Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
What an amazing job!
Also I have an A2991 16” M3 Pro that comes with only 512GB (4x 128GB) from the factory. I'd like to upgrade to 1TB. Somebody know what NAND model suites better? And when changing the NANDs do I need to pre-program them or do they just have to be blank?
just ensure the use of blank nands and the m3 pro is actual quite a special device as its the 1st one you can upgrade to the very max 4TB without the need to add any passive components. just the 4 nands of the correct capacity for the desired config

for the price and the effort level required id suggest going for the 4tb upgrade as will also make the device more valuable when it comes time to sell/upgrade.

to go from 512 to 1TB all nands will need to be replaced exactly the same as you would for a 4tb upgrade will only cost you slightly more in terms of nand chips
 
Last edited:
  • Like
Reactions: anakinfan8
The rail name is "SOC_FORCE_DFU", and needs to be shorted to 1.2V rail while plugging power into the machine (via USB-C power adapter). It needs to continue to be held shorted a few seconds after power is applied. It will pull about 100 mA at 5V from USB-C if done correctly.

Also you should remove UN400 as well.
I'm going to attempt that now thanks for the help.


also I don't mean to 2nd guess you here but I've double checked a MacBook M1 Max with a 4 and config and that keeps the UN400 present and it seems to be apart of the SSD_4L requirement as it shows as "Critical" on the schematic and from what I can see this 338S00600 “Ocarina” IC is needed to supply voltage to the only 4 nands that are installed.

the UN900 is the same 338S00600 “Ocarina” IC but is on the side with the vacant landing pads hence why it made perfect sense to remove it from that side when you suggested it... but you lost me totally when you suggested removing the UN400.
 
okay so I got into DFU mode after the suggested method and in the restore process I kept getting this error

Failed to restore device in recovery mode, libusbrestore error:2006 [com.apple.MobileDevice.MobileRestore – 0x7D6 (2006)]

most places talk about it being a USB connection issue. but I'm using an official apple wire that been used multiple time and have swapped it out with a like for like just in case, no joy.

made sure I'm using the correct DFU port (next to MagSafe port).

tries different ports on the host Mac tooo. still no joy



any ideas?
 
Last edited:
I'm going to attempt that now thanks for the help.


also I don't mean to 2nd guess you here but I've double checked a MacBook M1 Max with a 4 and config and that keeps the UN400 present and it seems to be apart of the SSD_4L requirement as it shows as "Critical" on the schematic and from what I can see this 338S00600 “Ocarina” IC is needed to supply voltage to the only 4 nands that are installed.

the UN900 is the same 338S00600 “Ocarina” IC but is on the side with the vacant landing pads hence why it made perfect sense to remove it from that side when you suggested it... but you lost me totally when you suggested removing the UN400.
Sorry, yeah UN400 should remain. Only the second Ocarina IC and the 5th NAND must be removed, everything else can stay.
 
Sorry, yeah UN400 should remain. Only the second Ocarina IC and the 5th NAND must be removed, everything else can stay.
Thanks for confirming that for me. so any idea what would be preventing the restore then?

it gets about half way through the installing process but keeps throwing up that error code.

Failed to restore device in recovery mode, libusbrestore error:2006 [com.apple.MobileDevice.MobileRestore – 0x7D6 (2006)]

have you seen much of these before, and how do you over come them? as I can't see no issues with the USB-C connection from host Mac to the one im upgrading

still getting the boot looping (power meter shaping that it is negotiating t 20v and then starts again on a constant loop, no image on screen)
 
Last edited:
just a little contribution, as for Louis Rossman not being comfortable with "donor" components

there is two things to say about that :

1 - don't ask me how, but there is for years Chinese suppliers for brand new, original caps, inductors and more importantly brand new NANDs and power supply ICs.
(those are KICM223R and 338S000600 IC's in the case of A2442/A2485 8TB configs)

The Chinese market for NAND upgrades for iPhones is a thing for years, it is established and huge, and there are quality and trusty sellers over there.
I think we can be grateful of the knowledges and quality of work of Chinese technicians for this.

