SSD Corupting/Losing data. SMART says OK.

[p4ck3tl055]

My rig has 2 Samsung EVO 840, 250GB drives in it. I use the first for the OS (Win 8.1 Pro) and basic apps. I use the second for games - mainly a modded Skyrim. I've had these drives for less than a year - the Skyrim drive is less than 6 months old.

Since Christmas I've been noticing that I'm either loosing files or am getting files corrupted on the Skyrim drive. I notice this because things in Skyrim don't work they way they should and re-installing the affected mod from its source (a zip file on a mechanical drive) fixes the problem. (I've done the same repair three times today for one mod... These are static files, a 3D mesh in this case.)

I've checked Windows, Samsung Magician and Crystal Disk Info. All three say there is nothing wrong with the SSD. I've checked my RAM earlier today - no issues.

So right now, I'm backing up the entire drive to a mechanical drive and I'm planning on doing a "secure erase" on the SSD. Why? Because I don't know what else to do.

Is there anything else I can do to check/validate the drive?

I don't trust this drive at this point and I really want to RMA it. But how can I prove to Samsung that there is something wrong when their (seemingly) only utility says otherwise?

Any help at this point would be welcomed.

 

[OlyAR15]

Try a new SATA cable. Also, ensure it is fully seated.

 

[MarkLuvsCS]

I would also run memtest to ensure there isn't an issue with the RAM.

 

[Elixer]

Test the SSD by zip/7z/raring up 300MB of whatever files on the HD, and then, copy it to the SSD that has problems, then copy that rar again from the SSD to itself a few times (so you end up with a few files), then, verify those archives, and they should come back with 0 errors.
SMART won't help in these situations.
If you have errors, then, you should check the RAM, as was mentioned, but use memtest86+ and leave it running all night, you should have 0 errors.
If that passes, then test the CPU with prime95 or occt for a few hours, and it should come back with 0 errors.

 

[Puffnstuff]

While you're at it get the samsung reconditioning utility for your evo and run it.

 

[redzo]

While there may be something related to ram corruption, I would eliminate that in our current situation. Why? RAM corruption usually brings other nasty things to table: things like random BSODs, sudden system resets and yes file system corruption also.

Good old checkdisk comes to the rescue. Just open up an elevated(run as administrator) cmd(command prompt) and type the following after backing up your important data residing on the scanned partions:
chkdsk <<driverletter>>: /r; replace <<driveletter>> with partitions located on the troubled SSD!

/r does a full complete scan of the filesystem partition. If data corruption occurs, chkdsk should point it out after completion.

It is not mandatory for smart to point out drive failures. Many drives get broken without any recorded smart failure event: although smart says everything is alright, the drive is physically borked. chkdsk reports should be taken into account on RMA.

 

[mikeymikec]

There's a firmware update for the EVO IIRC. I would probably try that after all the other suggestions already made and your data has been backed up first.

Furthermore, you could use Prime95 on the test that gives your RAM a workout (as well as an overnight test with memtest86+).

I assume that the system in question is in your sig - is Intel Rapid Storage installed? BIOS updates?

 

[coercitiv]

Do you have RAPID mode enabled for he "faulty" drive?

 

[redzo]

+1 for that. I would disable that if it's the case, and check how things work out without it.
Quite important... A sudden power loss, not enough time to flush data from ram back to disk => file corruption heaven.

 

[nk215]

Why not create a checksum for your entire game folder then verify the checksum from time to time?

Do this when you have everything backed up onto mechanical HDD. Use the mechanical HDD for actual use and keep the SSD plugged in for test.

 

[AdamK47]

While there may be something related to ram corruption, I would eliminate that in our current situation. Why? RAM corruption usually brings other nasty things to table: things like random BSODs, sudden system resets and yes file system corruption also.

Good old checkdisk comes to the rescue. Just open up an elevated(run as administrator) cmd(command prompt) and type the following after backing up your important data residing on the scanned partions:
chkdsk <<driverletter>>: /r; replace <<driveletter>> with partitions located on the troubled SSD!

/r does a full complete scan of the filesystem partition. If data corruption occurs, chkdsk should point it out after completion.

It is not mandatory for smart to point out drive failures. Many drives get broken without any recorded smart failure event: although smart says everything is alright, the drive is physically borked. chkdsk reports should be taken into account on RMA.

I would use chkdsk <driveletter> /r /f

It will fix errors on the disk, if there are any. I ran it on my 5TB SSD array. It took a couple hours with 4+ TB of data. The Intel controller limits me to only 1.5GB/sec.