ADATA SX8200 Pro 2TB SSD - Error: 23 data error (CRC)

[abstract6917]

I'm running tests on my SX8200 Pro 2TB SSD. I decided to do this because I noticed that the read speeds were very slow when I was migrating the data to a new SSD. It was around 60MB/s. Both SSDs are NVME drives so I expected much faster speeds. I tested doing the same file copy operations with my other SX8200 Pro and it was achieving more than 2000MB/s. File copy tests were performed with FastCopy, so I can rule out this being a Windows explorer issue.

The only difference between the drives I could think of is that the data on this slow drive was written a couple of years ago and has been almost untouched since then, and has been accessed very rarely. So my thoughts are the slow read speeds could be to do with the charge in the NAND cells running low over time.

I actually tested this theory by copying fresh data onto the drive and then measuring the speed when copying this fresh data to another drive. The speeds were very fast, 2000MB/s+, as I expected. So I think this theory is correct.

Anyway, I ran a surface read test and this showed a strange cyclic pattern in the read speeds. I don't know if this is anything to worry about? I thought it may be the SLC cache on the SSD filling up, but the read speed during this part of the test was less than 100MB/s, so I can't imagine that is the issue.

To investigate more, I am now running a Read + WRITE + read test to see if this can "fix" the slow read speeds for the data. However, it is now showing many bad blocks with the error:

Code: Select all

Error: 23 Data error (cyclic redundancy check)

The speed of the test is also extremely slow (less than 1MB/s). Is this an indication of a faulty drive? Strangely, the bad blocks appear to be in the empty part of the drive that has never held any data.

SMART info shows no errors and full health for the drive.

It's worth noting that my data is already backed up in multiple locations so I have no concerns about losing the data on this drive.

Screenshots are below. What should my next steps be? Should I be concerned that the drive is failing? Should I reinitialise the disk? Is there any point letting this current test complete? It's taking a long time...


Surface read test:

20240325-202022_R_ADATA_SX8200PNP-surface-full-noSN.jpg (295.76 KiB) Viewed 2513 times

Surface read test speeds and temps:

20240325-202152_R_ADATA_SX8200PNP-surface-full-st-noSN.jpg (264.33 KiB) Viewed 2513 times

Surface Read + WRITE + read test:

SX8200 CRC Errors-noSN.jpg (318.69 KiB) Viewed 2513 times

[hdsentinel]

> So my thoughts are the slow read speeds could be to do with the charge in the NAND cells running low over time.

Yes, you're absolutely correct: the data "harder" (= slower) can be read back.


> Anyway, I ran a surface read test and this showed a strange cyclic pattern in the read speeds.
> I don't know if this is anything to worry about?

No, this is normal, no need to worry about it.


> I am now running a Read + WRITE + read test to see if this can "fix" the slow read speeds for the data.

Yes, this is generally an excellent idea, as this test designed exactly to "refresh" a flash storage (SSD, pendrive, memory card) to make further reads faster.


> However, it is now showing many bad blocks with the error:
>

Code: Select all

Error: 23 Data error (cyclic redundancy check)

> The speed of the test is also extremely slow (less than 1MB/s).
> Is this an indication of a faulty drive?

Probably no, (especially because as you wrote SMART info shows no errors) "just" the SSD firmware may not tolerate the long/stressful operation and simply stopped responding for further commands, all following sectors show errors.
Usually this happens on overheat - but according the bottom of the image, the temperature was low (I see 34 C).


> Screenshots are below. What should my next steps be? Should I be concerned that the drive is failing?

No, probably not failing - just does not tolerate well really intensive disk operations (writes).


> Should I reinitialise the disk? Is there any point letting this current test complete? It's taking a long time...

No need to wait any longer, this test will probably not give better results, I'd surely cancel this test.

The Reinitialize disk surface is generally a good idea too in this case, but I worry that probably you'll somehow similar results, just maybe at different point.

Instead, personally I'd try the Disk menu -> Surface test -> Write+Read test.
This is a "light" version of the Reinitialize Disk Surface test (just performs a single pass overwrite and reads back for verification).
This re-writes all sectors, so generally refreshes the SSD. Ideally then all blocks will be green, should be no yellow or red blocks (darker green blocks are normal).

[abstract6917]

Thanks for your quick and detailed response!

I attempted to run the Write+Read test. At one point the window became unresponsive (greyed out, "loading" mouse pointer etc.) and, stupidly, I became impatient. I killed the task in task manager and now I think I bricked the drive. The PC hangs on boot if I have it connected to an M.2 slot. I have tried it in an external USB-C enclosure but the drive is still not detected. Serves me right for being impatient!

The info you provided is useful info for sure and will come in handy in the future. I will be more cautious when running these tests again.

[hdsentinel]

I can confirm that you did not make anything wrong. What you did is completely normal, probably everybody (including me) would do the same.

> At one point the window became unresponsive (greyed out, "loading" mouse pointer etc.)

This happens when the drive stops responding at all, "freezes" completely: it no longer accepts commands, stops processing anything (read / write).
Then any application (the disk test in this case) which attempts to use the drive seems "not responding" - because it is also waiting for the drive (which is not responding).

Probably slightly similar happened during the previous test too, that's why the SSD stopped then too. Your feelings that the SSD has problem (even if it seems correct) were right.

In similar situations, as we can't do anything else, yes, it is normal reaction that we try to cancel the test or even kill the task if required (as the drive stops working) or even try to shut down / restart.

In some cases yes, it is possible that the drive (after its firmware recovers) starts working again. So if you see it does not work when you try to boot, just allow some more time (even a hour or so) - not sure, but maybe it will start again later.

But if it does not start at all, then I'm afraid it may be failed. Please note that you did not make anything wrong and we can say it is better to fail during the disk testing - than failing when it would be filled with lots of data.
Generally the disk test did not make anything else just performed write (and then read) of the appropriate sectors, so probably you'd encounter similar if you'd simply begin to copy 2 TB data from a different SSD to this one.

I'd be just curious about the status of the drive, so a developer report (Report menu -> Send test report to developer option - when the drive was still working) could help lots....