M1 SSD high read and write usage per smartctl

The part of smartctl relevant to SSD life is the Percentage Used. You have about 1% used, but since it's rounded it could be anywhere from 1-2%, so you have between 98-99% of your SSD life left. Your Mac was 3 months old when you posted this, so you can expect 150-300 months life out of it.

Percentage Used:                    1%
Data Units Read:                    29,157,637 [14.9 TB]
Data Units Written:                 24,058,076 [12.3 TB]

My own 3.75 year old 15" MBP is showing Percentage Used: 21%, so I can expect another 14 years from it. It shows that my SSD TBW is estimated at 1000TB. If yours is also 1000TBW, you would expect about 250months life based on 12.3TB in three months

Percentage Used:                    21%
Data Units Read:                    513,695,304 [263 TB]
Data Units Written:                 444,977,968 [227 TB]

However, Robert Follis is showing a real issue

My M1 MBP 8/256 has used 21% of SSD life in 5 months writing 258TB !! to disk.

He has used as much of his SSD lifetime in five months as I have in nearly 4 years.

All I can suggest is he carefully check what's doing the writes. Activity Monitor -> Disk -> order by Bytes Read will show what's doing the most writing to disk.

Of course, the expected life is only an estimation. We would need to be able to track actual failed SSDs against the estimated TBWs. I'm sure manufacturers are doing that but I don't know how we, as consumers, would get real numbers.

There is also no telling how smartctl is getting

its information for "lifetime". Since it is not a published

number for Apple SSDs, they are likely using some low end

guess.

I have used smartctl to get info on SSDs, and different manufacturers

supply a wide variety of different information support and some

and not all adhere to the SMART standard and often take

"artistic license" in encoding their information. There was an iteresting

article on the DriveDx blog that pretty much said maintaining their software

is chasing a moving target, if it is even visible.

All that aside, while the smartctl absolutes could be more or less meaningless,

the relative numbers from one user to another (assuming they are on exactly

the same platform) are not.

My MBpro M1 16/1tb is showing 6.6TB written in less than 1 month of lite use. My windows laptop has been thrashed for 3 years,(serveral VMs, 2 permanently running DBs, thousands of app builds etc) and only has 3TB written. looking at activity monitor, the one which is always 100x more writes than any other process is kernal task (currently at 100GB for no apparent reason since reboot this morning).

Just checked and Kernel Tasks IS the culprit, 637GB written since the last reboot yesterday

Three different utilities all give 22% of disk life used since December 20

Tried Apple support who were not much help on this issue

Anyone know how to escalate with Apple this and get it taken seriously?

Sfromgi, I don’t think you or the OP have anything to worry about, but Robert.follis does.

Your use isn’t too far off my 45 month old MBP and I don’t have any concerns about mine. I’m pretty sure I can get at least the same amount of time out of it but I’ll probably replace it long before then.

it’s been 15+ years since I used Windows so I don’t know what it is you’re using to measure writes on your Windows laptop, but even if its TBW was at the lower end, 150TBW, that would be an expected lifetime of 100 years. So assuming 3TB in two years is correct, it’s overkill.

I’m willing to bet that the writes are almost exclusively due to swapping, so you need to look at the Memory tab in Activity Monitor. But go to the menu, View -> Columns, and turn on the Real Memory columns. macOS processes share a lot of memory, so looking at the Real Memory use tells you how much each process is using exclusively.

As for me, I’m a software developer who does a lot of work in AWS. I work with two different AWS accounts so I usually run two different browsers so I don’t have to switch accounts. I used Firefox and Chrome. Beginning last year I started to get a lot of memory pressure issues - causing a lot of swaps, and I could see that it was the Chrome tabs grabbing a lot of memory, so I switched to using Safari. I haven’t had any memory pressure issues since.

Robert, can you check Real Memory usage when you see kernel_task using a lot of CPU? Is there anything that stands out, eg Rosetta 2 apps being the main offenders?

