On the Mac Studio’s Extra SSD Slots | Hacker News Reader

On the Mac Studio’s Extra SSD Slots

512pixels.net

17 pointsingve11 comments

What strikes me is the entitlement of some Apple customers. Just because Apple put an extra SSD slot in, for unclear reasons, unadvertised and not listed anywhere as a feature, this entitles some customers to insist the ssd slot work the way they want it to work, while the peanut gallery non-Apple customers go nuts about it. How can Apple abuse their customers this way?! Oh, the humanity!

what strikes me about your comment is how you consider it entitled to want control over one's own hardware. Like, I paid money and have this hardware in my hands, I wonder if I can somehow fill up this empty slot and get some extra space. Fuck me, though, right?

Yes, that is entitlement to expect to be able to gain control of something that before 2 days ago no one outside Apple knew existed. For all we know it is for Apple's internal use only and can't be used otherwise without Apple's internal proprietary (perhaps diagnostic) equipment. Or maybe it is a fully functional ssd slot that's only purpose was redundancy to increase yields of functional Studios, and by design, only one or the other can be used at once. Or maybe you're absolutely free to use it, but you're going to have to figure out how to get it working yourself. It is entitlement to expect anyone else, including Apple, to make it work for you. Or maybe the slot works fine, but the ssd's won't play with any other ssd on an internal bus. No one really knows anything at all at this point, so it is entitlement to expect satisfaction when no one has any idea what the hell is happening.

There’s nothing stopping you from reverse engineering that slot and making it work in the same way Asahi Linux took on a much more challenging task by reverse engineering Apple’s GPU to run Linux.

It is pretty annoying though how in the Intel Mac mini they offer memory upgrades but if you go to a European Apple store they’ll tell you you can’t get the upgrade from them. And that would be even more annoying if there’s some artificial limitation to prevent you from doing it yourself.

I'm glad that this article doesn't do any editorializing. Perhaps I'll put this site on my reading list.

I can imagine that the impetus for this is exactly that of hard drives of olde. Back when Sun (and Netapp, etc) used to have Sun-specific firmware, ostensibly more thoroughly debugged or at least somehow optimized for a particular workload (imagine locking some key ufs data in the drive cache, no matter the access pattern). Same for netapp and friends. Of course, this also enabled them to sell the drives for an obscene markup, but perhaps you did get some value add from it. Their target customer (large enterprise) didn't care all that much about the money grab; computers were just a tool and costs get passed to the end user.

Back in ye olden days, I can completely believe that OEM drive firmware sucked in various ways, and that vendors actually improved upon it. Nowadays, I don't believe that for a second, for drives.

For SSD, I am 100% certain that there are more crap ones out there than quality ones. This is Apple protecting their brand. You don't want large numbers of random crashes from a bad SSD to be finger pointing at Apple. It makes perfect sense, apart from the blind spot they have about their own quality (lol).

As an aside, today vendors still play this hard drive markup game, but without the firmware value add. You can't buy the drive sleds apart from a drive and its 100% or more markup. Once the Chinese companies reproduce the sled, the vendors change to a new sled design. I don't quite understand why there is such a long lead time in cloning the sled. It can't be related to part complexity. There must need to be a certain volume of aftermarket buyers to make it worth reproducing, so around the time that used arrays become common enough in the market is around the time a new sled comes out.

I don't follow right to repair much, but I hope making sleds and other such parts available is part of the agenda.

>This is Apple protecting their brand.

I mean, it's certainly plausible that's part of it, and that's certainly giving them the benefit of the doubt. I suspect they are considering the financial incentives as well though. Also, I'm not sure making computers non-upgradable through software/firmware when there is a physical socket is good for their brand either. Their markup on SSD's is pretty high.

So, as it turns out, surprise, those slots in the Studio are not SSD slots, and the boot volume is not an SSD. They are slots for raw NAND storage, the cards are simply NAND memory chips. The SSD controller is integrated in the SoC. So an SSD will not work in those slots, period, not ever.

All this speculation, I’m shocked nobody has actually tried to put an SSD in that Second slot yet.

Don't think "SSD slot". The controller is on the M1 SoC (or the T2 for the Mac Pro). It needs raw access to the flash chips.

A better analogy would be if NVMe drives came with ZIF sockets for the flash chips so there was only one board part number regardless of capacity. The flash chips don't have controllers or any logic on them, they're just flash chips. The manufacturer programs the controller with how many chips there are and the striping, wear leveling, etc at the factory.

Now imagine someone says "I'm going to plug a different NVMe drive into the ZIF socket on the first NVMe drive!" That isn't going to go well. Or imagine you add more chips to empty ZIF sockets. The data is already striped and laid out for the original number of chips. To change the layout you'd need to reprogram the controller at minimum.

They have. There are a couple videos on YouTube of people trying to place the “drive” from another Mac Studio in the extra slot, as well as the primary slot.

The cards are linked to the motherboard and will only work in the system they came with. Maybe in the future some hacking could unlock it, but it doesn’t seem like Apple will ever officially support it.

Reply on news.ycombinator.com