Readyboost was never meant to increase RAM performance, it was to cache reads and writes to the "disk". (SSD and older type)
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
ssd swap - high usage of Terabytes Written
- Thread starter Forti
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
Nope.. They just use Swap like it if it was normal RAM.
Readyboost was there to cache data from slow HDDs to faster flash drives.
I meant an out of warranty repair at the user's expense, and I agree that a chargeable SSD failure would realistically mean scrapping the machine. However, it could well be that if numbers of M1 computers exhibit early SSD failures then Apple might offer a repair programme as they did with MBP GPUs, keyboards and I believe other issues.
The UK warranty Apple offer is, for better or worse, entirely normal.
Heavy users are self-defining, but I don’t suggest using Safari or any other browser of itself constitutes heavy usage. Multiple browser tabs have been mentioned in this thread, numbers of tabs I find hard to comprehend. We know tabs need memory and we know M1 computers maintain zippy performance by paging RAM to SSD very freely, so of course SSD writes will be high. But enough to impact the computer’s life expectancy? That’s the question.
'Entry level' performance was redefined up - by Apple!
We run more simulataneous apps/tabs on M1 now because we can. This leads to more memory usage, which leads to more swap, which leads to lower SSD longevity. This is only a problem now because the machines is so capable compared to before.
A 16GB/1TB Macbook Air M1 handles multiple simultaneous apps + *hundreds* of browser tabs + spotlight/photos indexing + time machine, while still remaining usable. Unheard of on a high-end Macbook Pro (intel). Extend SSD life by years through spending a few hundred dollars on more memory/storage if that's your use case.
Well, OK, so instead of 250 years of expected SSD life it seems that I will have to settle with less than a third of that.
Even a fifth will do.
Say, 50 years?
50 years is good to me. Even 15 would be.
We don't know for sure whether the cells in Apple's newest SSDs support 1000-1200 cycles (as you mention) instead of 3000 cycles as others. They could even be rated higher. I would rather be inclined to think more along 3000 than 1200, but again, it's a lot of life in my case.
And moreover, even if they are rated at 1000 it doesn't mean they will fail at 1000 cycles. Or even at double that figure. They can last less, of course. We are talking averages here. And there will be several warnings if the SSD is nearing doom.
Again, my point is that in my particular case (not abusing Safari, not using any non-native app) the SSD will be good for the whole life of the machine. Any other thing will fail sooner. Or I will declare the machine obsolete before the SSD fails.
But even after revisiting the numbers, I am more inclined to predict more than 50 years than 15.
Thanks for your comments on write amplification and wear leveling, in any case. Yes, I am used to keep a good amount of free space in every disk, too.
I am learning a lot about SSDs in this thread but, frankly, much more than I ever wanted to.
I have been following this thread daily for so long now. To be frank, it is starting to get on my nerves to see Apple being silent, and so many of you guys working hard to make adjustments, finding out how to reduce the numbers, etc. Do you see my point? All the anxiety of people who own a MBA M1, or about to buy one....the silence of apple....is just so frustrating. It feels like we are not getting anywhere, even with all the valuable debates we have here.
I dont know, after legitimately worrying about this for the longest time, im just starting not to care anymore. whatever happens, happens. I have wasted many many days worrying about this.
I dont know, after legitimately worrying about this for the longest time, im just starting not to care anymore. whatever happens, happens. I have wasted many many days worrying about this.
The thing is: If Apple acknowledges this as an issue, they "have to" fix it or people might have the right to refund their machines. Or worse: Class action law suit in the US. Because people will then think Apple knowingly sold them broken devices. Also: People who are unaware about this issue, might start to dig deeper. Apple doesn't want that.
Apple will not acknowledge anything unless it's inevitable. They might even silently fix it and not even list it in the change notes.
It's their best choice for them to be silent. They always did it that way.
Apple will not acknowledge anything unless it's inevitable. They might even silently fix it and not even list it in the change notes.
