"If you want to play around with those storage modules in a Studio, you should know that 1) you definitely need to do a full DFU erase, and 2) if you populate both slots, they definitely need to be the same size, and they might need to be the same vendor (there are several)."
I can’t speak for Apple’s design process, but for me, I’ve thought of and seen some personal projects use PCIe connectors because they’re cheap and have dozens of connections. I even saw a teardown on EEVBlog a while back of a piece of test equipment that used PCIe connectors for board-to-board interconnects.
For $2 or so in quantity 1, I can order an x16 connector from DigiKey and have 164 pins to connect two modules together. And being an edge connector, I don’t need one for each board (like “DuPont” or DIN 41612 does). If the board won’t be removed, you could even get away with the free tin plating (skipping gold).
> That's a nice way of saying "Deliberate hardware incompatibility"
As others have noted, that doesn't necessarily mean "incompatible primarily for the sake of breaking third-party upgrades."
It may very well be that there are advantages (such as better or more predictable performance, lower component cost, better power or heat dissipation, or higher reliability) to using raw modules with a custom controller rather than going through an SSD interface. Perhaps (for example) there are advantages to having the SSD's cache in the Mac's uniform memory and controlling when writes are performed.
(As an interesting aside, Steve Jobs famously convinced Shugart to make a stripped-down "SA390" floppy drive for the Disk II to save money. With Woz's controller and improved encoding they also managed to squeeze more data onto the floppy disks, so it saved [Apple] money while improving density and read rate - though possibly increasing the likelihood of bad sectors and data loss.)
Well we know they don't have significantly better power/heat because those aren't problems for regular SSDs. They may have lower component cost, but they certainly aren't acting like it (they're charging over 2x as much as high quality NVME SSDs). They might have better performance or reliability, but that is just wild speculation with no evidence to back it up.
There is also no evidence to back up the implication that Apple's primary reason was to prevent third party upgrades.
Based on what I've heard from people who work (or worked) at Apple, that isn't normally how they roll - usually there's a cost, performance, user experience, security, reliability, or functionality (etc.) reason and breaking third party upgrades, compatibility, or repairs is just a happy (for Apple) side-effect. ;-)
Placing the SSD cache in unified RAM would seem like a plausible performance and cost justification to me, not "wild speculation" as you claim. I've seen papers describing how to optimize the use of flash storage that require low-level control of the device as well as modifications to the OS/file system or even the cache and memory system - all things Apple could certainly do. However, as you note, evidence will tell.
Regarding whether Apple passes its own savings on to customers, I will note that the Disk II was said to be "obscenely profitable." In general Apple maintains some of the highest margins in the industry. Which is not to say Apple products are a poor value (I generally consider them an excellent value for what you get, and for my own use cases) but it is saying that they are very good at keeping money in their own pocket.
This isn't a competent counter-argument. All that means is that apple has spent literally millions of dollars deliberately engineering a worse solution than the NVME SSDs that have existed for almost a decade.
It seems like you can install a new module or a pair of identical modules, do a DFU restore, and install a new OS image. I bet eventually there will be aftermarket options.
It would likely be an NVMe or m.2 port, both of which I believe are much smaller ports that are standard on new motherboards that enable much faster bandwidth than anything over Sata. The most common internal SSDs that you can buy off the shelf and offer over 1GB/s read/write are going to be NVMe or m.2 drives (probably) and they're pretty fucking cheap, which is part of the reason people are progressively more upset about non-upgradable storage.
It's neither. These are not SSDs in the traditional sense, they lack the controller. They are raw storage modules that are paired with the M1's on die controller.
You might have been a bit eager to correct me. I wasn't referring specifically to the one mentioned in the article, just responding to the parent's more general question.
I'm not persuaded that the stated facts lead inevitably to the conclusion they reached. From another article (linked from the Verge post):
"* YouTuber Luke Miani decided to test whether or not the machine is actually upgradable through a series of comprehensive tests…"
"* Miani took the wiped SSD from the second machine and inserted it in the open slot on the first machine, but the Mac’s status light blinked SOS and it wouldn’t boot up"
"* He then swapped the SSDs between two Mac Studios and found again that neither would boot"
"* In a disappointing move, it appears that Apple has locked user-upgradability in software (the device recognizes the SSD, but Apple stops it from booting)."
Given Apple's recent attention to security at the hardware/firmware level, it is possible that, for example, Apple is requiring some type of hardware authentication from the SSD, and not simply to prevent the user from upgrading their SSD. (Along the same lines as the T2 chip, is what I'm imagining.)
I'm not a security person, so I don't know if this is likely. I'd be interested in the thoughts of any of you who do know this stuff. Is there a valid security purpose that could explain this?
T2-equipped macs do exactly this, with the idea being that nobody can pull the removable flash module and plug it into a compromised motherboard for easier cracking. The flash is cryptographically paired to the storage controller in The T2 (on Intel macs) and M1 chips.
> Miani took the wiped SSD from the second machine and inserted it in the open slot on the first machine, but the Mac’s status light blinked SOS and it wouldn’t boot up
Because the machine now sees flash that doesn't belong together in a single disk and (quite sensibly) refuses to assemble that into a volume. A DFU restore might fix this.
