Hello. I have this SSD for a year and half. Many apps such as CrystalDisk Info, HDSentinel, HDTune give me warnings about it. It runs much hotter than average of those who use the same SSD. My question is that if I try and see this SSD to RWA, would it be accepted?
If it’s still under warranty, go file a claim. Make sure you backup any files.
Can you post an equivalent screenshot for WD’s Dashboard?
The Media Wearout Indicator is specially coded to report the wear in terms of P/E cycles (attribute 0xAD, 173) and a second factor. This second factor appears to be TBW in the following thread:
https://forums.tomshardware.com/threads/ssd-health.3694631
However, in the present case this second factor appears to reflect the Block Erase Count (SLC) attribute.
MWI = 0x75FE010075FE → 0x7F5E / 0x100 / 0x7F5E → 0x7F 0x5E / 0x01 0x00 / 0x7F 0x5E → 127 94 / 01 00 / 127 94 → 127.94% / 1.00% / 127.94%
This means that the SSD has consumed 1.00% of its rated P/E cycles. The average P/E cycles is 5, so this means that the rated P/E cycles is 500.
The Block Erase Count (SLC) also appears to consist of 3 parts (min / max / average):
0x9E7BCD2FC7E8 → 0x9E7B / 0xCD2F / 0xC7E8 = 40571 / 52527 / 51176
The rated Block Erase Count (SLC) appears to be …
0xC7E8 / (127.94 / 100) = 40000
In short, your Media Wearout Indicator is reporting that your drive has already exceeded its rated Block Erase Count for its SLC cache (127.94%). This could be a firmware bug, or it could reflect some highly unusual usage pattern.
Here are two more SSDs with the same firmware version:
https://issue-tracker.miraheze.org/T9839
SSD #1
Block Erase Count (SLC) = 700091924664 → 0x00A300BB00B8 → 0x00A3 / 0x00BB / 0x00B8
Media Wearout Indicator = 197571117102 → 0x002E0028002E → 0.46% / 0.40% / 0.46%
Average P/E cycles = 2
SSD #2
Block Erase Count (SLC) = 330731225368 → 0x004D011E0118 → 0x00A4D / 0x011E / 0x0118
Media Wearout Indicator = 300652953670 → 0x004600500046 → 0.70% / 0.80% / 0.70%
Average P/E cycles = 4
The rated P/E cycles is once again 500.
The rated Block Erase Count for SLC cache is …
0x00B8 / 0.46% = 40000
… and …
0x0118 / 0.70% = 40000
Perhaps you could check if a firmware update is available?
1 Like
Hi. Thanks for your time to write all this, I appreciate it.
I have upgraded my SSD’s firmware an hour ago or so after seeing your message.
Positive: the health status of SSD has jumped to 31% from a simple “Bad” on CrystalDiskInfo. It shows 31% on the WD Dashboard app too. And it runs about 10 degree celsius cooler than before.
Negative: now the SSD’s performance is worse than my old HDD which runs at 5400 rpm. It’s really lagging so I’m even afraid to do a proper performance test through apps etc.
How do I fix this performance issue now?
Can you show us the new SMART attributes (both top and bottom halves of CDI’s screenshots)?
As for the performance issues, you could scan the drive using HDDScan or Victoria for Windows. This should find any “slow” blocks.
Edit:
Your first screenshot shows a Total Host Writes figure of 3136 GB. If we assume that this represents 31% of the TBW rating, then the drive is rated for 10 TBW. This seems far too low, so I could be completely wrong.
By scan the drive, you mean scan for “bad sectors”? I could do that. According to HDTune, the drive has 115 damaged blocks already (S.M.A.R.T. value).
The Reassigned Block Count and Grown Bad Block Count are both at 0x73 (= 115). These are probably responsible for the slowdown. Essentially, your SSD has started to degrade. I wouldn’t trust it with your data.
As for the Media Wear Indicator, WD has now redefined the values.
MWI = 0x450F0021450F → 0x45 0x0F / 0x00 0x21 / 0x45 0x0F → 69.15% / 0.33% / 69.15%
This means that the rated P/E cycle count has increased from 500 to 1500.
The other number (69.15%) appears to be showing the remaining life. I’m not sure what this calculation is based on.
As for scanning for bad sectors, I mean that the software will read every sector on the drive and report if there are any problematic sectors. Those 115 reallocated sectors have already been identifed and replaced by the drive.
115 reallocated sectors is high for a 480 GB SSD? How much more can it replace? Is there a fixed allocated number for it?
It’s scanning for bad sectors now. After reading and reporting bad sectors, the drive will automatically further reallocate them? Or is it something I need to do?
Lastly, what are the chances of getting replaced via RMA?
I don’t know WD’s criteria for RMAs.
IMO any number of reallocated sectors is too high for an SSD that has only used 5 of its 500 or 1500 rated P/E cycles. You should only start to see these grown defects towards the end of the SSD’s life.
As for reallocating bad sectors, the drive may reallocate a bad sector if it can read it successfully at least once. Otherwise, it needs to wait for the OS to write new data to that sector.
In any case, I find it suspicious that WD has played with the MWI attribute. Are they hiding something? Models like the SA510 have been plagued with problems. I wonder if your model is related?
I’m not sure if I did it right but this is the result.
I don’t know what they’ve done with this SSD but I wouldn’t have bought it if I knew these Green models are being made by Sandisk.
I’m ready to let this SSD go but I’m not sure if WD would let me.
There is a reason I didn’t recommend HD Tune. It is a tool that records a pass or fail for each read. That’s not what you want.
The tools I recommended will record the time taken to read each sector. If a sector is on the verge of going bad, it will require error recovery and one or more rereads. These “weak” sectors will be separately reported by the tool. HD Tune doesn’t understand the difference between a weak sector and a good sector.
1 Like
You still don’t understand.
This blog applies to a HDD, but it should also apply to SSDs.
https://www.technibble.com/hddscan-scan-for-bad-sectors/
https://www.technibble.com/articlecontent/2014/04/HDDScan.jpg
I gave up. I shouldn’t have to deal with this ■■■■. It’s lagging like crazy. I will send it for RMA and be happy with my 5400 RPM, healthy ancient drive instead.
