ssd swap - high usage of Terabytes Written

[jdb8167]

AppleInsider has some unofficial comment from inside Apple:

Questions raised about M1 Mac SSD longevity, based on incomplete data | AppleInsider

Complaints about the potential lifespan of Apple Silicon Mac storage are causing concern that the current crop of M1 Macs could stop working within a few years — and have been stoked by unanswered questions, unreliable data, and a silent Apple.

appleinsider.com

The quote by the anonymous Apple employee is:

"While we're looking into the reports, know that the SMART data being reported to the third-party utility is incorrect, as it pertains to wear on our SSDs" ... The source refused to elaborate any further on the matter when pressed for specifics.

It would be nice of Apple to supply documentation on what the specific SMART data means with offsets within the byte array to interpret the data. I've looked at the smartmontools code and they are using an outdated drive database as it pertains to Apple computers and even then, they are guessing (according to a code comment) about the offsets.

 

[fluxtransistor]

I have an M1 Pro, 16GB, 256GB, and here is my report:

SMART/Health Information (NVMe Log 0x02)
Critical Warning: 0x00
Temperature: 36 Celsius
Available Spare: 100%
Available Spare Threshold: 99%
Percentage Used: 0%
Data Units Read: 17,862,796 [9.14 TB]
Data Units Written: 9,680,256 [4.95 TB]
Host Read Commands: 209,160,040
Host Write Commands: 99,169,671
Controller Busy Time: 0
Power Cycles: 121
Power On Hours: 110
Unsafe Shutdowns: 12
Media and Data Integrity Errors: 0
Error Information Log Entries: 0

I've had it since Mid-December and I'm on it all day using Zoom and Safari.

 

[CMMChris]

@jdb8167 That statement is bullcrap. At least the TBR and TBW data reported by the drives firmware via SMART is correct. It absolutely matches writing stats of macOS itself.

 

[jdb8167]

@jdb8167 That statement is bullcrap. At least the TBR and TBW data reported by the drives firmware via SMART is correct. It absolutely matches writing stats of macOS itself.

I was thinking that this anonymous person was saying the percentage was incorrect not the amount of writes. To quote, "as it pertains to wear on our SSDs."

Certainly this statement does nothing to fix the ongoing PR problem that Apple is incurring. I hope they come out with an official statement and some tech documentation. There is no reason for these stats to be secret. SMART is a standard feature of SATA and NVMe drives.

 

[CMMChris]

Ah yes, I misread that. Wear percentage indeed might be incorrect but no one really cares about that value I guess.

 

[theSeb]

I have been wanting to buy the MBA M1 for so long. My plan is to just go ahead and buy it, then run DriveDx t o see how much is written and read. If the amount is reasonable, good. If not, return. This plan sounds ok?

Your plan does not sound ok. You should buy the M1 anyway, if that's what you were planning to do and use it. What is Ok/ Not Ok? What would be the threshold for the return in your eyes? There is no threshold, especially since Apple is claiming that the numbers reported are incorrect anyway. Buy a computer to use it, not to lose sleep over worrying about it.

For me this is an interesting thread because of the knowledge to be gained from it. It should not be used as a "point of concern" for existing, or prospective, owners.

 

[fluxtransistor]

Is it not possible that the values are off by a factor of 2 or 4, for example?
Also, I checked the Hours Powered On value after 2 hours, and it didn't seem to change.

 

[ian87w]

I guess it's good to wait up on the M1 mac purchase, at least until Apple figures this out.
Sometimes I dislike how Apple is so opaque about issues. Remember the butterfly keyboards, Apple PR was in complete denial until they themselves set up a repair program and finally ditched the design, proving that the issues were real.

 

[jdb8167]

Ah yes, I misread that. Wear percentage indeed might be incorrect but no one really cares about that value I guess.

I just verified that the "Data Units Written" is accurate. I started with:

starting units	12,724,656 [6.51 TB]
ending units	13,506,170 [6.91 TB]
total	781,514 [0.4 TB]

I wrote exactly 400 GB to my SSD and that is nearly exactly the number that you get multiplying 781,514 * 512 KB = 400.14 GB. You can see that the translation from units to TB is 12,724,656 * 512 KB = 6,515 GB and 13,506,170 * 512 KB = 6,915 GB.

The math checks out. The TBW numbers are correct. Whether or not the "Percentage Used" number is correct is yet to be seen.

 

[Spindel]

I just verified that the "Data Units Written" is accurate. I started with:

starting units	12,724,656 [6.51 TB]
ending units	13,506,170 [6.91 TB]
total	781,514 [0.4 TB]

I wrote exactly 400 GB to my SSD and that is nearly exactly the number that you get multiplying 781,514 * 512 KB = 400.14 GB. You can see that the translation from units to TB is 12,724,656 * 512 KB = 6,515 GB and 13,506,170 * 512 KB = 6,915 GB.

The math checks out. The TBW numbers are correct. Whether or not the "Percentage Used" number is correct is yet to be seen.

Or are your tools reading what MacOS is reporting (that might as well be written to DRAM on the drive) and not what is actually written to disk?

 

[Answer_i3]

Hmmm... maybe utils detects all actions with Unified memory? Not only ssd, but RAM?

 

[jdb8167]

Or are your tools reading what MacOS is reporting (that might as well be written to DRAM on the drive) and not what is actually written to disk?

Yes the tools use Apple's SMART monitoring system that is part of IOKit but the problem is that the details are not documented. The actual data is just a large blob of bytes and the software uses offsets into this blob to pull out data points. Not all drives have the same data offsets. That is why we don't know if the percentage used is correct because there is no way to correlate it with any user action.

