ssd swap - high usage of Terabytes Written

[IceStormNG]

sudo launchctl unload -w /System/Library/LaunchDaemons/com.apple.dynamic_pager.plist
sudo nvram boot-args="vm_compressor=2"

You have to disable SIP for the boot args to be changeable. After changing the setting, you can enable SIP again.
Reboot is required.

To revert:

sudo launchctl load -wF /System/Library/LaunchDaemons/com.apple.dynamic_pager.plist
sudo nvram boot-args="vm_compressor=4"

 

[Apple_Robert]

That is an excellent video. Thanks for posting.

Even with all the downloading and installing I have been doing, my average has been 31GB per day. And on the base model, (256GB) anything over 75GB per day for extended time periods, will slowly decrease the life expectancy.

I agree with him. There is a lot of clickbait fear mongering being perpetuated on YouTube (for revenue purposes) right now. The average user who doesn't have a memory leak program running etc. shouldn't have to worry about killing the SSD in 2 years.

 

[TheSynchronizer]

Whay Apple is silent on this topic? Any idea?

because
a) then they would be admitting it’s their mistake (lawsuits?)
b) they don’t have the fix yet or are working on it but it’s not conclusive
c) it would make the issue a lot more popular and a lot of people who don’t realise they’re experiencing it would notice -> more issues for apple.

it’s in their best interest to be as silent as a mouse, unfortunately.

 

[Dockland]

While I'm on an Intel machine that also has quite high write values, I did some testing here. As I have enough RAM, and I didn't liked how macOS swapped for no reason, I disabled swap.
But: You cannot fully disabled it. No matter what you set, it still swaps during sleep. But not while it's active. Now I have almost no writes except for what my applications write.

But: Do not do this if you're low on RAM all the time. My machine has 32GB RAM and all my apps that I use at the same time fit into that without any problems. If you have a 8GB RAM machine, do not disable swap. On 16GB and more you could try this if your apps all fit into RAM. If not, you might see system instability or crashes.

What I would like to see is a sysconfig value like linux's swappiness. So the OS doesn't swap just because it's bored but only if needed. Even if it doesn't effect my SSD lifetime, it's unnecessary to swap all the time and writing to SSD costs more power and takes longer than simply leaving stuff in RAM.

There's no need for "free up" RAM. Unused ram is wasted RAM.

Loved this part of your post "So the OS doesn't swap just because it's bored..."

 

[Leon1das]

sudo launchctl unload -w /System/Library/LaunchDaemons/com.apple.dynamic_pager.plist
sudo nvram boot-args="vm_compressor=2"

You have to disable SIP for the boot args to be changeable. After changing the setting, you can enable SIP again.
Reboot is required.

To revert:

sudo launchctl load -wF /System/Library/LaunchDaemons/com.apple.dynamic_pager.plist
sudo nvram boot-args="vm_compressor=4"

This doesnt work on M1.
After SIP disabling, reboot, Terminal & launching commands, reboot - I launched memory demanding app (Witcher 3 in CrossOver) and my swap is used (Activity Monitor shows use around 1Gb swap after 1min of play)

 

[ManicMarc]

While I'm on an Intel machine that also has quite high write values, I did some testing here. As I have enough RAM, and I didn't liked how macOS swapped for no reason, I disabled swap.
But: You cannot fully disabled it. No matter what you set, it still swaps during sleep. But not while it's active. Now I have almost no writes except for what my applications write.

But: Do not do this if you're low on RAM all the time. My machine has 32GB RAM and all my apps that I use at the same time fit into that without any problems. If you have a 8GB RAM machine, do not disable swap. On 16GB and more you could try this if your apps all fit into RAM. If not, you might see system instability or crashes.

What I would like to see is a sysconfig value like linux's swappiness. So the OS doesn't swap just because it's bored but only if needed. Even if it doesn't effect my SSD lifetime, it's unnecessary to swap all the time and writing to SSD costs more power and takes longer than simply leaving stuff in RAM.

There's no need for "free up" RAM. Unused ram is wasted RAM.

Are you sure that M1 Macs save their state to disk when they are put to sleep? There are other posts on here about data being lost when the battery dies while asleep.

