UPDATE:Nvme Health decreasing now with 3% for 5 days

[LittleBro]

Where did you get that drive from? Is it some reputable seller?

I'd make a backup clone of the drive just to be safe. You know, things may get worse.

 

[Wirko]

If anyone is still worried about the page file:

A blown up page file means that some insane app has requested an insane amount of virtual memory. The OS must either fulfill the request and reserve space to make sure there's space available, or deny the request and risk the app to crash. Reserving the space is done instantly even on an HDD, let alone an SSD. It just writes some metadata. This does not necessarily mean there's going to be a lot of actual writing and reading - unless and until the app starts using all that virtual memory for some processing or caching or whatever, but that is rare and improbable as long as the system has enough RAM.

 

[hermesa]

Where did you get that drive from? Is it some reputable seller?

I'd make a backup clone of the drive just to be safe. You know, things may get worse.

It's the biggest Bulgarian seller for IT products in Bulgaria.I've built my whole configuration buying products from them ,no single problem until now with this NVME. And I never do back ups or cloning my data.It's like couple of hours for me to retrieve it if It get lost.Nothing important , just some games , programs etc.

 

[3DVCash]

That's exactly what I am trying to explain to @ty_ger .All I did is normal use for the first couple of days of the disk.Not installing/uninstalling and then again installing games and large files for no reason...I guess I will keep it like that ,and just when it goes under -50% health i will buy new one on a different brand like samsung.With that tempo , I guess I will have at least 2-3 months of health until I buy the new one. Because If I RMA it , the chances of getting moneyback is none.They will just replace it with same drive ,with the difference it will be new one... The money back method is only available If I return the product to the first 14 days(2 weeks). Policy of the seller company I bought it from.And If I RMA it ,I will wait at least 1-2months for a replacement so it's more adequatte just to not do it at all.And buy a new one until this one dies.

I really wouldn't sweat it. I mean I'd def back up anything important on the drive, but that's just good practice regardless of drive health.

Drive health monitors are super vague, and as other posters have mentioned, there is probably some rounding going on behind the scenes.

Since you're beyond the refund period anyway, just use it! Enjoy it! Then check back in a few months to see how it's doing. Monitoring your drive health 24/7 won't make it degrade any slower, and you could be stressing yourself out over nothing so... give it time. You'll have a much clearer picture of what's really going on in 3-6months, and you can go from there.

 

[Vincero]

In order to loose 3% of lifetime, there must have been written to your disk more than 15 TB of data. (TeamGroup provides 700TBW for your disk.) I use HWInfo to check my disk for lifetime writes. Any time I do a benchmark, I can see that those writes changes accordingly.

The problem not seen and not reported by the drive SMART info is how much other write actions are happening outside of user actions - we literally have no way of knowing how many actual bytes / pages have been written to the NAND.
Thanks to pSLC caching and other seemingly 'modern' poor optimisations, write amplification on newer drives can be worse than older SATA SSDs from over 10 years ago depending on how data is used on them (note the use of the words 'can be' - if data is stored long term in TLC/QLC state, even as part of 'refreshing the cells', this packed data will keep the WA rating lower).
Is that 700TBW actual data from the host to the drive or does that include the drive reallocating written data out of caches, TRIM/GC routines, etc.? There isn't an actual metric of how much controller / physical NAND activity has gone on with this drive - it's all quite vague and not really in the user interest.

 

[Wirko]

Is that 700TBW actual data from the host to the drive or does that include the drive reallocating written data out of caches, TRIM/GC routines, etc.? There isn't an actual metric of how much controller / physical NAND activity has gone on with this drive - it's all quite vague and not really in the user interest.

This is data from the host. Of course a manufacturer needs some estimations and assumptions (about random vs. sequential access, etc.) to calculate the TBW rating. There's a JEDEC standard for that (JESD219) but it's paywalled, and SSD makers don't tell if their TBW figures are to that standard, or any other standard.

The program-erase cycle count for the memory chips themselves is much higher, it's around 3000 for TLC.

 

[LittleBro]

The problem not seen and not reported by the drive SMART info is how much other write actions are happening outside of user actions - we literally have no way of knowing how many actual bytes / pages have been written to the NAND.
Thanks to pSLC caching and other seemingly 'modern' poor optimisations, write amplification on newer drives can be worse than older SATA SSDs from over 10 years ago depending on how data is used on them (note the use of the words 'can be' - if data is stored long term in TLC/QLC state, even as part of 'refreshing the cells', this packed data will keep the WA rating lower).
Is that 700TBW actual data from the host to the drive or does that include the drive reallocating written data out of caches, TRIM/GC routines, etc.? There isn't an actual metric of how much controller / physical NAND activity has gone on with this drive - it's all quite vague and not really in the user interest.