That is not the case with writing to the drive. It is easy to write a specific amount to the SSD and then get the latest numbers from smartctl -a disk0. I wasn't writing to RAM or anything like that, I was writing a large block of data to the SSD. If it wasn't written to disk, then it wouldn't stick around after reboots. I'm not sure why you would think that writing to SSD is not persistent.

My methodology was pretty simple. I created an empty image file of 200 GB with the standard image format of a read/write disc. This allocates disk immediately. You can watch the disk writes using Activity Monitor Disk page and see that it writes just about 200 GB when creating the .dmg. I then created a a new volume on the same SSD and copied the 200 GB .dmg file to the new volume. This makes a duplicate (it took about 4 minutes to copy.)

 

[CMMChris]

Yes the tools use Apple's SMART monitoring system that is part of IOKit

SMART is not specific to Apple. It's a drive monitoring standard supported by hard drives and SSDs of all manufacturers and is present in most drives since the mid 90s.
It also has absolutely nothing to do with IOKit. On an SSD SMART stats are maintained by the drives controller that is managing the NAND memory. All those tools are just reading that information from the SSD controller. You can even read the SMART stats out while running Linux on the M1 and it gives the exact same values.

 

[ovenbakedlies]

@jdb8167 That statement is bullcrap. At least the TBR and TBW data reported by the drives firmware via SMART is correct. It absolutely matches writing stats of macOS itself.

My Machine has been up for 7 days since last reboot and this is what Apples own activity monitor is reporting (attached).

Over 5 TB written.

I have done virtually nothing of my own choosing that would write anything to the drive in that time (browsing, emails, a web client with no local data, that's it).

I believe the SMART data being reported is correct. The smartctl data agrees with their figure. Now I'm concerned that Apple's response is to blank the problem. Which means no solution.

 

[theluggage]

How does this prevent one from putting everything on an external SSD and booting from that bypassing this whole Soldered-in SSDs issue?

It doesn’t, but that’s hardly an ideal solution for a laptop, or even an ultra-small form factor computer like a Mini (which has plenty of space inside for a couple of SSD blades).

 

[CMMChris]

@ovenbakedlies What are the Apps that you have running constantly?

 

[tab0reqq]

I checked today again. Simple task manager and uptime.

3h from this uptime was Nvidia Ge Force Now app streaming Valheim They say, that with my settings up to 40GB per hour. So still that's huge writing/reading

 

[fluxtransistor]

Syspolicyd reads 353GB on my Pro.
What even is syspolicyd? 353GB sure is a lot.

 

[Maximara]

It doesn’t, but that’s hardly an ideal solution for a laptop, or even an ultra-small form factor computer like a Mini (which has plenty of space inside for a couple of SSD blades).

I disagree. External SSD are ridiculous small and with the size of many programs and their files odds are you are going to need one anyhow. SO why not have it as your boot up dravie? Sure is cheaper then Apple's option ($600 vs $350 for 1TB)

 

[Maximara]

View attachment 1735593
What even is syspolicyd? 353GB sure is a lot.

Google is your friend.

What exactly is syspolicyd and why is it running all the time?:

"syspolicyd was originally introduced in macOS 10.7.3 with the Gatekeeper feature. Its original purpose was to act as the centralized daemon for answering Gatekeeper questions. Today it still serves that purpose but its scope has greatly expanded. In addition to assessing applications before running, the daemon also handles authorizing the loading of KEXTs as well as tracking legacy applications that the user has run. In Mojave syspolicyd has expanded again and is responsible for handling software notarization checks as well. We’ll start with a very high level look at the daemon startup process and then dive deeper into each of syspolicyd’s subsystems."

 

[star trek]

Poisoned Apple

starring : Tim Cook

The most advanced OS in the world STRIKES BACK AGAIN !!!!!!

 

[theluggage]

I disagree. External SSD are ridiculous small and with the size of many programs and their files odds are you are going to need one anyhow.

They're still external (the clue is in the name) and having a laptop that is useless without an external device (and which will crash messily if that device is disconnected) is stupid. Either it's going to be hanging off on a cable which wrecks the whole portability thing, or it's going to be a dongle sticking out and risking damage to both the drive itself and the socket in the computer. Plus, it permanently uses up one of only two ports on the current M1 machines, so it's not even too smart on the Mini.

....and no, the fact that - if you need a lot of data on the road - you either have to sell a kidney upfront to get Apple's non-upgradeable SSD (at ridiculous prices even c.f. fast NVMe blades) or carry around an external drive is not a "good thing". Maybe excusable that the Air doesn't have a second SSD blade slot or even a replaceable main SSD - but Apple carry this all the way up to their high-end mobile workstations and Minis.

Sometimes the size and weight of a laptop needs to be offset against the number of bits and pieces you need to carry around to use it. Even on a desktop, less spaghetti and external boxes = good.

It's up to every customer to weigh these "cons" vs. the "pros" of having a Mac, but trying to pretend the "cons" don't exist is nonsense.

 

[Makkoo]

Since updating to Big Sur 11.2.2 this morning my swap hasn't been above 64 MB (M1 Mac Mini 16GB/512). So they must have included something in the update.

 

[Mc0]

Since updating to Big Sur 11.2.2 this morning my swap hasn't been above 64 MB (M1 Mac Mini 16GB/512). So they must have included something in the update.


View attachment 1735635

Was there a change in your current usage after doing the OS update? That’s a significant drop utilization wise. It’s holding me off in buying an M1. I need to replace my old laptop. I hope they fix this issue soon.

 

[Makkoo]

Was there a change in your current usage after doing the OS update? That’s a significant drop utilization wise. It’s holding me off in buying an M1. I need to replace my old laptop. I hope they fix this issue soon.

No, didn't use it differently than before the update. Just opened Final Cut and Logic Pro together with safari playing YT vids, swap still at 64MB, memory pressure around 25%.