ssd swap - high usage of Terabytes Written

[Maximara]

My experience was similar to jdb8167's except mine was a 28TB single event. These events may be happening much more often than we know because they will only be recognised as single events by people who track and plot their usage regularly.

I guess Santa wanted to give you some early coal (given the spike looks close to December 22).

Ok. Seriously, the fact it only happened that one time over a year of usage shows whatever this bug is, it requires thing to line up "just so" (not at ridiculous as the insurance policy in Fool Coverage but still very unlikely). Based on what I know about programming this is going to make it difficult if not impossible to find out what is causing it.

Though as writing this I just had a thought. How would the Mac or any tools that monitor SSD writes record wear leveling. Could this weird little spike be due to how Apple handles wear leveling?

People who look for the first time and report "xx TBW in yy months" have no way of knowing if that has been steady or a single event.

Yep and if it is the byproduct of some weird wear leveling algorithm it could be in the 'working as intended' category.

My hunch is that some of the exremely high usages reported in this thread have been caused by one or more events like jdb8167 and I have experienced. I don't think any of them have shown data supporting a steady increase to their very high usage.

That is what I have noticed. Still have to wonder what the TBW is actually set to given what the total TBW using Percentage Used part of the API back calculate to (totally off the wall insane in many cases)

 

[jdb8167]

I guess Santa wanted to give you some early coal (given the spike looks close to December 22).

Ok. Seriously, the fact it only happened that one time over a year of usage shows whatever this bug is, it requires thing to line up "just so" (not at ridiculous as the insurance policy in Fool Coverage but still very unlikely). Based on what I know about programming this is going to make it difficult if not impossible to find out what is causing it.

Though as writing this I just had a thought. How would the Mac or any tools that monitor SSD writes record wear leveling. Could this weird little spike be due to how Apple handles wear leveling?


Yep and if it is the byproduct of some weird wear leveling algorithm it could be in the 'working as intended' category.

That is what I have noticed. Still have to wonder what the TBW is actually set to given what the total TBW using Percentage Used part of the API back calculate to (totally off the wall insane in many cases)

This hasn't happened again so it isn't something new in my usage though I am working onsite now mostly on battery. That hasn't seemed to affect the SSD writes in anyway and this spike happened before I started the new contract.

Your question on wear leveling is interesting. If it is related, then it is clearly still a bug since I have a 1 TB drive. It is hard to imagine any wear leveling algorithm that would need 15 times the total drive size to balance out writes. I could see 2 TB but not 15 TB.

I do agree that it would be difficult to reproduce this bug but there are other ways to discover problems that don't necessarily require reproducing the bug—though that does usually make it easier to track down.

 

[Thistle41]

From what I found on the web this ssd should last for ~~150 TBW. After 8 days I have 0.9 TBW

-What’s the problem? You should compare and try: a chess gui with the Stockfish chess engine and all 3 to 7-piece syzygy endgame tablebases!!

So, let's say worst case scenario you get through 1TBW/week, 52TBW/yr the 150TBW life of the SSD gives you < 3 years. Nothing to worry about? My drive is at 83.3 after 18 months and being careful with auto tab discard etc etc as at the top of this thread. So my experience aligns with the figures you're quoting.

I do wish Apple would help with this but I note in another thread about hub connection issues any OS update seems to ignore known problems and I think this is another issue that is being ignored.

 

[Thistle41]

I guess Santa wanted to give you some early coal (given the spike looks close to December 22).

Ok. Seriously, the fact it only happened that one time over a year of usage shows whatever this bug is, it requires thing to line up "just so" (not at ridiculous as the insurance policy in Fool Coverage but still very unlikely). Based on what I know about programming this is going to make it difficult if not impossible to find out what is causing it.

Though as writing this I just had a thought. How would the Mac or any tools that monitor SSD writes record wear leveling. Could this weird little spike be due to how Apple handles wear leveling?


Yep and if it is the byproduct of some weird wear leveling algorithm it could be in the 'working as intended' category.

That is what I have noticed. Still have to wonder what the TBW is actually set to given what the total TBW using Percentage Used part of the API back calculate to (totally off the wall insane in many cases)

Agree, my 83TBW gives me 5% percentage used on a 256GB/8GB half filled SSD. So the figure quoted elsewhere 150TBW life of the SSD simply doesn't add up.

 

[evertjr]

From what I found on the web this ssd should last for ~~150 TBW. After 8 days I have 0.9 TBW

-What’s the problem? You should compare and try: a chess gui with the Stockfish chess engine and all 3 to 7-piece syzygy endgame tablebases!!

