Samsung 870 EVO - Beware, certain batches prone to failure!

[chrcoluk]

Have been using one for just over a year now as C drive , including page file on C.
No issues.

View attachment 299758

Thats a really nice read to write ratio, especially considering its OS drive.

 

[Oswald]

Hi,

I have read the first and last page of this thread.

I have an EVO 4TB from 2021 which had his reallocated sectors amount growing quite quickly for a new SSD in the first weeks (it's now around 1000).

But I had no issue "in real life" so I told myslef that's fine as the warranty lasts 5 years.

1 year and half later, I had my first corrupted files impossible to move to a backup drive. I noticed I couldn't read the files at all (photographs). I checked and saw ECC count around ~15k.

I saw this thread and felt concerned so I opened a ticket and requested an RMA.

My SSD has just left my city to Samsung support. I'm a bit worried since Samasung Magician shows a healthy drive despite the critical state of ECC.

Hopefully they will send me a brand new drive.

Question: are current EVO 4TB still affected by the early 2021 issue? When I read the last page of this thread it doesn't look so comforting...

Thanks

 

[SPDIF]

Hi,

I have read the first and last page of this thread.

I have an EVO 4TB from 2021 which had his reallocated sectors amount growing quite quickly for a new SSD in the first weeks (it's now around 1000).

But I had no issue "in real life" so I told myslef that's fine as the warranty lasts 5 years.

1 year and half later, I had my first corrupted files impossible to move to a backup drive. I noticed I couldn't read the files at all (photographs). I checked and saw ECC count around ~15k.

I saw this thread and felt concerned so I opened a ticket and requested an RMA.

My SSD has just left my city to Samsung support. I'm a bit worried since Samasung Magician shows a healthy drive despite the critical state of ECC.

Hopefully they will send me a brand new drive.

Question: are current EVO 4TB still affected by the early 2021 issue? When I read the last page of this thread it doesn't look so comforting...

Thanks

I have not seen any defects with the new batch of drives i got back, just make sure it has the latest firmware and you should be fine. AFAIK, it was a mix of cheap NAND and bad firmware. Just about when the pandemic begin. Many manufacturers delivered bad quality during that time.

The new drives also don't go suddenly to 99 percent and then stay like that. They now all showing 100 percent life even after written 1 TB.

I think you will get a new drive for sure.

 

[BBP]

The system has xp/win7 and win10 x64(mostly for backup and things that don't work on xp/win7) installed on the SSD. The issue is present in windows 10 (the update problem/latest version of SM) and the blocks can't be fixed also (read error). Also i tried resizing the windows 10 partition (have 4: xp, win7, win10 and storage) which was working before the bad blocks appeared and i get the the same 'read error' at ~80% of the process, so i really think it's a hardware issue.

Maybe win7 is the issue, unsupported to do it's thing probably..

Firmware update worked on windows 10 with internet connection from my phone and also one bad block went away, thank you.

 

[SPDIF]

The system has xp/win7 and win10 x64(mostly for backup and things that don't work on xp/win7) installed on the SSD. The issue is present in windows 10 (the update problem/latest version of SM) and the blocks can't be fixed also (read error). Also i tried resizing the windows 10 partition (have 4: xp, win7, win10 and storage) which was working before the bad blocks appeared and i get the the same 'read error' at ~80% of the process, so i really think it's a hardware issue.


Firmware update worked on windows 10 with internet connection from my phone and also one bad block went away, thank you.

In your place I would quickly make a backup of that SSD! Especially if there is data that you do not want to lose. Keep an eye on that reallocated sectors value, if it gets even higher now, then your SSD is dying. Then you have to RMA it to Sammy.
Don't trust your precious data to a sick SSD! A man forewarned is worth two.

Bad blocks are not gone, they are only internally marked as unusable. What is lost remains.
The bad blocks are then replaced by spare blocks, but they will run out at some point.
Many people still wrongly think that bad blocks are suddenly gone... They are not, they are replaced by spare blocks, the bad ones stay and never go away.

Anyway a good SSD should never have bad sectors... Whenever you see this, then backup, backup, backup! At that moment just as an old HDD, you have a bomb ready to explode in your hands. I would never trust such an SSD or HDD in my business, and my customer's. I have enough experience with this in about 30 years.
You have a greater chance of losing everything from data, the longer you postpone everything with such an SSD.

But if you like playing on the edge of the abyss, go ahead! And i say it again; A man forewarned is worth two...

Prices are currently spectacularly low for some SSDs right now, so I wouldn't wait for to pay double for that.

If you see this on any HDD or SDD, i wish you good luck...

In this case an SSD with 329 bad sector's, do i need to say more?

 

[SuperConker]

Thats a really nice read to write ratio, especially considering its OS drive.