Our luck is that Applesilicon Macs (even the M2 Mac Pro) do use the exact sames NANDs found in every iPhone
(M1 mac use the same NANDS as iPhone 13, most M2 Mac use the same NANDs as in iPhone 14, etc).

You can check the references I have collected in the table of the first post of this thread.

It's simple logic : Apple achieves economies of scale, better pricing and overall quality by sourcing only one type of NAND for all its products: iPhones first, iPads second, and Macs third (and accessories at last).

So, in our misfortune that Apple solder and glue NAND chips, we are fortunate that there has been for years a mature iPhone upgrade market in China for years that we can take advantage of.



2 - The NAND chips used in iPhones and Apple Silicon Macs are really solid.

I mean really really really solid.

As a test, I've been trying for months to "kill" two Mac Mini M1s I owned before upgrading them to 2 TB.

I wrote with a script 700TB of data before their SMART "remaining life" status went to 0% which is remarkable for 256GB SSDs. This is a 700TBW "official" lifetime.

But there's more : I've continued the test and is is still not over.
At that time, my 2 mac minis both continue to accept writes at the nominal rate (3GB/s on the 1st third and 700MB/s thereafter as they're in TLC).

I'm up to 2.3PB of written data on one mac mini.
And 4.5PB on the other I started earlier (and still running).
4.5PBW... for a 256 2-chips SSD...
And both are still running day and night, just like the first day.

View attachment 2346747

what will be the lifetime of the 8TB SSD ? I'm going to test it in the Mac Studio cards clones I'm starting to built.


I'm writing this message on a 13" M1 MBA that I upgraded to 2TB a year and a half ago. With a quite intensive use, I am at ~60TBW so around 40TBW a year.

Of course due to my job I'm a backup maniac, so that's a reason not to be worried.
But I'm also not worried at all about reaching my 2TB SSD's lifespan any time soon because it should take me anything between 10 and 100 years to reach the lifetime of this 2TB SSD : all other parts of the mac would be dead before the NANDs will start to fail...

Forgive me if I'm wrong, but even if I was soldering used NANDs on a customer board, there would be no real danger here.

The real problem is that someone CAN solder and sell used NANDs.
No one should do this and I will never because in my case I have a trusty seller and always put new NANDs.

But the real problem is that once reprogrammed, used NANDs will show a false "data unit written" value of 0...
Hello, I am curious if you have determined the point at which an SSD "fails"?
 
hi, so im trying to upgrade the nand on my mac mini m4. i successfully replace the nand, managed to install the os via DFU but now my fan is spinning at max volume the moment i turn it on. i tried different ssd module and the fan is normal.

i know we dont have the schematic for this mac mini nand board, but can someone maybe point to me wich resistors or pin is responsible for detecting nand temperature.

the picture attached is from m2 pro schematic i assume because they are using the same type of nand, i could figure out why my board is acting up but i just dont understand which line control the temperature sensing. maybe anyone here can enlighten me. thanks
 

Attachments

  • IMG_0104.jpeg
    IMG_0104.jpeg
    187.6 KB · Views: 8
  • IMG_0106.png
    IMG_0106.png
    1,003.7 KB · Views: 9
Hello, I am curious if you have determined the point at which an SSD "fails"?
Hi,

Here are my results yet :
  • Mac Mini M1 A2348 - 256GB SSD (2 samples) : 0% lifetime after 660TBW, SSD dead after 10 PBW on average (one made 9,500 TBW the other 10,200 TBW)
  • MacBook Air M1 A2337 - 2TB SSD (1 sample) : still no sign of death at 5.1PBW, is at 10% lifetime remaining
  • Mac Studio M1 A2615 (6 samples) : only 10% of lifetime used at 2.6PBW
So far, the only SSD I managed to make dead was the Mac mini SSD, at 15x the time of its official lifespan reported by the SMART status. Its performance remained stable at 3GB/s read, 3GB/s write inside the SLC cache, 650MB/s outside the SLC cache up to ~9PBW when it then degraded.