 

[wirtandi]

11.3 Beta 4 has been significantly better for me in terms of swap used and disk writes.

Yay, another one with success on 11.3

 

[DeanL]

because
a) then they would be admitting it’s their mistake (lawsuits?)
b) they don’t have the fix yet or are working on it but it’s not conclusive
c) it would make the issue a lot more popular and a lot of people who don’t realise they’re experiencing it would notice -> more issues for apple.

it’s in their best interest to be as silent as a mouse, unfortunately.

Or because
a) Apple doesn't issue press releases or support documents about every single thing MacRumors Forum users are concerned with.
b) This forum is an echo room.
c) No one here actually has proof that the data being reported using the third party apps they have used is actually accurate/that the data is being interpreted correctly, or that the SSDs in Apple computers wouldn't be able to withstand the amount of data written if the data reported turned out to actually be accurate.

 

[osplo]

Dont get me offensive or something - I just hate when someone is telling me "this is not an issue, this is normal". No, it's not normal

I am sorry to hear that you still have the issue. I never denied it. I just say it is just a handful of you guys left with severe symptoms. And I stand by the fact that Apple already sold millions of M1 machines. If the issue were widespread there would surely be more than about 12 people trying to find a solution for it, here and everywhere. Even discounting the non-technical fellows and such, as you pointed out.

As I see that your Intel machine also has the issue, I agree with you that the problem could reside not in M1 chips by themselves but in Big Sur or a combination of Big Sur and the apps/workflow. It could even be a faulty disk controller or even some SSDs (yes, it could be a hardware issue or a combination of hardware and software) which is somehow aggravated by Big Sur handling of swapping or something.

I do want it to be fixed for you and whoever has it, by Apple or by third-party app developers if they are the culprit or the accomplice.

Again, what I am trying to say to people who need a machine today is that they should definitely go and buy an M1 computer. It is really fantastic. And if they happen to have the abnormal write behavior they can return it after a few days.

BTW, when you bought the M1, did you start from scratch, with a clean install, or moved data from your Intel machine or Time Machine? I started from a clean install, and my Mac is barely writing to the SSD. Did you try to see what happen in Safe Mode? Did you try to work with a single app at the time to try to zero in the culprit (if it is an app)? Just trying to help here.

 

[Apple_Robert]

Or because
a) Apple doesn't issue press releases or support documents about every single thing MacRumors Forum users are concerned with.
b) This forum is an echo room.
c) No one here actually has proof that the data being reported using the third party apps they have used is actually accurate/that the data is being interpreted correctly, or that the SSDs in Apple computers wouldn't be able to withstand the amount of data written if the data reported turned out to actually be accurate.

You bring up some very good points.

A lot of people seem to be jumping on the blame Apple bandwagon with this topic, when we don't have any corroborative facts.

 

[DeanL]

You bring up some very good points.

A lot of people seem to be jumping on the blame Apple bandwagon with this topic, when we don't have any corroborative facts.

Especially given that it started as an M1 issue (this thread is located under Apple Silicon) but then people seem to exhibit this "issue" with Intel Mac.
It's frustrating to not get answers from Apple but at the same time imagine the volume of inquiries of that kind they get.
I trust that that kind of issue would have been picked up a long time ago given all the sorts of analytics collected by Apple (including bug reports filed by people).

 

[TheSynchronizer]

Or because
a) Apple doesn't issue press releases or support documents about every single thing MacRumors Forum users are concerned with.
b) This forum is an echo room.
c) No one here actually has proof that the data being reported using the third party apps they have used is actually accurate/that the data is being interpreted correctly, or that the SSDs in Apple computers wouldn't be able to withstand the amount of data written if the data reported turned out to actually be accurate.

Well yes, ofcourse, in the grand scheme of things it’s too early to tell whether this issue is an actual problem or whether these SSDs are designed to be written to so much. Just like it was too early to tell when people were rushing to buy their 2016 MBPs with butterfly keyboards.

Also, this issue isn’t just isolated to this forum, as a matter of fact it’s been documented and discussed now by many media outlets, tech channels, and other forums such as reddit. So it’s not just inside our macrumours echo chamber

Do also keep in mind that just like the butterfly keyboard, not everyone was, is, or will be affected, but there are a few that definitely are and that does not make this any less of an issue.