The M1 ssd certainly lasts a lot more than 150TBW, in this thread there's multiple users with 300TBW+ and even one with petabyte written. Either those SSDs are insanely over provisioned or macOS is incorrectly reporting some stuff as data written making those numbers useless. I guess we'll only know when ssds starts falling (I think none so far?) or Apple announce some kind of repair program, but I feel like its working as intended, you can't compare the M1 SSD with anything else because they are exclusively made for Apple, there's no similar chip.

 

[Maximara]

From what I found on the web this ssd should last for ~~150 TBW. After 8 days I have 0.9 TBW

-What’s the problem? You should compare and try: a chess gui with the Stockfish chess engine and all 3 to 7-piece syzygy endgame tablebases!!

Agree, my 83TBW gives me 5% percentage used on a 256GB/8GB half filled SSD. So the figure quoted elsewhere 150TBW life of the SSD simply doesn't add up.

That ~~150 TBW is what the drive is warranted for not how long it will actually last as demonstrated by The SSD Endurance Experiment: They’re all dead article back in 2015. For example, the Samsung 840 Pro (73 TBW) last to at least 2.5 PB or 33 times what it was warranted for.

The M1 ssd certainly lasts a lot more than 150TBW, in this thread there's multiple users with 300TBW+ and even one with petabyte written. Either those SSDs are insanely over provisioned or macOS is incorrectly reporting some stuff as data written making those numbers useless. I guess we'll only know when ssds starts falling (I think none so far?) or Apple announce some kind of repair program, but I feel like its working as intended, you can't compare the M1 SSD with anything else because they are exclusively made for Apple, there's no similar chip.

Well the tools are using Apple's own APIs and they are open source so if you know how to program you can check if the programming in them makes sense. Though, I have been pointing out from the get go that the TBW that the percentage used produces doesn't make any degree of sense.

 

[Thistle41]

The M1 ssd certainly lasts a lot more than 150TBW, in this thread there's multiple users with 300TBW+ and even one with petabyte written. Either those SSDs are insanely over provisioned or macOS is incorrectly reporting some stuff as data written making those numbers useless. I guess we'll only know when ssds starts falling (I think none so far?) or Apple announce some kind of repair program, but I feel like its working as intended, you can't compare the M1 SSD with anything else because they are exclusively made for Apple, there's no similar chip.

OK agree that a) we don't know the actual SSD spec, and b) the reporting could be somehow skewed. However, if there are failures of the SSD in the average M1 user population, how will **we** know it is the SSD at fault? The user will experience random presumably crashes or some cryptic error message so will not know why. It won't be mentioned in this forum unless by chance one of us here experience it and know what's happening.

My M1 Mac Mini is also monitored and has about 1/2 the TBW and does not progress at anything like the rate the MBA does even though it's the same spec (AFAIK but are the internals actually identical for 256/8 spec?).

 

[Maximara]

Agree, my 83TBW gives me 5% percentage used on a 256GB/8GB half filled SSD. So the figure quoted elsewhere 150TBW life of the SSD simply doesn't add up.

The quick and dirty TB*100/percentage = TBW gives 1,666 TBW Which isn't too off the wall compared to some of the other values I have calculated.

 

[stigman]

Tha's my usage on MBP 256GB/16 since 29.10.2021. I've implemented most of tips from first page of this thread and it seems like it helps out a bit. I'm not heavy user though, but tab abuser for sure with 30+ tabs in safari

 

[altaic]

Tha's my usage on MBP 256GB/16 since 29.10.2021. I've implemented most of tips from first page of this thread and it seems like it helps out a bit. I'm not heavy user though, but tab abuser for sure with 30+ tabs in safari

Huh, my 16” MBP 64GB/4TB from 2021-10-26 looks a little worse at 6.31 TB, though the spare/used numbers are the same as yours (i.e. basically new). I do kill off leaky processes, though, so I usually don’t have any swap in use.

I wasn’t sure how many tabs I have open currently, so I just wrote a little AppleScript:

Guess it’s time to do some spring cleaning 😬

 

[gank41]

So, let's say worst case scenario you get through 1TBW/week, 52TBW/yr the 150TBW life of the SSD gives you < 3 years.

I've got an M1 MBP 16GB/2TB from November 2020, showing 124 TB written over 4410 'Power On' Hours (184 Days). Still running like a champ.

 

[Maximara]

So, let's say worst case scenario you get through 1TBW/week, 52TBW/yr the 150TBW life of the SSD gives you < 3 years. Nothing to worry about? My drive is at 83.3 after 18 months and being careful with auto tab discard etc etc as at the top of this thread. So my experience aligns with the figures you're quoting.