I’m afraid, unfortunately, that 8GB RAM isn’t enough for you. I’m pretty sure it’s swapping to disk.

woodmeister50 wrote:

I have used smartctl to get info on SSDs, and different manufacturers

supply a wide variety of different information support and some

and not all adhere to the SMART standard and often take

"artistic license" in encoding their information. There was an iteresting

article on the DriveDx blog that pretty much said maintaining their software

is chasing a moving target, if it is even visible.

I agree. The drive manufacturers of hard drives and SSD have shown over the years they sometimes have broken SMART implementations. For example for years I have seen the "Power On Hours" for hard drives to show one value while the time stamp for a log entry is completely different. This is very basic functionality and reporting yet the drive manufacturers get this wrong all the time. Sometimes the timestamps for the logs are all the same or even zero. There are other examples where the SMART attributes are implemented outside of spec. Most of these drive manufacturers don't even document these health attributes. It is a shame that such a useful monitoring feature has so little attention to detail and that the health monitoring of NVMe SSDs has become so simplified. I'm assuming they are hiding the details because with each advance on SSD NAND tech the speed and life expectancy of the NAND gets worse & worse so they want to hide this from users.

I personally believe the smartctl utility is most likely properly retrieving the information from the SSD. I personally believe the SSD is most likely reporting the information incorrectly. However, like all the drive manufactures Apple will never tell us if this is the case and it will most likely remain unknown until we see the SSDs fail or we see these TBW exceed the common values for SSDs.

Most NVMe SSDs only report the bare basics unlike the older SATA SSDs where we had the actual known values & thresholds to translate them into easily understood indicators. Only the "RAW" values were to be questioned since the utilities were calculating the "RAW" value which typically included the exact number of errors or the exact amount of data transferred.

Until someone verifies that the values for "Data Units Written" actually matches the amount of data truly written to the SSD through a test we have no idea if this value is truly accurate or whether we are off by a certain unit of measure. We'll also have our answer once one of these SSDs wears out if these reported values are actually correct. Most standard SSDs typically have a max TBW (TerraBytes Written) of 300TB (of course this varies especially by the size of the SSD).

Opening Safari with 1 tab spikes Kernel Tasks from 0% to over 5000% and 1 tab uses 1.03GB of memory

Only other app showing any major memory use in BitDefender

Just uninstalling BD to see what happens

FYI other people with 16GB RAM are reporting exactly the same problem - See @sfromgi in this thread "My MBpro M1 16/1tb is showing 6.6TB written in less than 1 month of lite use" That's even worse than me

In terms of kernel task using up so much writes, the kernel task

is doing all the "low level" work that is required by all apps that

make any system requests and all other tasks to keep the

computer working.

So, it is quite possible you have an app that is generating runaway

requests of the kernel that itself does not show up directly as

the apps disk use, such as apps disk cache or scratch disk use.

Also, look at the data you collected from those apps and do the math

and you will see that they reports must be questionable. If

your drive has in fact done 260TBW and a lifetime is generally

300 TBW (Apple states no spec on that) it would mean you have 87%

of life used not 22% and if you take the 22% used and the writes

reported, that would mean the drive has a life span of nearly

1200TBW which isn't even matched by high quality enterprise

SSDs which are in the 700-800 TBW range!

So, what is reported is severely suspect by all apps!

Robert Follis wrote:

6.6TB a month = 792TB annually which is the total life of some SSDs

He's reporting 6 point 6 TB per month, not 66TB.

Anyway, I think what we're forgetting here (myself included) is that the Percentage Used number is based on what the Apple SSD itself is reporting, so if it is reported as 1000TBW, that's pretty much what Apple is guaranteeing.

So, in the case of Robert, I'd say your best bet is to just use the Mac and forget about the SSD usage, and aim for it to reach 100% before it's out of warranty., as long as it isn't impacting your use of the Mac. If Apple doesn't honor its hardware warranty and/or implicit 1000TBW guarantee then you and anyone else in this situation will have a good legal case. Just make sure you have a backup.

As for sfromgi, such an option is little less clear, but I'd suggest forgetting about "lite use" and just go ahead and use it as normal, and keep regular backups. It's unlikely that sfromgi's SSD will fail due to writes. Like I've pointed out, mine is 227TBW in close to four years and has zero errors, but if his usage with other than light use is much higher, he'll also be in line for a warranty replacement.