I've never seen (personally) such a fast degradation even with all stuff working in the background of the NAND controller. 3% per 5 days is really bad. My oldest NVMe is from late 2018, it's 500 GB Samsung EVO 970 and it still has more than 91% life remaining (used as OS and partially as gaming drive with Win10 for 3 years). The most extreme thing I saw was 0.7% degradation over a week caused by browser storing/caching too much to disk. This was an extreme scenario, user regularly opened few hundreds of tabs and almost never restarted PC. You can turn off disk caching in hidden browser options (that's what I always do along with disabling pagefile) but you will need more RAM to cover this hundreds-opened-tabs approach.

 

[Vincero]

This is data from the host. Of course a manufacturer needs some estimations and assumptions (about random vs. sequential access, etc.) to calculate the TBW rating. There's a JEDEC standard for that (JESD219) but it's paywalled, and SSD makers don't tell if their TBW figures are to that standard, or any other standard.

Exactly, but even regardless of what standard they use (or don't use), actual physical disk usage is hidden information - it may not even be logged outside of some obscure running total which indirectly informs the media lifespan / wearout data, and the firmware marking off which memory pages are getting harder to program / shifted to spare capacity. Without seeing that we have no idea if the 3% is some arbitrary action or there is some physical usage pattern that accounts for it.
This is something that some drive and controller vendors did very well and others don't bother with / seem to actively obscure (and it only seems to be getting worse), probably because the below activity whilst great for performance isn't a friend of longevity.

The program-erase cycle count for the memory chips themselves is much higher, it's around 3000 for TLC.

Yeah, and for some QLC it's just 1000, which seems like a false trade-off.
Problem with that is writing to the pSLC will use 1 of those cycles, then re-packing that data into TLC/QLC storage levels is also going to use 1 of those cycles (albeit for a fraction of the same amount of pages). Not so bad (but not great) for TLC drives, but a bit concerning for any QLC drives, especially if used for boot or storage that might have frequent temp/scatch data use.

This doesn't even touch upon long-term storage / data retention quality/performance of the drives.

 

[nageme]

I've never seen (personally) such a fast degradation ... My oldest NVMe is from late 2018, it's 500 GB Samsung EVO 970 and it still has more than 91% life

But I suspect you've never used a "TeamGroup Cardea A440" drive.

There's no reason to assume different drives, controller chips, firmware versions, report the same thing.
I'm not sure where this 3% figure comes from. The OP posted a SMART screenshot in the previous page with an item titled "Percentage Used" (edit: presumably SMART 5, so probably that).

(BTW, too bad SSD-Z is no longer being actively maintained.)

 

[chr0nos]

just return the drive or RMA....

EOF

 

[Frick]

So anyway had a look on my SSDs. The Kingston is rated for 80TBW, and I've done 20TBW in 5 years, so that's pretty fair. The WD Black is 2 years old.

 

[A&P211]

Since everyone is post about their SSDs, here are my from my normal laptop and work laptop.

work laptop, don't ever buy a P3 plus SSD, they suck.


I dont know why everyone is trying to compare their own SSD usage or lifespans vs the OPs when every SSD is different. Maybe, just maybe the OPs SSD is bad and needs an RMA, but from the looks of it. It looks normal.

 

[lexluthermiester]

when every SSD is different.

Not THAT different.

They will just replace it with same drive ,with the difference it will be new one...

Chances are you'll get a good working drive as a replacement, so you should still do it. Sucks that you have too, that's life sometimes..

 

[A&P211]

Not THAT different.

I meant on usage and its different but the same. Or maybe I meant the quality of the NAND.

 

[LittleBro]

But I suspect you've never used a "TeamGroup Cardea A440" drive.

God bless, not. Anyway, a firmware should not report remaining lifetime lower by 3% just after 8 TBW on that drive (it has 700 TBW endurance).
If a drive reports such inaccurate statistics, it's actually absolutely pointless to make any further assumptions, as we lack enough proper data (we can only guess).

I'm not sure where this 3% figure comes from.

What is the current SMART report? Does it continue to degrade so fast?