> He then swapped the SSDs between two Mac Studios and found again that neither would boot
Because the data is encrypted to a key unique to the original machine's SoC, so the other machine can't read it. A DFU restore is likely to fix this.
No I don't think so. It wouldn't stop the PC from booting up. Also, the scenario is different here. What will happen if you took two identical HP systems and swapped their disks?
Considering all hard drives eventually break down, even SSDs, this makes Macs disposable products.
Eventually that SSD is going to fail, but by then five or six years have passed and they'll just tell you to buy a new Mac. I'm very happy with my Mac Mini m1, and I'll probably buy a MacBook Air when the next one comes out, but I'm doing this with the understanding this is not going to last forever.
Unless you have a soldering iron, you're stuck, this is so frustrating because it artificially limits the lifetime of an expensive product. Hypothetically would say my m1 Mac mini breaks down in 5 years, I strongly doubt Apple will replace the SSD out of the goodness of their heart.
32 comments
[ 3.3 ms ] story [ 100 ms ] threadhttps://twitter.com/marcan42/status/1506022589357260801
"If you want to play around with those storage modules in a Studio, you should know that 1) you definitely need to do a full DFU erase, and 2) if you populate both slots, they definitely need to be the same size, and they might need to be the same vendor (there are several)."
https://twitter.com/marcan42/status/1506033231069855747
For $2 or so in quantity 1, I can order an x16 connector from DigiKey and have 164 pins to connect two modules together. And being an edge connector, I don’t need one for each board (like “DuPont” or DIN 41612 does). If the board won’t be removed, you could even get away with the free tin plating (skipping gold).
As others have noted, that doesn't necessarily mean "incompatible primarily for the sake of breaking third-party upgrades."
It may very well be that there are advantages (such as better or more predictable performance, lower component cost, better power or heat dissipation, or higher reliability) to using raw modules with a custom controller rather than going through an SSD interface. Perhaps (for example) there are advantages to having the SSD's cache in the Mac's uniform memory and controlling when writes are performed.
(As an interesting aside, Steve Jobs famously convinced Shugart to make a stripped-down "SA390" floppy drive for the Disk II to save money. With Woz's controller and improved encoding they also managed to squeeze more data onto the floppy disks, so it saved [Apple] money while improving density and read rate - though possibly increasing the likelihood of bad sectors and data loss.)
Based on what I've heard from people who work (or worked) at Apple, that isn't normally how they roll - usually there's a cost, performance, user experience, security, reliability, or functionality (etc.) reason and breaking third party upgrades, compatibility, or repairs is just a happy (for Apple) side-effect. ;-)
Placing the SSD cache in unified RAM would seem like a plausible performance and cost justification to me, not "wild speculation" as you claim. I've seen papers describing how to optimize the use of flash storage that require low-level control of the device as well as modifications to the OS/file system or even the cache and memory system - all things Apple could certainly do. However, as you note, evidence will tell.
Regarding whether Apple passes its own savings on to customers, I will note that the Disk II was said to be "obscenely profitable." In general Apple maintains some of the highest margins in the industry. Which is not to say Apple products are a poor value (I generally consider them an excellent value for what you get, and for my own use cases) but it is saying that they are very good at keeping money in their own pocket.
https://twitter.com/marcan42/status/1506039392884109312
"* YouTuber Luke Miani decided to test whether or not the machine is actually upgradable through a series of comprehensive tests…"
"* Miani took the wiped SSD from the second machine and inserted it in the open slot on the first machine, but the Mac’s status light blinked SOS and it wouldn’t boot up"
"* He then swapped the SSDs between two Mac Studios and found again that neither would boot"
"* In a disappointing move, it appears that Apple has locked user-upgradability in software (the device recognizes the SSD, but Apple stops it from booting)."
Given Apple's recent attention to security at the hardware/firmware level, it is possible that, for example, Apple is requiring some type of hardware authentication from the SSD, and not simply to prevent the user from upgrading their SSD. (Along the same lines as the T2 chip, is what I'm imagining.)
I'm not a security person, so I don't know if this is likely. I'd be interested in the thoughts of any of you who do know this stuff. Is there a valid security purpose that could explain this?
I think you’re on to something. I suspect we’ll get a tech note or presentation at WWDC that explains what’s happening.
Because the machine now sees flash that doesn't belong together in a single disk and (quite sensibly) refuses to assemble that into a volume. A DFU restore might fix this.
> He then swapped the SSDs between two Mac Studios and found again that neither would boot
Because the data is encrypted to a key unique to the original machine's SoC, so the other machine can't read it. A DFU restore is likely to fix this.
Eventually that SSD is going to fail, but by then five or six years have passed and they'll just tell you to buy a new Mac. I'm very happy with my Mac Mini m1, and I'll probably buy a MacBook Air when the next one comes out, but I'm doing this with the understanding this is not going to last forever.
Not at all. Apple can replace the flash module (it's not really an SSD) and, eventually, there will be some aftermarket options.
https://www.macrumors.com/2021/04/06/m1-mac-ram-and-ssd-upgr...
Unless you have a soldering iron, you're stuck, this is so frustrating because it artificially limits the lifetime of an expensive product. Hypothetically would say my m1 Mac mini breaks down in 5 years, I strongly doubt Apple will replace the SSD out of the goodness of their heart.