And personally I was experiencing this issue heavily on my M1, which accounts for the 21.3TB it has written since launch, but now I’m averaging writes of 20-30GB per day while maintaining the same workflow and using the laptop for the same things, as a result of multiple fixes and tuning which I believe all contributed to fixing my write problem. So this is in my opinion most definitely caused by software, and from what i’ve observed specifically by various parts of Apple’s software, including Big Sur and what came with it eg Rosetta 2.

Whether they’re planning to fix it or it’s working as intended only Apple knows. But I know i’d rather have SSD writes on the level of my previous Intel Macbook Pro, rather than these new weekly Terabyte writes that I was experiencing.

 

[DeanL]

Also, this issue isn’t just isolated to this forum, as a matter of fact it’s been documented and discussed now by many media outlets, tech channels, and other forums such as reddit. So it’s not just inside our macrumours echo chamber

Can you let me know which media outlet has discussed this beyond the usual "MacRumour users are reporting..."?
Do these media outlets base their reporting on the same basis I have already questioned? (i.e. using software that might interpret incorrectly the data)

Do also keep in mind that just like the butterfly keyboard, not everyone was, is, or will be affected, but there are a few that definitely are and that does not make this any less of an issue.

The Butterfly keyboard issue is not comparable. It was observable with the naked eye (i.e. if your keyboard is broken... It's broken+teardowns could show the design flaw).
This supposedly SSD issue is not. First, it's only observable with third party software that again, might be misinterpreting the data. If the data is accurate, the lifespan calculation could be made on outdated SSD life expectancy models as many people have pointed out before.

 

[nobackup]

Well yes, ofcourse, in the grand scheme of things it’s too early to tell whether this issue is an actual problem or whether these SSDs are designed to be written to so much. Just like it was too early to tell when people were rushing to buy their 2016 MBPs with butterfly keyboards.

Also, this issue isn’t just isolated to this forum, as a matter of fact it’s been documented and discussed now by many media outlets, tech channels, and other forums such as reddit. So it’s not just inside our macrumours echo chamber

Do also keep in mind that just like the butterfly keyboard, not everyone was, is, or will be affected, but there are a few that definitely are and that does not make this any less of an issue.

And personally I was experiencing this issue heavily on my M1, which accounts for the 21.3TB it has written since launch, but now I’m averaging writes of 20-30GB per day while maintaining the same workflow and using the laptop for the same things, as a result of multiple fixes and tuning which I believe all contributed to fixing my write problem. So this is in my opinion most definitely caused by software, and from what i’ve observed specifically by various parts of Apple’s software, including Big Sur and what came with it eg Rosetta 2.

Whether they’re planning to fix it or it’s working as intended only Apple knows. But I know i’d rather have SSD writes on the level of my previous Intel Macbook Pro, rather than these new weekly Terabyte writes that I was experiencing.

Just FYI ... i already independently came to the same conclusion as one of the YouTube vids posted here , sorry can post a link as YouTube is blocked here at the moment (Myanmar) .. i found i could get both my 8GB and 16Gb to behave more like the previous intel devices simply by using only M1 optimized apps ... cut down on rosetta 2 ... i do work with BIG VMs and constantly clone them locally before deploying ... plus the usual office 365, safari ...

so whilst not a "Power" content creator all seems fine ... worst case on my 8GB (used 1% of life in 3 months) still means 25 years to go ... before i reach 100% of stated life ... so why worry ..

 

[Maximara]

Or because
a) Apple doesn't issue press releases or support documents about every single thing MacRumors Forum users are concerned with.
b) This forum is an echo room.
c) No one here actually has proof that the data being reported using the third party apps they have used is actually accurate/that the data is being interpreted correctly, or that the SSDs in Apple computers wouldn't be able to withstand the amount of data written if the data reported turned out to actually be accurate.

The problem, which is the post that is quoted in the link, is these tools are giving two values: Percentage used and Data Units written. If you look at both numbers as I did they go off the rails, Also another video posted here revealed that the Data written in Activity Monitor represents all data written to all drives. So if you have a time machine drive hooked up then the Data written number represents the date written to your SSS AND that written to the Time Machine drive. People have been saying the Data written confirms the numbers that the third party tools are giving them even if they have time machine (one would assume and external drive) active.