 

[TechBuyingHavoc]

Yes, it would be helpful to get a update from OP on how the health is decreasing or not.

Concerning page files, I set my page file to a fixed amount and just keep it on the SSD, unless you happen to have a HDD. Modern SSDs have enough endurance that a Windows page file should not affect it in any substantial way, unless you are extremely RAM-starved.

Personally, if you are spending the extra money to get a SSD instead of a HDD for the same amount of storage, you should also be maximizing the amount of RAM that you can use on your system.

I will have to check my PC for how much page file space it actually uses after a few hours of gameplay but I currently have the page file set at 4096 MB. I used to keep it at the same amount of RAM but since I bought more RAM for my PC, I think the page file issue will be even less important.

 

[A Computer Guy]

So where are we at now? Still at 3% or did it jump again?

 

[eidairaman1]

That's exactly what I am trying to explain to @ty_ger .All I did is normal use for the first couple of days of the disk.Not installing/uninstalling and then again installing games and large files for no reason...I guess I will keep it like that ,and just when it goes under -50% health i will buy new one on a different brand like samsung.With that tempo , I guess I will have at least 2-3 months of health until I buy the new one. Because If I RMA it , the chances of getting moneyback is none.They will just replace it with same drive ,with the difference it will be new one... The money back method is only available If I return the product to the first 14 days(2 weeks). Policy of the seller company I bought it from.And If I RMA it ,I will wait at least 1-2months for a replacement so it's more adequatte just to not do it at all.And buy a new one until this one dies.

Well buy 1 and them immediately rma the faulty drive, but before you do back up your data.

 

[hermesa]

God bless, not. Anyway, a firmware should not report remaining lifetime lower by 3% just after 8 TBW on that drive (it has 700 TBW endurance).
If a drive reports such inaccurate statistics, it's actually absolutely pointless to make any further assumptions, as we lack enough proper data (we can only guess).


View attachment 366741

What is the current SMART report? Does it continue to degrade so fast?

I reinstalled windows and all the games , except this time I made 10% unallocated space and installed only 2 large games on it (other games are on 2nd SSD + HDD),no page file ,now there are ~3900GB(0.53% of the700TBW) written with 8126GB reads and still on 97% with 212 Power On Hours ...for now.I hope it stays like that for at least 1-2 months.I don't have any intentions to install more games on it.Right now is 505GB free out of 840GB.Maybe I can install one more game just to test ,but still not sure if I want to do that. BTW , can a pirate copy of windows activated from a kms server cause this problems or any viruses ?

 

[A Computer Guy]

... BTW , can a pirate copy of windows activated from a kms server cause this problems or any viruses ?

Well...consider the possibility of a rouge program burning up SSD writes. Honestly I don't know why we don't hear more reports of malware burning up SSD's as an attack vector to help inflate manufacturers bottom line in sales but I'm sure we will get there someday.

Putting aside my imagination for a moment I would be highly suspect of any OS and/or software not from official sources. You could try reinstalling from scratch an official un-activated Windows install for awhile and see if you notice different behavior.

 

[ty_ger]

More deletes and more writes in order to figure out where all the writes are going.

 

[lexluthermiester]

What is the current SMART report? Does it continue to degrade so fast?

It's a few pages back, the OP ran TEAM Groups own utility.

 

[Marcus L]

More deletes and more writes in order to figure out where all the writes are going.

You can't make it up

 

[eidairaman1]

I reinstalled windows and all the games , except this time I made 10% unallocated space and installed only 2 large games on it (other games are on 2nd SSD + HDD),no page file ,now there are ~3900GB(0.53% of the700TBW) written with 8126GB reads and still on 97% with 212 Power On Hours ...for now.I hope it stays like that for at least 1-2 months.I don't have any intentions to install more games on it.Right now is 505GB free out of 840GB.Maybe I can install one more game just to test ,but still not sure if I want to do that. BTW , can a pirate copy of windows activated from a kms server cause this problems or any viruses ?

Just stop it!

 

[hermesa]

Just stop it!

Who cares if it's faulty drive anyway... I am gonna keep it until I buy a new one and then just RMA it and get a brand new of the same model which i will sell in order to try get at least 70% of the price I spent on it...I will even make more writing and reads but not now because I already got pissed off it ,in order to test and verify if it's software bug or not.I already accepted that the money spent on the drive are thrown in the trash but I will RMA it until it's in warranty...Just can't RMA it now until I buy a new one like Samsung pro 990 or something like that :]