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.
First M2 Macs Expected to Launch Later This Year Using TSMC's 4nm Chip Process
- Thread starter MacRumors
- Start date
- Sort by reaction score
Seeing how Apple handles their chips. What I can expect is to see the speed from the M1 go up, but I don't know if the basic M2 will beat an M1Pro chip. If it brings the whole line to where the bottom speed is M1Pro speed (with less GPU cores because Apple), that would be good enough. If it will take years for the low end to get to that speed then why bother upgrading till then?
Also, they could make the M1Epic (M1Omni? M1Alpha? M1Omega?) chip for the Mac Pro have four M1 chips on one die. Or skip that entirely and put four M2 chips on one die and slowly put M2 chips in everything else afterwards.
I hope Apple can just change the chips in the computers without changing the design at all. That way they can focus on making them faster at the same price every year rather than making them look different and charging you more.
Also, they could make the M1Epic (M1Omni? M1Alpha? M1Omega?) chip for the Mac Pro have four M1 chips on one die. Or skip that entirely and put four M2 chips on one die and slowly put M2 chips in everything else afterwards.
I hope Apple can just change the chips in the computers without changing the design at all. That way they can focus on making them faster at the same price every year rather than making them look different and charging you more.
Speculation is very difficult, since the chip cadence is being rearranged to accomodate both the new broader chip design roadmap and TSMC's upstream process. There's obviously no guarantee that there will be a new M series for every A series release, for example - but we also don't know if the current pipeline of M1 upgrades (M1 -> Pro/Max -> Ultra) or even that sort of tiering is going to be true for future chips.
What is likely from Apple operational behavior is that they will try to grab as much of TSMC's capacity for a new process as possible, so it is likely they will continue to slate A series chips for that. As capacity ramps up, Apple will have capacity to produce other chips on that process, and will probably have a M-series chip ready to go.
What applies pressure is that Apple wants to have lots of new hardware in the channel at the year end, both for consumers (Christmas presents of the new shiny) and businesses (end of year spending, new year budget). They also want to have phones available for Christmas. If nothing changes, that means that A series and M series have to launch pretty close to one another, often on a brand new process. Launching a high volume product like a phone on a new process means risk to Apple's biggest quarter, both due to TSMC execution and yield.
I actually suspect we may see an inversion, where M series chips start to roll out on new processes earlier than their A series counterparts.
Does Apple already cover SVE2 and security similar to Confidential Compute Architecture? Those are probably among the most interesting features of armv9.
Just rehashing old rumors that so far haven’t been validated. M2 could have been M1 Ultra (dual M1 Max) just as the M1X was M1 Max. Apple completely taking the 27” iMac off the marketplace was totally unexpected. Still expect it sometime after sales of Mac Studio have caught up with demand and a 27” promotion display is being made in adequate quantities. It’s very likely Apple considers the 24” iMac acceptable placeholder until conditions are right for the M1 Max iMac to ship.
Apple doesn’t need to capture more value from the M1 as its already moved down to the $599 iPad Air. Moving to M2 on the MBA and iMac for Q3 BTS sales is a more logical next step. The M1 Pro, Max and Ultra are with us for some time to come and the iPad Pro won’t get update to M2 until Q4/22 or Q1/23, giving apple
Plenty of M1 sales including the Air 5. Leaving the M1 past 2 years (November 22) in the MBA/MBP/mini and iMac is probably not a great idea marketing-wise, but Apple is used to doing this very delicate dance and doing it well. Moving the M1 to the Air is the first piece of the M2 migration puzzle.
Same for me. I can't handle the big form factor of the Studio mini. Also, I mount my mini under the desk so thinner is better. I am hoping for an m2 mini with an even thinner case and external power brick with Ethernet.
Aww, I'm waiting for a light and fanless MacBook Air with a native dual monitor support! I was hoping for an M2 MBA during the spring event. Fingers crossed for June's WWDC!
I'm going to go with M2 being A15-based, albeit on the rumored 4nm process rather than 5.
Let's look at it this way...
Start with the most basic and widely-used implementation. That would be iOS (iPhone and lower-end iPads), which outnumbers the units manufactured of all other products by a wide margin. Fewer GPU cores, less RAM, less need to support peripherals.
Then then move to the M-series for devices manufactured in the highest numbers. More GPU cores, more memory, support for external displays, USB-C/Thunderbolt. First released is the "basic" M-series for the most widely-sold Macs and higher-end iPads. As more and greater capabilities are added, the number of Macs that will use those implementations continues to drop, until we reach the highest-end Mac.
As long as the A-series comes out annually, M-series will also be refreshed annually, with the last iteration of M-x arriving just before or coincident with A-x+1. Tick tock, tick tock.
Apple didn't take control of its silicon to then hold back on advancing the tech every year in every product. That was one of the problems they were solving by ditching Intel.
Let's look at it this way...
Start with the most basic and widely-used implementation. That would be iOS (iPhone and lower-end iPads), which outnumbers the units manufactured of all other products by a wide margin. Fewer GPU cores, less RAM, less need to support peripherals.
Then then move to the M-series for devices manufactured in the highest numbers. More GPU cores, more memory, support for external displays, USB-C/Thunderbolt. First released is the "basic" M-series for the most widely-sold Macs and higher-end iPads. As more and greater capabilities are added, the number of Macs that will use those implementations continues to drop, until we reach the highest-end Mac.
As long as the A-series comes out annually, M-series will also be refreshed annually, with the last iteration of M-x arriving just before or coincident with A-x+1. Tick tock, tick tock.
Apple didn't take control of its silicon to then hold back on advancing the tech every year in every product. That was one of the problems they were solving by ditching Intel.
M series has already exceeded annually. It’s very rare for Macs to be updated on a yearly basis. Even iPads isn’t quite that.
Why is everyone saying that M2 will be based on A15/A16? At this point I think it's much more likely that the M chips come first and the A chips will be based on them.
Thus the M2 may well be a new chip and a slightly cut down version is then produced as the A16. Initially I thought that Apple could update all its Macs every year by just bumping the chip up the next iteration similarly to how they worked back in the old days of Intel actually managing a significant chip update each year. Now I'm wondering if Apple is on a 2-3 year cadence like the iPhones so we get year 1: new M chip, year 2: nothing (perhaps RAM or SSD silent bump), year 3: new form factor and then back to a new chip and the cycle continues.
Thus the M2 may well be a new chip and a slightly cut down version is then produced as the A16. Initially I thought that Apple could update all its Macs every year by just bumping the chip up the next iteration similarly to how they worked back in the old days of Intel actually managing a significant chip update each year. Now I'm wondering if Apple is on a 2-3 year cadence like the iPhones so we get year 1: new M chip, year 2: nothing (perhaps RAM or SSD silent bump), year 3: new form factor and then back to a new chip and the cycle continues.
According to Anandtech it's a little more than that. Given the load of security issues that require emergency OS updates we've been seeing it sounds like a (potential) necessary security improvement.
Based on silicon die complexity, the A series is much simpler. Lower CPU and GPU core counts, slower memory controllers with lower bandwidth needs, no need for things like Thunderbolt or PCI controllers, etc. The A series also power's Apple most-important product: the iPhone.
So to my thinking, it makes far more sense for Apple to focus on the A-series first and then use that as the foundation for all the other Apple silicon products.
If it's based on the A15 then it will still be ARMv8. ARMv9 was only finalized a year ago and these chips are likely designed years in advance.