 

[Maximara]

Or because
a) Apple doesn't issue press releases or support documents about every single thing MacRumors Forum users are concerned with.
b) This forum is an echo room.
c) No one here actually has proof that the data being reported using the third party apps they have used is actually accurate/that the data is being interpreted correctly, or that the SSDs in Apple computers wouldn't be able to withstand the amount of data written if the data reported turned out to actually be accurate.

The problem, which is the post that is quoted in the link, is these tools are giving two values: Percentage used and Data Units written. You can't say one is valid and throw out the other. Either this tool is valid or it isn't.

 

[TheSynchronizer]

Can you let me know which media outlet has discussed this beyond the usual "MacRumour users are reporting..."?
Do these media outlets base their reporting on the same basis I have already questioned? (i.e. using software that might interpret incorrectly the data)

The Butterfly keyboard issue is not comparable. It was observable with the naked eye (i.e. if your keyboard is broken... It's broken+teardowns could show the design flaw).
This supposedly SSD issue is not. First, it's only observable with third party software that again, might be misinterpreting the data. If the data is accurate, the lifespan calculation could be made on outdated SSD life expectancy models as many people have pointed out before.

Yes there are many articles mentioning ‘Users have reported that’ but there are also many with their own take, many tech channels doing their own tests, and one certain one who’s doing a survey that has 15K responses last time they did an update, to see how many people are affected. You type ‘M1 Mac SSD’ and you find plenty of videos on YouTube, but here’s the survey one specifically

Yes the lifespan calculation (percentage wear) could be inaccurate, but most importantly it can go beyond 100% of wear by design i.e. it was never an accurate metric in the first place, because no one can ever know when an SSD will die as they all have slight differences. It’s indeed just based on some value, whether working correctly or not, that’s irrelevant.

So what’s relevant? Well, would you call Activity Monitor third party software? No? Well when comparing sysmontools results at boot up, and then repeating the report after Activity Monitor reports 1TB+ written, sysmontools shows the same increase in its total SSD writes, that indicates that that metric is accurate.

What else is relevant is that someone’s M1 Macbook with sysmontools reporting 600TB written had an SSD failure. Sure, that’s a sample size of 1 - basically useless, but as I said, it is way too early to determine whether these writes are problematic or not. But give time and all the other users who are completely unaware of this issue, and if more and more M1 SSDs fail then you can see how this could be similar to the butterfly keyboard issue. Yes it’s not hardware, but unfixed software might as well be.

So until Apple fixes this issue themselves, and/or notifies everyone and their mother to check their Activity Monitor to see if they’re affected (unlikely), or at least gives some system notifications that indicate when swap is being abused, this is a developing story. I’ve taken steps to reduce my writes drastically, something I never had to do on previous macs, but I don’t have to be worried whether these writes are problematic or not as I will not be writing anywhere near enough to have an SSD failure while I own this Mac (like putting a silicone cover on a butterfly keyboard 😌).

 

[TheSynchronizer]

Just FYI ... i already independently came to the same conclusion as one of the YouTube vids posted here , sorry can post a link as YouTube is blocked here at the moment (Myanmar) .. i found i could get both my 8GB and 16Gb to behave more like the previous intel devices simply by using only M1 optimized apps ... cut down on rosetta 2 ... i do work with BIG VMs and constantly clone them locally before deploying ... plus the usual office 365, safari ...

so whilst not a "Power" content creator all seems fine ... worst case on my 8GB (used 1% of life in 3 months) still means 25 years to go ... before i reach 100% of stated life ... so why worry ..

I’m not worried either, I do the same things on my laptop as I’ve always done and yet I’ve reduced my writes from 10-12GB an hour to 1-2GB an hour.

Someone’s M1 Macbook failed at 600TBW - mine will reach that value in around 40 years.

Do note that the percentages all seem inaccurate as I believe they had something like 30% wear and yet an SSD failure. Of course that could be an anomaly, and maybe most of these SSDs are designed to survive much higher TBW if the percentage is meant to be accurate.

So i’d rather go by instinct and the actual TBW writes previous SSDs reached before failing. And at the current rate I’m going I have nothing to worry about for around the next 50 years

 

[Maximara]

Yes there are many articles mentioning ‘Users have reported that’ but there are also many with their own take, many tech channels doing their own tests, and one certain one who’s doing a survey that has 15K responses last time they did an update, to see how many people are affected. You type ‘M1 Mac SSD’ and you find plenty of videos on YouTube, but here’s the survey one specifically

Yes the lifespan calculation (percentage wear) could be inaccurate, but most importantly it can go beyond 100% of wear by design i.e. it was never an accurate metric in the first place, because no one can ever know when an SSD will die as they all have slight differences. It’s indeed just based on some value, whether working correctly or not, that’s irrelevant.