It's their best choice for them to be silent. They always did it that way.
As I said before it may be more problematic than that. Remember that this all started through some third part tools and three are been statements that at that time the tools were not coded for the M1 or Apple's controller. Also people have pointed to Chrome and or Adobe being suspects and Apple has no control over what Alphabet or Adobe code their programs.
Right now the only example of an M1 SSD failing we have is a bank using one for a postgres server which is totally insane. 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." You can't even justify that nonsense with "testing" because again, these are entry level machines.
More over the tools give results that are all over the place and in many respects when you use basic math you get crazy values. Heck, Intel went out of its way in Technical Advisory – TA 340062-001US to explain their Optane SSD DC P4800X/P4801X would cause Percentage Used to behave non-linearly early on - implying normally that Percentage Used is normally linear.
An older version (2019-2020) of the 2021 webpage (posted a long time ago and called "old") said this:
In regards to the attribute named “Percentage Lifetime Used” (sometimes referred to as “Percent Lifetime Remaining”), this is simply a metric for how much wear life is left on your SSD. A solid state drive, like any flash memory-based storage device, has a limited amount of data which can be written to the memory blocks before they start to lose their reliability, and eventually go into read only mode. Your Crucial SSD will keep track of this life with SMART attribute 173, “Average Block Erase Count.” The Lifetime Used is a reflection of the block erase count in terms of a percentage. For example if your drive is rated for 3000 block erases and you have a total of 100, your Percentage Lifetime Used would be 100/3000, or 3-4%. For percent lifetime remaining we would simply take (3000-100)/3000 = 96-97%. These attributes are not a full picture of the health of a drive, but an expectation of how much usable life is left.
Everything points to Percentage Used normally functioning in a linear manner with the only counter argument so far a wikipedia article. Heck, even wikipedia doesn't consider its own articles "reliable" as sources so why should we be any different?
Last edited:
Different use-cases, to be sure. As you point out, the power between them has been leveled to an almost imperceptible degree with M1 (which is one reason it's so impressive, of course). This also means all of them are affected by this issue (although storage capacity/RAM have more to do with how badly than the model itself, which I'll come back to).
My main point in the post you quoted, though, was in how this relates to my feeling that increasingly the idea has been woven into this thread, and the wider narrative surrounding this issue, that it is heavy use which is responsible, when in fact the practices producing high TBW seem to be just about the most everyday use imaginable - browsing the web with (shock, horror) lots of browser tabs open and having lots of apps open in the background.
This is not 'heavy use' would normally be defined. This is how a vast majority of ordinary, everyday users have their PCs running. As I mentioned previously, this is how the most computer illiterate people I know use their laptops/i-devices and judging by how I see friends and family in general use their devices I'd say it's completely normal behaviour for the average user. As far as I'm aware, there has never been any idea that anyone was trashing their SSD by doing so.
So, coming back to the implied tradition in the hierarchy of Mac computers; with this particular issue, it would likely be the low end machine buyer/user who would stand to be the worst affected - e.g., less computing knowledge (so probably not aware of the issue/forums and taking steps to mitigate), using their device in an everyday way (as above), and with lower-spec machines in the model range (less RAM, lower capacity SSD), which would speed the demise of the SSD if they did have this issue.
Except that this is also occurring with just one program running - e.g., one of the worst culprits is Adobe Lightroom CC, which has been Silicon native for a while and still having issues...
My thinking is if it were a non-issue (if the stats were way out, or if Apple had a very highly specced/heretofore unseen SSD lifespan) it would have been in Apple's interest to just say so and put the issue to bed. In my mind, their silence can only mean the opposite - there's a real problem, chance of failures, but they stay silent because it means they will be able to wriggle out of as many as possible without losing a lot of money.
UK's VAT doesn't change the fact the first batch of M1s were entry level machines.
As I have previously pointed out there are other practical reasons for Apple keeping silent.
