ssd swap - high usage of Terabytes Written

[Maximara]

Nice try but Surface Pro X has storage controller on ARM SoC and has replaceable SSD.

And that Surface Pro X runs Windows for ARM at nearly half the speed the M1 does...though virtualization - Apple M1 runs Windows nearly 2 times faster than Microsoft’s hardware

Having replaceable SSDs or RAM or both clearly causes a slow down as I seriously doubt the M1 on its own is double the speed of whatever Microsoft is using.

More over as Kung gu pointed out:

"You can actually buy a few 2230 M.2 SSDs (the size of M.2 that MS uses) from KIOXIA and Western Digital as a non-commerical customer, but they were clever enough to build the Surface that standard 2280 M.2 SSDs don't fit inside. If they would, you could just go and buy inexpensive off the shelf SSDs. I was about to applaud Microsoft for not soldering the SSD and make it replaceable, but this way it doesn't matter."

That replaceable SSD is actually smoke a mirrors as if the case has been purposely designed so you cannot replace them with off the self SSD then it might as well be soldered because odds are that version of the SSD is so expensive that booting off an external SSD (like with the M1 Macs) is more economical.

 

[daverdfw]

another update, 43hrs uptime. 151GB written by kernel_task. so not sure if it was running Onyx or 11.3b4 but things are MUCH improved.

 

[Mc0]

another update, 43hrs uptime. 151GB written by kernel_task. so not sure if it was running Onyx or 11.3b4 but things are MUCH improved.

May I know your workflow? I'm a casual user so it surprises me to see this much data. I use my mbp to code web apps so I only hit around 30 - 40GB per day. I think my max is 60GB when I do a lot of sync tasks.

 

[Fishrrman]

I don't have an m1 Mac.
But...
... my guess is that the out-of-control writes to the drive has something to do with how virtual memory works on the m1, and also may have something to do with the "changed" way that memory is hard-wired to the CPU (direct memory access??). Think of it as "VM swapping gone crazy".

My thinking -- which I've already posted in this thread more than once -- is that this can be addressed by TURNING OFF virtual memory disk swapping. (if that's possible on the m1 Macs, again, I don't have one with which to try this myself)

Once again, I request that someone who has an m1 just try this for a day or so, and see if that changes anything...

 

[gregpod9]

Would anyone of you buy an Apple Certified refurbished MBA M1 or MBP M1 even with this SSD wear issue?

 

[daverdfw]

just knowledge worker stuff, I work in IT pre-sales. So lots of office, web browsing, terminal windows. No coding or content creation.

 

[IceStormNG]

Would anyone of you buy an Apple Certified refurbished MBA M1 or MBP M1 even with this SSD wear issue?

Nope. But not because of the SSD.

And that Surface Pro X runs Windows for ARM at nearly half the speed the M1 does...though virtualization - Apple M1 runs Windows nearly 2 times faster than Microsoft’s hardware

Having replaceable SSDs or RAM or both clearly causes a slow down as I seriously doubt the M1 on its own is double the speed of whatever Microsoft is using.

And what exactly has the SSD speed to do with regular "speed"? Which comes mostly from the fact that the ARM SoC in the Surface pro is way weaker than the M1. The M1 Mac would be probably equally fast as it is right now if they used a NVMe Blade instead of soldering it.

Maybe swapping and therefore writing the hell out of the SSD would be slower then ahaha

 

[JMacHack]

That replaceable SSD is actually smoke a mirrors as if the case has been purposely designed so you cannot replace them with off the self SSD then it might as well be soldered because odds are that version of the SSD is so expensive that booting off an external SSD (like with the M1 Macs) is more economical.

I’m going to play devils advocate here, a replaceable, though non standard ssd is a step in the right direction. If you think of it like restoring an old car, you often have to cannibalize another old car to do it. Even if you have to cannibalize another Surface to fix one, it’s still fixable.

 

[AndyMacAndMic]

That replaceable SSD is actually smoke a mirrors as if the case has been purposely designed so you cannot replace them with off the self SSD then it might as well be soldered because odds are that version of the SSD is so expensive that booting off an external SSD (like with the M1 Macs) is more economical.

But if you want to repair the device is it more economical to replace the whole motherboard in the M1 or replacing only an SSD? You can't replace the motherboard of the M1 with an off the shelf motherboard either. I still think that a new proprietary 'expensive' SSD you are talking about is way more economical than replacing a proprietary M1 motherboard.

