There's another 970 Evo with 1 media error on the first page. Looks rather peculiar and more like some sort of firmware issue. My Evo+ shows 0. Those errors should appear around the time after the SSD exhausted its reserves, not with just 3.5TB written which is nothing.
A quick look on google and I found some other Evos with rather early errors on that field. One with 17TB of writes and 94% available spare had 14 errors.
Somewhat interesting. 17TB is still nothing to these drives, at least this is the story from the few endurance tests I've seen, in which I believe these drives, the 250GB variant, lasts around 4-5PB under non stop stress testing. I've also seen a 960 Evo 250 used to "150%" with just 45TB, which look quite disappointing. No media errors on it, "spare" was 100%.
I have my suspicions that the reporting is hardly accurate and that not all drives under the same label are functionally similar, not even close.
Also, it looks like Samsung honors RMAs for drives within warranty that start to show up media integrity errors, so there's that.
Here's the best endurance testing article I've read, use translate as it's Russian.
Бытует мнение, что одним из самых существенных недостатков твердотельных накопителей выступает их конечная и притом относительно невысокая надёжность.
3dnews.ru
At lot has happened since that post.
First the current status. 51 data and integrity errors. Also down to 93% spare.
So what has happened.
I had noticed a couple of months back erors had creeped up to the teens, but at the time no ill effects noticed.
Then I got my 3080FE, and it has no Win8 drivers so had to update my OS.
I proceeded to make backup of my system drive and also the second partition as I wanted to resize the OS partition larger, both of these on the 970 EVO.
Macrium failed to read data on the second partition, CRC read error again and again.
Each time the error popped up the medi aand data integrity error counter went up, and sometimes it also triggered the available spare to go down.
I then started running chkdsk /r as advised by macrium and this massively accelerated both stat changes but eventually the data became readable and I did the backup.
I then moved the ssd to ryzen rig, tried to beat it down with reads and writes, not a single SMART issue logged.
I didnt RMA to samsung as my experience with them has been poor (they seem incredibly strict, rejected previous RMA with 100s of CRC logged in smart), as with brexit would probably be huge shipping delays/costs as RMA centre not in UK.
I did some testing on main PC, reduced system agent voltage which fixed CRC errors on SATA ports, since I fixed that issue I became confident the problem was the NVME signalling rather than the drive, as it also was finein ryzen rig.
Since I have put back in this PC there has been no stat changes, and its been in here since mid February. So looks like was a system agent voltage been too high issue, but I have been checking it now and again.
Older image below for when put back in main pc.
Also since moving to Win10 the unsafe shutdown no longer goes up on shutdown anymore either which is good.
(also you may notice one of my spindles was in warning status in feb), I reformatted that drive with full format to try and get reallocated sectors (was loads of pending sectors), and they all cleared out instead, which is why its back to blue now. This rig has had some weird i/o issues for sure.
