Really Strange Corsair MP510 NVME SSD Stability and Write Issues

[ltron]

This Corsair MP510 NVME drive (960GB version) has been working flawlessly until very recently. I noticed Control was stuttering a bit more than usual so I ran Crystal Disk Mark and the PC BSODed but there was no code because the dump could not be written due to an I/O error. Now pretty much every time I run Crystal Disk Mark it will do this and/or complain about I/O errors. In a separate incident Steam even refused to install a patch because it aborted the process due to a write error.

However, despite all of this SMART says the drive is good and its remaining lifespan is 99%, temperatures are in the 30s-50s C under load. However, I'm convinced the drive is faulty and will probably die fairly soon:

I took my Samsung SM961 512GB NVME SSD which is on a PCIE adaptor from my old PC and put it in this machine and it runs Crystal Disk Mark absolutely fine many times in a row.

I moved the MP510 from the CPU NVME slot to an NVME slot connected to the X570 chipset instead and it still malfunctions.

I did an in place repair install of Windows and downgraded the AMD chipset drivers to the previous version and still no joy.

I disabled Windows Defender's realtime protection and tried an older version of CDM, locked my CPU to its baseclock by disabling turbo and disabled C States and that didn't work either.

What I'm going to try next:

A clean install of Windows 11 and test with nothing installed.

Moving the MP510 to my old PC which only has a slower 20Gb/s NVME slot and seeing whether it still fails.

Then I will RMA, I just wanted a sanity check to see whether I'm right and that it is the drive. It only lasted two and a half years, it has a very high endurance rating for the flash cells (which is one of the main reasons I bought it) but still seems to have begun to fail prematurely in some way. SMART is useless though.

Aida 64 disk bench read suite works although scores are lower than they were initially and AS SSD works sometimes but gives me a timing error on other occasions. After malfunctioning in CDM the drive can go into a sort of protection mode on reboot where read speeds are limited to around 1000 MB/s for a few minutes before recovering.

CPU and memory stress tests pass with flying colours.

AMD Ryzen 3900X
32GB Crucial Ballistix 3600MHz RAM (4*8GB)
Gigabyte X570 Master BIOS F35 AGESA 1.2.0.5
Noctua NH-D15S with 2 fans
Nvidia RTX 3080 FE 512.15 with ReBAR and HAGS enabled
Corsair RM1000i PSU
Corsair MP510 960GB SSD
Phanteks P600S
Windows 11 Pro 64bit

 

[ltron]

Got 2 of those 960gb mp510
Never had any issues with mine
Could try testing with a live USB
Either Linux or a Windows PE version

Thanks for those tips. I do find it strange that SMART is saying the drive is fine but putting it under stress in a benchmark will cause an almost instant BSOD.

 

[ltron]

Yeah that's why I have windows on a 2nd drive too
Any weird issues I boot to that drive
And see if it's replicated or not
After that I use bootable USB like mentioned

Well, as a secondary drive in my old PC and in a half speed NVME slot the problem has been confirmed, the drive seems to disappear or lock up. See this screenshot where it has scored 0 MB/s for numerous tests:



and the drive often becomes inaccessible after malfunctioning in this manner:



What puzzles me is that SMART says this is a healthy drive with 99% remaining life.

 

[ltron]

Yeah definitely looks wonky then
No idea how smart tests it or what parameters
My next thing I would try
Would be a secure erase if corsair toolbox actually does it properly now
I know in the past when I tried it that it didn't
Do it properly
I had to resort to Linux
Not sure if corsair toolbox was just doing an overwrite

Thanks for the suggestion, I just did this through the Corsair Toolbox (it allows you to do so only if it is a secondary drive). No difference unfortunately, it is still malfunctioning.

I was actually thinking about upgrading to a PCIE 4 NVME drive anyway and this gave me the excuse to do so. I've got a Seagate Firecuda 530 coming tomorrow, I'll RMA the Corsair at some point.

 

[ltron]

Out of curiosity
Did corsair toolbox do it properly?
Ie secure erase should be instant
If its still only overwriting would take a little bit of time

Actually it did take about 5 minutes, but it said it was erasing the cells and Corsair called it 'secure erase' so I had assumed so. It's not very impressive if their secure erase feature doesn't actually secure erase.

 

[ltron]

Yeah after I used it previously I had doubts as it took a while
So I ran recovery software on it
And it found the files
Did it in Linux or parted magic and it was instant
It's meant to be as it just sends a voltage spike to the cells
Couldn't recover a single file after doing it with that

That's shocking, thanks for letting me know. I thought Corsair would secure erase an SSD when that's what they are telling you they are doing.

 

[ltron]

Hi sorry for the very delayed post. I'm having a similar issue, BSOD with Corsair MP510 960gb when using crystaldiskmark test in windows 11. Just wanted to check how you got on, and if the drive did end up failing?

I replaced it with a Seagate Firecuda 530 because I tested it using Crystal Disk Mark in another PC as a secondary drive and it would disappear and drop out (that's why it BSODs when used as a primary drive).

If you can try it in another PC that would be a good idea.