is there a real limitation with this architecture or are they just making a business decision so they can sell multiple models over 2-3 years instead of creating what the market wants immediately
is there a real limitation with this architecture or are they just making a business decision so they can sell multiple models over 2-3 years instead of creating what the market wants immediately
It's just about scale for the time being, probably.
The M1 chip is now in 5 different devices. That is 100% worth not fulfilling everyone's RAM needs for the fantastic economies of scale and engineering simplicity that brings. They'll come out with an M2 soon enough, which will undoubtedly have more RAM and they'll stick in the iMac Pro, Mac Pro, and MBP 16" (and possibly "backport" to other devices).
That being said I just "downgraded" from an Intel 16" MBP with 32GB RAM to an M1 13" with 16GB RAM and haven't noticed much of a difference, even though I was regularly getting to 80-90% RAM utilization on my 16".
I think that only applies to their mobile devices. I have had Intel Macs that can't run the latest MacOS but can still Boot Camp into Windows 10. Of course, this is all moot in a post-Intel Apple world.
Also, Windows 10 in 2021 is far from being exactly like Windows 10 in 2015. There have been a lot of changes, Microsoft just doesn't increment the major number any more.
If you're saying that Windows 10 from 2021 is Windows 10 from 2015 with "minor updates", you haven't been paying attention to Windows updates :-)
Windows 10 updates are probably bigger than MacOS updates.
This is like saying Mac OS X is 10 year old software.
I agree with the latter, but tell that to their previous default of 4gb ram and 120gb ssd (for so long) and 16gb of storage on iPads.
Every single device ever made has to make tradeoffs. No device can be everything to everyone. Apple has (correctly) identified that RAM is really not the end-all-be-all that people obsessed with specs seem to think it is. If you spend more time working on other aspects of the device you can get away with less RAM and still have a device that people resell years later for good money.
Would you have bought a MacBook air or pro in 2017 with 4gb of non-replaceable ram for $1000? No, because it's obselete and you know that. What if they told you it was a great computer, and you didn't know anything about computers? Probably.
That'd then be 1 module = 8GB, and 2 modules in a daisy chain or T-topology gets you 16GB, and there just isn't anything more to be done without a bigger memory controller (more channels) or bigger DRAM modules.
That's likely the largest that LDDR4x is ever going to go because the tech is outdated. If they had a 24GB option, you'd hear a lot fewer complaints (and the cost difference isn't massive).
They announced 16GB (128Gb) LPDDR5 chips were in mass production in Q3 of last year. These both use less power and have better performance than LPDDR4x.
Both Intel and Qualcomm have support for LPDDR5 in their chips that launched last year. No excuse for Apple here.
That said, I did straight up run out of RAM the other day, but my editor was chewing up something like 32gb of swap. Still no slowdown, though. (I suspect a memory leak, it's very new software)
Not currently doing any Android work so I can't comment.
They have cut CPU costs with M1 so keeping RAM costs down ready for material price cut on these models next year.
So as the sibling said: CPU is binned (but apparently only for the 7/8 GPU cores), RAM is added separately in the package, not on the die.
From the marketing, it's clearly a consumer-oriented device. Presumably there will be "pro" versions at a later time.
Micron and Samsung make 12 Gb LPDDR4X chips, but Apple probably decided that having a 24 Gb option wouldn't be worth it, or would be an ugly number, or couldn't get enough supply, or all of that.
The M1 is for "consumer" tier devices anyway. Wait for M1X or Mwhatever – when they port the larger MBPs to their own silicon, that will have larger memory.
I would go with supply combined with having enough supply of the 8gb modules but not enough of the 12gb ones that it would create strange tiers, while they would lean towards a 24gb machine
While it would be fun to be able to run a complete k8s cluster locally that's quite a niche usecase. There's not much software on the consumer side where having 32gb+ really makes a huge amount of difference. At least not the software Apple is showing in their presentations. Lots of people think they want a huge amount of memory, but macOS seems to be able to handle memory pressure (with transparant compression and swapping) quite well, even on machines with little memory.
I don't know why it's so hard to find a 3840x2400 24" display.
Apple is a company. It is beholden to focus on profits for it's shareholders.
Why give users more RAM for free?