Brand new SSD is already at 89% health.

Hardware
1

Should I be worried?

Long story short, this morning my laptop’s 1TB Seagate Firecude hybrid SSHD (basically a HDD with an 8GB flash storage module) that I’ve been using for the last four years starting throwing errors, hanging on reboots, slowing down shutdowns to 5+ minutes. So I ran hdsentinel on it and it showed 12% health with two days estimated lifetime left.

So I went out and bought a long overdue upgrade, a 1TB ADATA SU650 SSD. Came home, plugged it in and I was suprised to see it came at 97% health out of the box. Just formatting it for bringing in my backups, I noticed it went down to 94% health in less than an hour, also mentioning 6 errors ocurred during data transfer. So I went back to the store, returned it and exchanged it for another one of the same kind.

Now this second one, was 100% healthy out of the box. I formatted it, I checked for errors and since there were none, I copied my backed up data onto it - about 440GB of mostly images and videos. So after that operation, this second SSD is claiming to be 89% healthy, with 11 errors ocurred during data transfer. To make matters even strangers, hdsentinel claims 15+TB written, while I only wrote 500GB of data onto it.

Out of curiosity, I checked my original disk again, the one I replaced for being 12% healthy and having two days left, and lo and behold, that is now 99% healthy with about 1k days lifetime left, but still mentioning a few hundred errors.

I should mention this is not a system drive, I have a dedicated M.2 SSD for my system.

Now, I could exchange this again, and maybe this time go for a Kingston, or a Samsung, but I wanted to ask for feedback first.

Is this worrisome?

2

Check the cables running to the drive and maybe try a different one?

3

On another computer, see what happens.
Otherwise, yes - try to replace the cable. Both.
For now.

But in general, the batch of disks could have been like that. The races came together.

Of the options, what to buy if all stars, patriot, intel, Samsung. As far as I’m concerned.

4

Digital vampirism. Ha ha.

Random thoughts: I wonder if

  • There is a software (driver), hardware (controller) or compatibility (OS) issue. They do smart things to mitigate wear. The opposite could happen if something isn’t working properly.
  • The data or drives are broken, thus transfer is making multiple attempts to copy certain data.
5

Have you checked the temp of the drive when checking its health?

SSDs do get quite hot around the flash chips and controler, and when they’re hot, they have a tendency to bring false reports. I’m not saying that’s the problem, but just to check it.

If you ever change your SSD to store personal data (images, docs, videos, …), I would go for an MLC type of flash chips (rare to find nowadays) and then for TLC drives. I would never go for a QLC drive, just for peace of mind…

I personally am quite happy with my Crucial MX500 drives. Old model, but trusty.

6

Yes, temp is a major factor. Mine are older models too. I also don’t use ones with high storage. I rather have several smaller drives.

7

It did get a little warm when I was copying files onto it, but I don’t think it got over 35C. And speaking of flash chip types, the one I got is SLC, which is supposed to be the most durable of the bunch…

I think I will exchange it again today… Maybe third time’s a charm…

8

Sorry, but isn’t that Adata drive a TLC drive with SLC cache?

A real SLC drive is very expensive, if I’m not wrong… (but would be an ideal choice, on the other hand)

9

You may be right, the manufacturer does mention an SLC cache and nothing else…

10

Could alignment be an issue? Bad alignment can cause a high value of physical vs logical writes, but the difference you list is huge – ’ hdsentinel claims 15+TB written, while I only wrote 500GB of data’).

https://wiki.debian.org/SSDOptimization

or

11

Could these 15TB be because of swap? Dunno how aggressive your OS is with it but in macOS when the new M1 machines were coming out, was writing hundreds of TB to disk in swap, due to the 8GB of memory, in the lower end systems, and wanting to keep things stable.

In linux I recommend every everyone to run zram, it’ll compress your memory and allows you to basically double your ram with only a 1-3% penalty to your CPU performance. The “download more RAM” joke is now true indeed.

12

This is just a storage drive. My system is on a separate M.2 drive, as is the page file.

1 Like
13

Other people have already mentioned your sata cable, but it might also be worth a try to switch your sata power cable, if possible on a modular power supply, you could also swap to a different port.

14

A little update

I took it back to the store and they gave me a third copy of the same drive. After running through the paces, copying my data onto it, this one appears to be better behaved. It’s 99% health, with just 1 errors occurred during data transfer. Oddly enough, it still reports crazy amounts of written data - 8.26TB, which is impossible since I’ve only had it for an hour, I know exactly what I wrote to it, and that is a fraction of that.

However, I’m thinking there is an issue with how hdsentinel calculates stuff.

I ran smartctl -a /dev/sdb and under Host_Writes_32MiB it mentions 8457. Now 8457*32=270624 And 270624 MiB is about 284 GB, which kind of makes sense, because the actual amount of data that I’ve written to the disk is 285.5 GB

1 Like
15

Let us know if anything changes a week later. Hope it is all good news from here on out.

1 Like
16

A bit of an update.

The third ADATA disk started degrading again, faster than I’d have wanted. A couple of weeks later it was down to 96%, with many errors so I decided to take it back again, pay a little extra and buy the new Samsung 870 QVO.

This one reports correctly the amount written and for now The status of the solid state disk is PERFECT. with 100% health. Here’s hoping for the best.

2 Likes