First, the tools that caused all this in the first place were third party and it had beens stated the versions first used had incorrect specs as baselines.
Second, only some people have reported a problem with the main culprits being Chrome, Adobe, and other third party stuff or people trying to entry level machines as if they were high end ones. The only Apple program that has been clearly identified as a possible source is Safari and I will concede that that has been a flaky mess (on my iMac 2023) for the last six months with certain sites going all wonky with the weird black triangle problem under Catalina and then people reported similar problems on later Macs with Big Sur and other programs as well
Finally, even activity monitor will produce weird numbers when external drives are plugged in (even if not mounted)
Apple cannot go off and make statements until they nail down the issue.
Can I prevent somehow big writtings caused by Paralles VM? After started a WIN10 in 5 minutes I've got like 10gb of writes caused only by VM
Last edited:
Depends on how Paralles handles its VM. If you can put it on an external drive that is certainly better then having it on the internal SSD.
That's pretty normal. VMs are write intensive, because they write to the virtual disk which is then mapped to APFS blocks and then to SSD blocks. So it naturally has a high write amplification. Especially with time machine's copy on write snapshots and snapshot rotation.
If the VM has not enough RAM for what it does, it has to page out (inside the VM). Then you have swap inside the VM which is even worse than swap on the host.
One benefit we have in the U.K., as far as the warranty is concerned, is the Limitations Act, a part of the Consumer Rights Act.
While it can be a pain in the butt sometimes to actually enforce it, the act states that a product must be fit for purpose and last for a reasonable amount of time. That law overrides any company warranty and is effective for a period of 6 years. Though it is your burden to prove the fault existed at the time of purchase.
Essentially then, should the SSD fail within, let’s say 4 years just for the heck of it. You would have little problem in your argument that such a component should reasonably last far longer than 4 years and the generally accepted lifespan of any computer, should be more than 4 years.
So, in warranty or not, you have a chance of repair or replacement under the Limitations Act - if you could be arsed doing it. I’ve done it once, for a very expensive television, which went completely tits up after 3 years. On that occasion I was pretty lucky I think, the manufacturer didn’t argue at all against the CRA, so that was good.
thanks to both of you
and thanks for everyone involved in this tread. Due to that daily reading of what you guys saying, my english improved so much. <3
Last edited:
I definitely agree with this. I’ve always run Parallels VM’s, even on my Intel Macs, on an external SSD.
Playing around with the beta Parallels for M1, I’ve got it running from a 1TB Samsung T7 and the performance is excellent.
I've come to the conclusion that the excessive swapping/disk usage is not "a bug"... it's "a feature".
In other words, it's something that Apple's engineers have intentionally designed to "be there". Simply stated this is how the m-series Macs work. Not by "mistake", but "by design".
HOWEVER... it's not something they want to talk about.
Has Apple addressed this issue directly yet?
In any way?
The amount of "investigation" in this thread, trying this, trying that to reduce the amount of data written, is probably pointless.
Again, because the way we see things working, is "the way they are designed to work".
And once more, I'll offer my solution:
TURN OFF VM disk swapping using terminal,
and
TURN OFF compressed memory using terminal.
See what that does for you over the course of 2-3 days, then report back with the results.
In other words, it's something that Apple's engineers have intentionally designed to "be there". Simply stated this is how the m-series Macs work. Not by "mistake", but "by design".
HOWEVER... it's not something they want to talk about.
Has Apple addressed this issue directly yet?
In any way?
The amount of "investigation" in this thread, trying this, trying that to reduce the amount of data written, is probably pointless.
Again, because the way we see things working, is "the way they are designed to work".
And once more, I'll offer my solution:
TURN OFF VM disk swapping using terminal,
and
TURN OFF compressed memory using terminal.
See what that does for you over the course of 2-3 days, then report back with the results.
May I ask you - how I can make it possible to run a Parallels on external drive?
Since a couple of weeks I learnt more about my computer, especially how everything works then in my entire life, trust me