So what’s relevant? Well, would you call Activity Monitor third party software? No? Well when comparing sysmontools results at boot up, and then repeating the report after Activity Monitor reports 1TB+ written, sysmontools shows the same increase in its total SSD writes, that indicates that that metric is accurate.

What else is relevant is that someone’s M1 Macbook with sysmontools reporting 600TB written had an SSD failure. Sure, that’s a sample size of 1 - basically useless, but as I said, it is way too early to determine whether these writes are problematic or not. But give time and all the other users who are completely unaware of this issue, and if more and more M1 SSDs fail then you can see how this could be similar to the butterfly keyboard issue. Yes it’s not hardware, but unfixed software might as well be.

So until Apple fixes this issue themselves, and/or notifies everyone and their mother to check their Activity Monitor to see if they’re affected (unlikely), or at least gives some system notifications that indicate when swap is being abused, this is a developing story. I’ve taken steps to reduce my writes drastically, something I never had to do on previous macs, but I don’t have to be worried whether these writes are problematic or not as I will not be writing anywhere near enough to have an SSD failure while I own this Mac (like putting a silicone cover on a butterfly keyboard 😌).

Apple M1 SSD Lifespan Ageing. Do YOU have the problem? shows that Activity Monitor is only useful if you have no other drives connected either physically or via wifi. This is because Activity Monitor shows all data written to all drives. If you have time machine active (which should be an external drive because IMHO not doing so defeats the whole purpose of the feature) then Activity Monitor will show the writes to the SSD AND to the external Time Machine drive. I assume the same is true of iCloud and Time Capsule.

This makes Activity Monitor very limited as a check tool and given nobody I know of pointed out this fact before the Apple M1 SSD Lifespan Ageing. Do YOU have the problem? video came out means people would misread the data and freak out over that even if the 10TB they see is actual 1 TB to the SSD and the rest (9 TB) is to the external Time Machine drive.

I mean it has come out that Windows 10 freaking defrags the SSD it boots from. Say what?! I know the Windows file structure is old but that is insane. As Macworld stated "If you have an SSD defragging will have no benefits and will damage your drive". Were is the outrage from the much larger windows community about this piece of stupid?

 

[Maximara]

I’m not worried either, I do the same things on my laptop as I’ve always done and yet I’ve reduced my writes from 10-12GB an hour to 1-2GB an hour.

Someone’s M1 Macbook failed at 600TBW - mine will reach that value in around 40 years.

I think this is a variation of what Ryan Hileman quoted on twitter:
"I have someone with a failing M1 disk. 4mo old, 2% spare, 10% thresh, 98% used, 600TB write 500TB read, 200h "on", 10,000 "Media and Data Integrity Errors". Machine had an inconsistent glitch in my app. 16/512GB machine, typical RAM use 9GB. Working on RCA, ama."

Well it was claimed this genius was using this as the postgres server for a bank. As I said "I have to ask just how stupid was this bank to run a postgres server on any of what are entry level machines? That like using a crowbar in place of hammer; yes you can use it that way but it is really really dumb.

Something I didn't notice and now do is that 2% spare; that starts out at 100%. So they have blown through 98% of the base capacity and 98% of the spare. Based on other results there is something way wrong with those numbers.

Do note that the percentages all seem inaccurate as I believe they had something like 30% wear and yet an SSD failure. Of course that could be an anomaly, and maybe most of these SSDs are designed to survive much higher TBW if the percentage is meant to be accurate.

Who is saying this? Also "Percentage of the Rated Lifetime Used. This is the opposite of SSD Time Left. 1 means the drive is 100% healthy, while 100 means that 100% of the drive’s lifetime is used up, and the drive can be used as a small doorstop." There is something seriously wonky about a SSD at only 30% going belly up.