Basically your 'smoke and mirrors' argument is much less smoke and much less mirrors than the M1 motherboard's 'smoke and mirrors'. Because, let's be honest, if your internal M1 SSD fails are you really going to boot from an external drive permanently, or are you going to seek for repairs?

 

[Mistborn15]

Would anyone of you buy an Apple Certified refurbished MBA M1 or MBP M1 even with this SSD wear issue?

In hindsight, I should have waited since I am very much in Adobe ecosystem. But I've accepted that I won't get any resale out of this M1 Air. If Adobe release Apple silicon version of their apps by end of this year and the laptop works till 2025, it would have paid for itself

 

[k-hawinkler]

It's all about RAM managing as far I know. Users with 8GB ram have much higher TBW than those with 16GB.

That and bugs/features in Big Sur, no?

 

[Spudlicious]

Would anyone of you buy an Apple Certified refurbished MBA M1 or MBP M1 even with this SSD wear issue?

I wouldn't because Apple UK's refurb prices are brutal. Current example:

Refurbished 13.3-inch MacBook Air Apple M1 Chip with 8‑Core CPU and 8‑Core GPU - Space Grey​

• £1,059.00

That's the 512GB model I bought for £1149.97 from Amazon back in January, and not even in the gentleman's Silver choice of MacBook colour. It's used, it's second-hand, and no way would I pay so close to the new price for anything someone else had rejected. If I was such a mug as to buy an Apple refurb M1 the first thing I would do on receipt, and even though I've expressed cynicism in this thread about the whole SSD write shock/horror/calamity, would be to install the command line tools and paste smartctl -a disk0 into Terminal. Who would accept a used car without knowing the mileage?

 

[mi7chy]

Would anyone of you buy an Apple Certified refurbished MBA M1 or MBP M1 even with this SSD wear issue?

Make sure it has a transferable warranty and avoid 8GB DRAM. There's a reason why 8GB is dumped by previous owners and is always on sale new.

 

[Maximara]

I don't have an m1 Mac.
But...
... my guess is that the out-of-control writes to the drive has something to do with how virtual memory works on the m1, and also may have something to do with the "changed" way that memory is hard-wired to the CPU (direct memory access??). Think of it as "VM swapping gone crazy".

My thinking -- which I've already posted in this thread more than once -- is that this can be addressed by TURNING OFF virtual memory disk swapping. (if that's possible on the m1 Macs, again, I don't have one with which to try this myself)

Once again, I request that someone who has an m1 just try this for a day or so, and see if that changes anything...

People have reported similar (but not as bad) behavior on Intel Macs so whatever this is it is not unique to the M1.

If we go back to Lonehorn, which everybody and his brother seem to be quoting, we have 1% in about 2 months - that is 200 months (16.6 years) for 100%.

Dan Moren's numbers are even better as 1% in 3 months is 300 months or 25 years.
His 2017 iMac's 14% in 42 months (3.5 years) product the same life span - 300 months (42*100/14).

If these numbers real these SSDs will last over a freaking decade. Tell me again why everyone is freaking out?! Is it because most of the people reporting these numbers are NOT working through the numbers and realizing 'oh if I used y% in x months then I have a minimum of 100 * x/y' months before I hit 100%'?

 

[majormike]

So the consensus is that the high memory swap comes from unoptimized Rosetta 2 Apps? Which means Rosetta 2 is a dead end and unless any app is available natively, it will lead to long term durability issues?

 

[thadoggfather]

So the consensus is that the high memory swap comes from unoptimized Rosetta 2 Apps? Which means Rosetta 2 is a dead end and unless any app is available natively, it will lead to long term durability issues?

I haven't kept up fervently with this 55 page thread because I don't seem impacted (funnily enough as someone using only AS apps, only had one and found the AS version right away when I got the machine in December)

BUT

that doesn't seem the consensus at all. Seems its not isolatable: some with tons of tabs, some with video editing and such, some with casual use, some with TM backups, etc.


--

393 W / 918 R (@ la kernel_task)
35 days uptime
macOS 11.2.1
M1 Air base

 

[Tev11]

 

[majormike]

That Ryan Hileman Tweet from February 28th has previously been reproduced in this thread. Please note that Hileman has not mentioned the matter since then, and neither has the guy he quotes. A small puff of smoke does not mean there's a big fire.

This is really bad though. My M1 MacBook Pro will arrive in about 5 days and it looks like I'll be sending it back and buy an Intel Laptop is that's what will happen with these machines.