Mine has a near perfect 1:1 ratio lol.

It's been running 24/7 for 2 years and 6 months, and is only down to 92% so far.

 

[veodko]

We did a little research at work out of curiosity and found next 4 dying/dead Samsung SSDs in some PCs:
Three 870 EVO's 1TB, all manufactured in 2021 running factory SVT01 software.
One 980 PRO NVME 250GB that had just 3TBW...

What's funny is that the disks seemed perfect when looking at the SMART values either in CrystalDiskInfo or Samsung Magician. Until we started copying some files from them and doing full scans in Magician....
Then suddenly the ECC error rates started jumping up to even 11k, and magician was hanging for some long time on certain blocks just to mark them as Bad.
One 870 died completely (despite working normally before this) and Windows started just freezing, not booting etc with it plugged in. So complete trash.
The rest was somehow "saved" by doing a update of the software, secure erasing them and doing a full format in Windows. The bad blocks disappeared, and doing further scans in Magician stopped increasing the error rates in SMART.
What's interesting is that the 870's were updated through Magician, but to SVT02 only (and it claims its the latest version). Wheareas a brand new 870 produced this year came with SVT03B6Q software.

We'll throw some non-sensitive data onto them and monitor how it goes after some time.

 

[SuperConker]

When it comes to the 870 EVO failures, CrystalDiskInfo does NOT tell you the full truth about the Health of these drives.
CrystalDiskInfo might even make it look like the drive is perfectly fine (even though it's failing).

I would recommend checking it with "Hard Disk Sentinel" (HDSentinel) to get a more correct picture of what's going on.

 

[SPDIF]

Remember that BAD blocks never disappear if you do a secure erase. They are still there but internally replaced with SPARE blocks. The SSD will then stop writing to this bad block.
Spare blocks are limited like HDD, once used up it's done for. Like with old HDD, once you have bad blocks on an SSD, take a backup ASAP and have a new disk ready. If you don't care for your data, then you can go on use it until it dies.

The new 870EVO SSD with 1TB space are now shipped with SVT03B6Q firmware, not available online because they use new NAND packages. The 870EVO 500GB and 4TB are still shipped with SVT02B6Q firmware.

 

[P4-630]

The new 870EVO SSD with 1TB are now shipped with SVT03B6Q firmware, not available online because they use new NAND packages. The 870EVO 500GB and 4TB are still shipped with SVT02B6Q firmware.

Do you work for Samsung?

 

[SPDIF]

Nope i have my own shop over here. After customers complaining with bad 870EVO SSD, i think i have enough experience with it. We actually opened two SSD and the NAND packages have a totally different number.

When you ask SAMMY, same answer, they are looking into it...

I have no problems with any SSD now from SAMSUNG and i sell them again like before. No problems with 870EVO anymore, just still old batches that are returned DEAD.
Problems with 980 and 990 PRO SSD are also fixed, no complaints anymore. 990Pro SSD are also shipped with new firmware not available online.

Before i sell them i make sure they all have the latest firmware installed.

 

[Tetrahedrix]

Like many my Samsung Evo 1TB was purchased in 2021 and only started giving me issues in April this year with a small incident of bad blocks moved. I was then able to update the firmware which helped for a bit then like others the corruption has already begun. Mines gone from 300ish uncorrectable ECC error rate attributes to now 1346 over a course of 2 months, thankfully its not very frequent and has not hindered anything. I've moved a huge chunk/already had backups or file syncs of my files off this drive for awhile now. But, like many going to have to decide soon and get another drive and start over again. In my case this is the only attribute rising at odd times of the day every now and then. Slower shutdowns at times have been noticed and I suspect its the drive doing something or windows. If it help, so everyone should be prepared for sure with or without updating the firmware on the older drives, I can confirm once its begun the update may slow it down but it still is like a virus/corruption is going to occur.

I hope the newer drives are indeed better, thinking of upgrading to the 2 or 4TB 870 Evo's.

We did a little research at work out of curiosity and found next 4 dying/dead Samsung SSDs in some PCs:
Three 870 EVO's 1TB, all manufactured in 2021 running factory SVT01 software.
One 980 PRO NVME 250GB that had just 3TBW...

What's funny is that the disks seemed perfect when looking at the SMART values either in CrystalDiskInfo or Samsung Magician. Until we started copying some files from them and doing full scans in Magician....
Then suddenly the ECC error rates started jumping up to even 11k, and magician was hanging for some long time on certain blocks just to mark them as Bad.
One 870 died completely (despite working normally before this) and Windows started just freezing, not booting etc with it plugged in. So complete trash.
The rest was somehow "saved" by doing a update of the software, secure erasing them and doing a full format in Windows. The bad blocks disappeared, and doing further scans in Magician stopped increasing the error rates in SMART.
What's interesting is that the 870's were updated through Magician, but to SVT02 only (and it claims its the latest version). Wheareas a brand new 870 produced this year came with SVT03B6Q software.