I first used an applescript that triggered "Amorphous DiskMark" benches, but it was too long and would have taken 10 years to reach 10PBW. So a Mac Studio user (Max Liu) created from my instructions a complete test software (generate and write true random data to a RAM disk, then write the random data sample many times to the SSD, and report data by email).

It is incredibly difficult to wear out these SSDs.
 
Hi,

Here are my results yet :
  • Mac Mini M1 A2348 - 256GB SSD (2 samples) : 0% lifetime after 660TBW, SSD dead after 10 PBW on average (one made 9,500 TBW the other 10,200 TBW)
  • MacBook Air M1 A2337 - 2TB SSD (1 sample) : still no sign of death at 5.1PBW, is at 10% lifetime remaining
  • Mac Studio M1 A2615 (6 samples) : only 10% of lifetime used at 2.6PBW
So far, the only SSD I managed to make dead was the Mac mini SSD, at 15x the time of its official lifespan reported by the SMART status. Its performance remained stable at 3GB/s read, 3GB/s write inside the SLC cache, 650MB/s outside the SLC cache up to ~9PBW when it then degraded.

I first used an applescript that triggered "Amorphous DiskMark" benches, but it was too long and would have taken 10 years to reach 10PBW. So a Mac Studio user (Max Liu) created from my instructions a complete test software (generate and write true random data to a RAM disk, then write the random data sample many times to the SSD, and report data by email).

It is incredibly difficult to wear out these SSDs.
Fascinating data! Thank you for your effort in analyzing this information. Does a "dead SSD" imply the entire Mac is inoperable as well? Is it possible to boot from an external disk or perform any other actions with a Mac that has a "dead SSD"?
 
Does a "dead SSD" imply the entire Mac is inoperable as well?
Yes, absolutely. A dead internal SSD will cause the Mac to stop working, and if it has an LED, it will flash orange with an SOS morse pattern.
All T2 and Apple Silicon Macs from the iMac Pro up to the latest M4 Mac won't boot without a working internal SSD.

Is it possible to boot from an external disk or perform any other actions with a Mac that has a "dead SSD"?

No unfortunately not. The secure boot process involves a hidden partition inside the first NAND chip. There is no way for any Apple Silicon or T2-based Intel mac to boot without it.
 
  • Like
Reactions: ducnsh and Borowski
Yes, absolutely. A dead internal SSD will cause the Mac to stop working, and if it has an LED, it will flash orange with an SOS morse pattern.
All T2 and Apple Silicon Macs from the iMac Pro up to the latest M4 Mac won't boot without a working internal SSD.



No unfortunately not. The secure boot process involves a hidden partition inside the first NAND chip. There is no way for any Apple Silicon or T2-based Intel mac to boot without it.
Thank you. I have a concern about the SSD on my Mac. I am using a 16" M1 Max with 32GB RAM and 1TB storage. The S.M.A.R.T analysis of the SSD appears unusual. It has recorded approximately 19TB written, yet the percentage used is already at 2%. I am uncertain about the reason, as I have meticulously cared for my Mac since the first day.
 
Thank you. I have a concern about the SSD on my Mac. I am using a 16" M1 Max with 32GB RAM and 1TB storage. The S.M.A.R.T analysis of the SSD appears unusual. It has recorded approximately 19TB written, yet the percentage used is already at 2%. I am uncertain about the reason, as I have meticulously cared for my Mac since the first day.
Some perspective: You have used 2% over about three years, assuming you bought the M1 Max in fall 2021. If you continue writing to your SSD at the same rate you have been, it will take (100%/2% * 3yr) = ~150 yr to raise "percentage used" to 100%. The computer will be a museum piece you'll not be interested in using anymore in a mere 20 years (probably much less).
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.