I do wish Apple would help with this but I note in another thread about hub connection issues any OS update seems to ignore known problems and I think this is another issue that is being ignored.

Again that value (150TBW) is what the SSD is warranted for. It doesn't really tell you the actually life as demonstrated by the 2015 test. For example the Samsung’s 840 Pro 256GB is rated at 73 TBW but in the actual test it lasted until 2.5 PB or slightly over 33 times what it was warranted for.

Look at the percentage used — that rather than the raw TB written tells you more about the lifespan.

 

[Maximara]

I've got an M1 MBP 16GB/2TB from November 2020, showing 124 TB written over 4410 'Power On' Hours (184 Days). Still running like a champ.

View attachment 2009689

2% over 184 Days = 9,200 days for 100% or 25 years of life for the SSD

 

[ambient_light]

In the meantime ... 12.5 Beta1 swapping is visibly more efficient than in previous Monterey versions. Fingers crossed it could return to Big Sur level after 11.4. Perhaps they finally merged back some fixes

 

[Pedro1223]

Hello, here is my latest update. I have been using the computer since March 30 2021, and I have only this TBW used. My work is as a researcher (NLP) so it does not see much highly demanding process but, I do have 60+ tabs in Opera GX with tabsuspender, I try to use mostly native apps, when I need teams I use the web version. Opera GX is amazing! Also I have noticed that the app using most ram is RStudio but since it is native it does not write to much to the SSD but when it is finished it kind of deletes it.

 

[Maximara]

Hello, here is my latest update. I have been using the computer since March 30 2021, and I have only this TBW used. My work is as a researcher (NLP) so it does not see much highly demanding process but, I do have 60+ tabs in Opera GX with tabsuspender, I try to use mostly native apps, when I need teams I use the web version. Opera GX is amazing! Also I have noticed that the app using most ram is RStudio but since it is native it does not write to much to the SSD but when it is finished it kind of deletes it.

Using the percentage I get about 16 years. (1% in two months)

 

[evertjr]

Looks like 12.5 introduced massive improvements to memory management. On my 8gb M1 Air the memory pressure stays green even with a large amount of tabs, 3 vscode instances and node server running and im shocked how little its writing to the ssd.

 

[pshufd]

Looks like 12.5 introduced massive improvements to memory management. On my 8gb M1 Air the memory pressure stays green even with a large amount of tabs, 3 vscode instances and node server running and im shocked how little its writing to the ssd.

I'm giving some consideration to upgrading my M1 mini to Monterey. I will do some testing on my 2021 MacBook Pro to verify that memory management has improved and will then go ahead.

 

[thadoggfather]

Looks like 12.5 introduced massive improvements to memory management. On my 8gb M1 Air the memory pressure stays green even with a large amount of tabs, 3 vscode instances and node server running and im shocked how little its writing to the ssd.

any downsides to 12.5 so far?

still on 12.3.1 on base M1 Air

 

[pshufd]

any downsides to 12.5 so far?

still on 12.3.1 on base M1 Air

See the Monterey 12.5 thread in the news section. One user on Reddit had their 2021 MacBook Pro bricked - he's going to bring it to an authorized reseller. One person in the MacRumors thread says the installation gets stuck about two-thirds of the way through. Vast majority have no problems with it.

 

[thadoggfather]

that always happens with someone with some update. that alone wouldn't necessarily worry me

 

[pshufd]

that always happens with someone with some update. that alone wouldn't necessarily worry me

It didn't worry me. I updated after I read it. I watched it like a hawk though.

I can just drive to my local Apple Store (I have three of them), if a device bricks. It has only happened to me twice in 25 years and they took care of it in ten minutes both times.

 

[939248]

13.0b MBA M1 8/256 (11/20)

 

[ambient_light]

I'm giving some consideration to upgrading my M1 mini to Monterey. I will do some testing on my 2021 MacBook Pro to verify that memory management has improved and will then go ahead.

Unfortunately from my observations not only it didn't improve in comparison to late Big Sur releases, but rather deteriorated further. Now on MBA M1 16Gb / 1TB I constantly observe written volume > 50% of read, primarily from kernel_task / paging, even with low pressure or swap size, which is unjustifiable. The paging bug is still alive and well in Monterey 12.5.

 

[osplo]

Looks like 12.5 introduced massive improvements to memory management. On my 8gb M1 Air the memory pressure stays green even with a large amount of tabs, 3 vscode instances and node server running and im shocked how little its writing to the ssd.

That's great. At this point I suspect that the few cases that still have issues with this have something special with their setup. (I mean, Electron apps, bad apps, or a faulty SSD driver or anything).

My M1 Mac has written less than 10 TB in year and a half. It will last far more than me.