We'll throw some non-sensitive data onto them and monitor how it goes after some time.

I was able to update mine and it did turn into the when I first learned of this to that exact version "SVT02B6Q" despite the SSD being from the 2021 batches, perhaps mine a later batch of those bad batches. Mine is increasing though at a slower rate since that update.

 

[Solid State Soul ( SSS )]

Can we please close this thread, what ever issues plagued the early batches am sure its fixed by now and it bothers me seeing this getting activity more than a year and half

 

[SuperConker]

Can we please close this thread, what ever issues plagued the early batches am sure its fixed by now and it bothers me seeing this getting activity more than a year and half

The issues are not fixed as drives are still failing to this day.

Whether it bothers people who are unaffected by this is irrelevant to us who lost data due to this fiasco from Samsung,
i personally lost 250 GB of data due to this.

Thread stays open until it's actually been fixed.

 

[Oswald]

Agreed. Plenty of failing SSDs are still in use and their owners are yet to discover it in a few weeks, months or years. The warranty is up until 2026 so I guess this thread can still help until this date.

 

[Ahhzz]

Agreed. Plenty of failing SSDs are still in use and their owners are yet to discover it in a few weeks, months or years. The warranty is up until 2026 so I guess this thread can still help until this date.

I assume you got your drive replaced at this point?

 

[Oswald]

I assume you got your drive replaced at this point?

Yes, it wasn't easy to find the RMA process in Europe, but once I did it took only 1 week to send and receive a new disk

Fingers crossed now!

 

[SuperConker]

I do believe the title of the thread could be updated a little to reflect just how widespread the issue really is.

Pretty much all of the batches of the 870 EVO made between 2020 and 2023 are affected by this issue
(so it's not just "certain batches" anymore).

Even if you update them to the newer SVT02B6Q firmware right out of the box, they will still fail.


And the new batches from 2023 (with the brand new SVT03B6Q firmware) are still too new to know if it has fixed anything yet.

 

[Oswald]

I do believe the title of the thread could be updated a little to reflect just how widespread the issue really is.

Pretty much all of the batches of the 870 EVO made between 2020 and 2023 are affected by this issue
(so it's not just "certain batches" anymore).

And the new batches from 2023 (with the newer SVT03B6Q firmware) are still too new to know if it has fixed anything yet.

I'm still surprised all this "story" can be corrected by a firmware. Hopefully Samsung did correct the hardware level issues which we will never know about.

 

[Nicholas Steel]

I ordered 2 of these as they were part of the Amazon Prime sale thing, bringing them close to MSRP but then I got reminded of this fiasco and cancelled the order... what are some good alternative 4TB SATA SSD's at a reasonable price?

 

[Count von Schwalbe]

I ordered 2 of these as they were part of the Amazon Prime sale thing, bringing them close to MSRP but then I got reminded of this fiasco and cancelled the order... what are some good alternative 4TB SATA SSD's at a reasonable price?

MX500 are $200.

 

[Nicholas Steel]

Thanks, that's what I ended up going with.

 

[P4-630]

I'm getting an MX500 2TB soon...

Also as far as I know, Crucial is the only company having SSD's with power loss protection, which is a nice thing to have.

 

[AusWolf]

Imo, Samsung SSDs are overrated, especially the Evo ones that aren't any better than your regular Crucial MX/BX drives, but come with a price premium for some reason.

 

[chrcoluk]

We did a little research at work out of curiosity and found next 4 dying/dead Samsung SSDs in some PCs:
Three 870 EVO's 1TB, all manufactured in 2021 running factory SVT01 software.
One 980 PRO NVME 250GB that had just 3TBW...

What's funny is that the disks seemed perfect when looking at the SMART values either in CrystalDiskInfo or Samsung Magician. Until we started copying some files from them and doing full scans in Magician....
Then suddenly the ECC error rates started jumping up to even 11k, and magician was hanging for some long time on certain blocks just to mark them as Bad.
One 870 died completely (despite working normally before this) and Windows started just freezing, not booting etc with it plugged in. So complete trash.
The rest was somehow "saved" by doing a update of the software, secure erasing them and doing a full format in Windows. The bad blocks disappeared, and doing further scans in Magician stopped increasing the error rates in SMART.
What's interesting is that the 870's were updated through Magician, but to SVT02 only (and it claims its the latest version). Wheareas a brand new 870 produced this year came with SVT03B6Q software.

We'll throw some non-sensitive data onto them and monitor how it goes after some time.

What you just described mirrors the old 840 issue, data read problems due to data rentention issues, fixed by fresh write or new format (temporarily on the 840 without fixed firmware).