are identical between the Pro and Max.
For a few years apple did deliver dual designs to Samsung and TSMC. I guess metals were similar enough for that to work but not exactly the same
Ironically, a few people disagreed with me and responded to my post saying that Apple would be able to launch the MBP by the end of *this* year.
You may very well be right and a second quarter launch in 2023 would happen--I'm no expert and I'm just speculating for fun. However, doesn't Apple always get pre-production samples of components so that they can plan out assembly ahead of time? Additionally, not much is supposed to change other than the M2 upgrade and perhaps a bump in base storage/RAM--so not a huge engineering challenge.
I still think a 1st quarter launch in 2023 is feasible-but maybe Apple will want to wait for other reasons.
It's not stupid at all. Apple is naming them based on the microarchitecture rather than the node size:
M1: based on A14 (Perf. cores = Firestore, Eff. cores = Icestorm)
M2: based on A15 (Perf. cores = Avalanche, Eff. cores = Blizzard)
Indeed, this naming approach makes sense because Apple's responsible for advancing the microarchitecture, while TSMC is responsible for advancing the node size. So by using this naming scheme, Apple is taking credit for what they accomplished rather than what TSMC accomplished.
3nm... Simply amazing. I remember reading back around Y2K about how they were thinking 90nm was likely going to be the physical limit for CPUs and they would have to find other ways to make CPUs go faster since throwing more transistors at the problem would no longer be offset by advances in miniaturization and would start making the chips progressively larger, hotter and more power hungry. Moor's Law may have fallen by the wayside, and the approach to speeding up chips has changed, but obviously they were very wrong about that 90nm barrier.
What is important to you? Boot Camp support or running Windows? Because those are not the same thing!
Boot Camp support requires supporting all the Apple hardware, requiring MS to write many custom drivers.
Running Windows simply means using a virtualization app (there are already dozens of those, but Parallels probably makes it the easiest) and using a standard version of ARM Windows that communicates with the standard virtio drivers. And you can do it today.
But of course the quest to complain never ends, so I expect there is some super important reason why virtualization is not and never will be an acceptable solution...
According to Wikipedia, S4 through S7 have all used various flavours of 7nm?
Apple silicon - Wikipedia
There’s 25 other letters in the alphabet to move to when that day 20yrs down the line comes and the 14” is less expensive than a Thinkpad x1 carbon so not seeing your point.
Apple already uses A, H, M, S, T, U, and W.
1.8 is probably happening in three or four years.
If you want to get a 14 (it's what I'm typing this on) and not use it professionally, you do you, but I'm betting most people wouldn't buy a $2000+ machine for that.
the problem was that 3nm wasn't ready for the M2 so they had to stay at 5nm. That made the chip relatively underwhelming in terms of the upgrade from the M1 and also made it run hotter and have to be much larger than it should have been. Due to the chip size of the m1 pro/max already being much larger than the m1 since they are basically multiple m1s stitched together they just don't have room inside the MacBook Pro to do a 5nm m2 pro/max chip since it would be too physically large for them to fit an m2 max on the board especially for the 14. If they are forced to stick at 5nm you might get a 14 that can only come in a pro and not a max version
I'm skeptical of this argument.
Let's do a simple back-of-the-envelope calculation: The M1 Max chip is 420 mm^2. Approximating it a square for the purposes of this basic analysis (it's not--it's a non-square rectange), that's 20.5 mm on a side. The M2 has 20B/16B = 25% more transitors than the M1. If the M2 Max is larger by the same proportion, that would increase its die area to 420 mm^2 x 1.25 = 525 mm^2. Again approximating it as a square, that works out to 22.9 mm on a side.
So effectively you're saying that the 14" MBP doesn't have room for a chip to increase in size by 2.4 mm (= 1/10") on each side. That seems highly unlikely!
I expect the 14"/16" M2 MBPs will use 3 nm chips. But if they stayed at 5 nm, size would not be an issue.
Last edited:
Can it mean Apple will employ similar strategy like with iPhone 14? Basic M chip will be build on older technology compare to Pro/Max/... to make difference more noticeable? Or they just need to keep thermal performance manageable?
My gut feeling says that TSMC 3nm process was delayed and Apple was left with the choice of either delaying the M2 or manufacture it using the "old" 5nm process.
You may be right. It would explain thermal throtling of 10 core GPU under heavy load.