Same for me 4 months old SSD, now 44 TB of SSD writes, 50 TB of reads. I do not even transfer large files or use time machine, this is crazy... My Macbook Air 2013 catalina os has written a lot less...

Watch this: https://www.youtube.com/watch?v=09iO4TuVkCA

All those people saying its not an issue take note: The M1 is demonstrably writing 10-60x more to the SSD than an intel mac doing the same thing with the same apps and the same memory, including M1 native apps from Apple.

In my case, at the current write rate, the SSD will fail in under 2 years, but outside the warranty, and irreparably.

Apple are silent - they dont want to admit liability, but whats not clear to me is what, from a technical point of view, could possibly be causing this? Is it a major flaw in the apple silicon architecture, or a bug in the OS? If it were a bug in the OS, presumably Apple would have addressed it already.

This is an opportunity for a 3rd party to analyse the problem and produce some sort of fix. Either HW - e.g. a TB connected SSD to take the fall, or SW (e.g. limiting the number of apps in memory to a low number).

I am lucky, I kept my old 15" 2017 intel MBPro when I bought the M1, so I can switch back to it when the SSD's time comes. I should switch back now - being able to use 2 monitors again would increase my productivity significantly, but I do like the battery life and small size of the M1. Tough.

sfromgi wrote:

Watch this: https://www.youtube.com/watch?v=09iO4TuVkCA

Interesting video. Thanks. Here is another video by the same person that includes some brief information about how SSDs work as well as trying to identify the maker of the SSD itself and the SSD's possible endurance and what it may mean for the M1's SSD and how to minimize the SSD's wear (hopefully until Apple fixes the issue if possible). The interesting part for me begins at 4:12 with attempting to determine the manufacturer of the SSD.

https://www.youtube.com/watch?v=e8aCgnb_RQs

Apple are silent - they dont want to admit liability, but whats not clear to me is what, from a technical point of view, could possibly be causing this? Is it a major flaw in the apple silicon architecture, or a bug in the OS? If it were a bug in the OS, presumably Apple would have addressed it already.

Apple is always silent as they love their secrets. Apple is unlikely to admit to the issue unless they can actually fix the issue, otherwise it opens them up to lawsuits. You may have to actually wait until a lot of people have their M1 Macs' SSDs fail prematurely so that Apple is unable to hide the premature failures. According to the video I linked it is likely to happen to high use professionals in a about a year of use. The video mentioned at least one M1 Mac has already died due to a worn out SSD although he did not mention where this news was reported.

I am lucky, I kept my old 15" 2017 intel MBPro when I bought the M1, so I can switch back to it when the SSD's time comes. I should switch back now - being able to use 2 monitors again would increase my productivity significantly, but I do like the battery life and small size of the M1. Tough.

Definitely keep the old laptop especially if your workload is with lots of data usage.

Has anyone here seeing excessive writes raised a support ticket with Apple?

If enough people do this instead of just railing about perhaps Apple will

investigate the situation.

I for one have not been seeing this issue, at least not enough for concern.

And I know I push my machine often to the point of memory swapping.

My usage has been an average of ~27GB/day since the end of last

November to date. Based on that, I have an SSD life expectancy of

somewhere around 19 years if one is assuming a 200TBW life time which

is likely lowballing it for a 512 GB drive (more likely 300-400). That is

also assuming that the SSD chips Apple is using are consumer level chips.

So, there are possibilities that there is a user element in the phenomenon

in terms of how the machine is physically used by the individuals with

the issue as well as what they have installed on their systems. These

days, who knows how apps are written and what things they do

"behind the scenes". And certain combinations could be leading

to "runaway" scenarios.

I might be lucky I guess.

System reports a time since boot of 48 days and kernel_task has only written 44,06GB for me inside activity monitor. Weirdly enough in the bottom it says written data 196GB and 404GB read. Swap usage is 716MB with 6GB RAM being used according to the memory tab while I have nothing open in the background.

I don't do heavy tasks on my macbook, atleast not regularly so I believe the 44GB written is much better than what I see in the couple hundred GBs or even TBs of writes. And that's for 48 days of use.