So i’d rather go by instinct and the actual TBW writes previous SSDs reached before failing. And at the current rate I’m going I have nothing to worry about for around the next 50 years

Remember all that percentage is how much the drive is warranted for. Also the controller chip could have gone belly up and the SSD itself was fine. I still like to know just what insane thing they were doing to use 600TB in less then four months.

 

[TheSynchronizer]

I think this is a variation of what Ryan Hileman quoted on twitter:
"I have someone with a failing M1 disk. 4mo old, 2% spare, 10% thresh, 98% used, 600TB write 500TB read, 200h "on", 10,000 "Media and Data Integrity Errors". Machine had an inconsistent glitch in my app. 16/512GB machine, typical RAM use 9GB. Working on RCA, ama."

Well it was claimed this genius was using this as the postgres server for a bank. As I said "I have to ask just how stupid was this bank to run a postgres server on any of what are entry level machines? That like using a crowbar in place of hammer; yes you can use it that way but it is really really dumb.

Something I didn't notice and now do is that 2% spare; that starts out at 100%. So they have blown through 98% of the base capacity and 98% of the spare. Based on other results there is something way wrong with those numbers.

Who is saying this? Also "Percentage of the Rated Lifetime Used. This is the opposite of SSD Time Left. 1 means the drive is 100% healthy, while 100 means that 100% of the drive’s lifetime is used up, and the drive can be used as a small doorstop." There is something seriously wonky about a SSD at only 30% going belly up.

Remember all that percentage is how much the drive is warranted for. Also the controller chip could have gone belly up and the SSD itself was fine. I still like to know just what insane thing they were doing to use 600TB in less then four months.

Thank you for this, and I stand corrected as I must admit I couldn’t find the tweet myself and I was just referring to it from my memory. Clearly that case is completely invalid then, and can be disregarded, as that is not something the system was designed to do at all. So we’re currently at 0? reasonable known SSD M1 failures, which is lovely, and hopefully this stays that way for a very long time.

With the percentage I was just referring to the fact that it is really based on some unknown preset number by the manufacturer, for warranty purposes. For example I have a working SSD pulled from a machine that has 125% wear last time I checked. I’ve only replaced it to not have to worry about the inevitable, but it does indeed still function completely normally in every sense, and shows no errors in SMART or any other sense. It was used 24/7 and constantly had many reads and writes for a very long time. The percentages are nothing other than for warranty purposes.

Truly no one knows when their SSD will fail, however it’s very reasonable to think it will at least last for it’s 100% wear as that is the manufacturers claim. It’s very possible that all of these M1 Macs will last for even 2PB of writes to the SSD as that is completely possible, and then none of this is really an issue at all as even with a TB+ written every week you are looking at many years of use. Maybe the problem lies in the uncertainty of the future?

 

[Fishrrman]

Can someone confirm if this terminal command works on the m1 series:
sysctl vm.swapusage

Using this command changes nothing and will hurt nothing.
All it does is render a report on VM swap.

If VM is off, report should be:
vm.swapusage: total = 0.00M used = 0.00M free = 0.00M

That's what I see on my INTEL-based 2018 Mini with VM disk swapping disabled.

I was wondering if this reporting command works on the m1 side of things...

 

[vcsyc]

Can someone confirm if this terminal command works on the m1 series:
sysctl vm.swapusage

Using this command changes nothing and will hurt nothing.
All it does is render a report on VM swap.

If VM is off, report should be:
vm.swapusage: total = 0.00M used = 0.00M free = 0.00M

That's what I see on my INTEL-based 2018 Mini with VM disk swapping disabled.

I was wondering if this reporting command works on the m1 side of things...

vm.swapusage: total = 3072.00M used = 1654.50M free = 1417.50M (encrypted)

It's interesting that my activity monitor shows only used (1,6gb) amount as swap.

 

[pistonpilot]

You bring up some very good points.

A lot of people seem to be jumping on the blame Apple bandwagon with this topic, when we don't have any corroborative facts.

I have a fantastic idea, let's blame IBM. We wouldn't want to blame Apple, would we?

 

[Apple_Robert]

I have a fantastic idea, let's blame IBM. We wouldn't want to blame Apple, would we?

At this point, we don't know if Apple is to blame. If facts come forth that show Apple is to blame, I would have no problem adding my voice to those blaming Apple. However, I am not going to blindly blame Apple for something I don't know to be true. Doing that would be illogical and stupid, in my opinion.