My sister and her boyfriend have a 13" Macbook Pro from 2014 and it is still going strong while they actually have the ability to swap the SSD and with M1 you don't.

I don't want to, given the worst case, send my M1 to Apple after 3 years and pay 700 to replace the complete logic board.

 

[majormike]

No is is NOT "pure greed" As I said in another thread:

To everyone moaning and groaning about Apple using soldered RAM and SSDs read The scourge of fully soldered and non-upgradeable laptops.

"First and possibly foremost is manufacturing efficiency, which includes both quality control and cost reduction. Every additional removable piece, especially including a SODIMM slot, introduces more cost and another potential fail point. Plus, an actual RAM socket requires an actual human being be there to plug a RAM chip into every laptop that goes down the assembly line, further adding to cost.

More pertinent, though, is the fact soldered RAM can be placed just about wherever engineers decide. With proper research and development, this can lead to streamlined mainboard design as well as increased thermal efficiency. It also means there's no need to include a bulky connector or an access door, and all told, these benefits combine to let laptop designers shave millimeters off case thickness"

I'm sorry my friend but that's complete ********, they could add a swappable SSD and it would not increase the laptops thickness, it is, for one reason or another, "greed".

It is made for 2 reasons:
1: to completely control the investment of a user to pay for storage (don't allow them to install better and cheaper third party storage) and 2: to make sure they have to buy a new laptop earlier than before or make them pay to "repair" it by replacing an entire logic board (it'd be at least acceptable if they actually had real techs who replace the worn ssd).

 

[k-hawinkler]

So the consensus is that the high memory swap comes from unoptimized Rosetta 2 Apps? Which means Rosetta 2 is a dead end and unless any app is available natively, it will lead to long term durability issues?

Nope. I don't see a consensus.

 

[Maximara]

I'm sorry my friend but that's complete ********, they could add a swappable SSD and it would not increase the laptops thickness, it is, for one reason or another, "greed".

It is made for 2 reasons:
1: to completely control the investment of a user to pay for storage (don't allow them to install better and cheaper third party storage) and 2: to make sure they have to buy a new laptop earlier than before or make them pay to "repair" it by replacing an entire logic board (it'd be at least acceptable if they actually had real techs who replace the worn ssd).

And an external SSD knocks that claim in the head.

 

[Maximara]

This is really bad though. My M1 MacBook Pro will arrive in about 5 days and it looks like I'll be sending it back and buy an Intel Laptop is that's what will happen with these machines.

IMHO you are panicking regarding what is happening to a small percentage of M1 users.

My sister and her boyfriend have a 13" Macbook Pro from 2014 and it is still going strong while they actually have the ability to swap the SSD and with M1 you don't.

I don't want to, given the worst case, send my M1 to Apple after 3 years and pay 700 to replace the complete logic board.

External SSD; Problem solved. And where're you getting this 3 years from?

Lonehorn shows 1% in about 2 months - that is 200 months (16.6 years) for 100%.

Dan Moren's numbers are even better as 1% in 3 months is 300 months or 25 years.
His 2017 iMac's 14% in 42 months (3.5 years) produces the same life span - 300 months (42*100/14).

People seems to be looking at just the raw TBW and not the percentage. If we can trust the TB numbers then why can't we trust the percentages these very same tools produce?

 

[wyk126]

So the consensus is that the high memory swap comes from unoptimized Rosetta 2 Apps? Which means Rosetta 2 is a dead end and unless any app is available natively, it will lead to long term durability issues?

Not necessary in my opinion, see most of the my bytes written was done by the browser - Edge, as it's caching most of the web and media content to the SSD instead of the memory. I think it's depending on how the app handles the memory / cacheing . I'm using 16GB ram btw.

 

[Forti]

This is really bad though. My M1 MacBook Pro will arrive in about 5 days and it looks like I'll be sending it back and buy an Intel Laptop is that's what will happen with these machines.

My sister and her boyfriend have a 13" Macbook Pro from 2014 and it is still going strong while they actually have the ability to swap the SSD and with M1 you don't.

I don't want to, given the worst case, send my M1 to Apple after 3 years and pay 700 to replace the complete logic board.

Don't return it, yet Even if you will have this issue - it will still last for years.

 

[Maximara]

Don't return it, yet Even if you will have this issue - it will still last for years.

For most of the ones that actually given a percentage I have seen produce more than a decade when you do the math. Longhorn clocks in at 16.6 years and Dan Moren at 25 years.