Question Brand new SSD's health dropped to 94% in a month

[kornsoud]

Hello! Should I panic?

Spoiler: CrystalDiskInfo

April-29-2024
Health Status : Good (99 %)
Power On Hours : 112 hours
Power On Count : 109 count
Host Reads : 137 GB
Host Writes : 481 GB

May-17-2024
Health Status : Good (97 %)
Power On Hours : 395 hours
Power On Count : 151 count
Host Reads : 588 GB
Host Writes : 1398 GB

May-22-2024
Health Status : Good (96 %)
Power On Hours : 478 hours
Power On Count : 156 count
Host Reads : 1339 GB
Host Writes : 1895 GB

May-31-2024
Health Status : Good (95 %)
Power On Hours : 615 hours
Power On Count : 175 count
Host Reads : 3295 GB
Host Writes : 2554 GB

June-06-2024
Health Status : Good (94 %)
Power On Hours : 701 hours
Power On Count : 181 count
Host Reads : 3426 GB
Host Writes : 2691 GB

 

[Lafong]

2.2 terabytes written in 5 weeks.

At that rate, 23 per year.

More than most;

I'd look into it, but not sure it would rise to the level of "panic".

I think that is an Axiom drive with TBW of 400.

 

[kornsoud]

2.2 terabytes written in 5 weeks.

At that rate, 23 per year.

More than most;

I'd look into it, but not sure it would rise to the level of "panic".

I think that is an Axiom drive with TBW of 400.

Edge Memory 1400TBW. Also CrystalDiskMark test consumes 400 GB of Total Host Writes.

 

[USAFRet]

Also CrystalDiskMark test consumes 400 GB of Total Host Writes.

Not from what I see.

This is a 1TB SanDisk SATA III drive.
Before and after a full CDM test.

 

[kornsoud]

Not from what I see.

This is a 1TB SanDisk SATA III drive.
Before and after a full CDM test.

What settings? I have NVMe SSD in settings (not Default), Test Count: 5, Test Size: 1GiB.

 

[USAFRet]

What settings? I have NVMe SSD in settings (not Default), Test Count: 5, Test Size: 1GiB.

I think a lot of that has to do with C drive or not.

CDM 8.0.4, same settings as you.

NVMe #1, C drive - 275 Total Host Writes
NVMe #2, H drive - 111 Total Host Writes
SATA III SSD, E drive - 43 Total Host Writes

 

[USAFRet]

In any case....

The first month or two, a drive incures a LOT more writes than later.
That curve levels out.

So far, your use is 2691GB in 2.5 months.
Maintaining that current use, extrapolating out....
Approx 1TB/month.

Your drive has a warranty of 1400TBW?
That means 1400 months.
You will go past that TBW number sometime in mid 2140.

 

[kornsoud]

You will go past that TBW number sometime in mid 2140.

Is that possible to update the firmware despite of non existence of official update tool (edge memory doesn't have one)? I found some *.bin files for the controller (PS5018-E18) I have on my SSD. Should I update the firmware?

 

[USAFRet]

Is that possible to update the firmware despite of non existence of official update tool (edge memory doesn't have one)? I found some *.bin files for the controller (PS5018-E18) I have on my SSD. Should I update the firmware?

I wouldn't worry about it.

 

[fzabkar]

The Percentage Used raw value will hit 100% after about 45 TBW.

2691 GB / 6 x 100 = 44850 GB = 45 TB​

​

One wonders whether the performance will be throttled when that happens.

 

[kornsoud]

will hit 100% after about 45 TBW.

Should be 1400 TBW: https://us.edgememory.com/Home/inde...dge-nextgen4-pro-m-2-pcie-solid-state-drive-2

I'm wondering, is it because of the overvoltage, broken firmware, defective elements or something with windows? I use fresh windows 2021 LTSC, never updated. The only way I'm thinking about is firmware updating with a .bin file I found on the internet. But I don't know is it safe to use the same firmware for different SSDs with the same controller (Phison E18 in my case)?

 

[fzabkar]

It could be that the write amplification is very high. That is, the SSD may be writing much more data to NAND than is being written by the host.

 

[kornsoud]

It could be that the write amplification is very high. That is, the SSD may be writing much more data to NAND than is being written by the host.

Meaning I have not Host Writes : 2691 GB but 10 times more? I can't see Total NAND Writes...

 

[kornsoud]

New info.

I installed HDSentinel (last stable version) and run Short Self test on SSD. In the middle of the test I got bluescreen WHEA_UNCORRECTABLE_ERROR. Writing data about bluescreen was stuck at 0%. After that the computer rebooted itself and went right into bios, no SSD here, not detected. I rebooted the PC once and still no SSD in bios. After that I turned off the PSU and waited 5 minutes. After that the windows has been loaded successfully but no bluescreen dump (BlueScreenView).

I backed up the whole system (important files has been already backed up). What was that? SMART data for my SSD is still the same:

Spoiler: SMART

Some facts:
-OS: Windows 10 2021 LTSC (I used Optimize-Offline to "optimize" an original iso file)
-I use an old Firefox 56 browser with a lot of extensions and userscripts
-Motherboard, SSD and CPU was under overvoltage (6-pin PCIE cable to 8-pin CPU slot, with an old PSU)

My guess is something is broken on SSD or motherboard. Also I want to reinstall vanilla Windows 10 2021 LTSC, not optimized by Optimize-Offline.

What do you suggest?

 

[USAFRet]

What do you suggest?

Use it until it actually breaks.
Maintain a good backup routine in the meantime.

 

[kornsoud]

Use it until it actually breaks.
Maintain a good backup routine in the meantime.

Should I run Short Self test again? What made the SSD to disappear, what this Short Self does? Also yesterday I removed the SSD from M.2 slot and cleaned the contacts with isopropanol (they were clean though). I made sure that it sits tight in the slot.

 

[USAFRet]

Should I run Short Self test again? What made the SSD to disappear, what this Short Self does? Also yesterday I removed the SSD from M.2 slot and cleaned the contacts with isopropanol (they were clean though). I made sure that it sits tight in the slot.

If it works right now, leave it alone.

 

[NickyB]

To me, that is NOT normal. I have a 990 pro that I installed win11 onto and have done a <Mod Edit> ton of game installs and data. I am at 100% after 3 months. To me something is not right there.

 

[kornsoud]

To me, that is NOT normal. I have a 990 pro that I installed win11 onto and have done a <Mod Edit> ton of game installs and data. I am at 100% after 3 months. To me something is not right there.

Yeah, but I don't know what to do next. I want to update the firmware but I don't know how. Also it could be dangerous to use some non verified .bin files, the only source I found is some russian website: https://www.usbdev.ru/files/phison/ps5018fw/

 

[MWink64]

Should be 1400 TBW: https://us.edgememory.com/Home/inde...dge-nextgen4-pro-m-2-pcie-solid-state-drive-2

I'm wondering, is it because of the overvoltage, broken firmware, defective elements or something with windows? I use fresh windows 2021 LTSC, never updated. The only way I'm thinking about is firmware updating with a .bin file I found on the internet. But I don't know is it safe to use the same firmware for different SSDs with the same controller (Phison E18 in my case)?

I would not play around with random firmware.

I'm not familiar with Edge Memory and that site does not inspire confidence. There's a Silicon Motion logo, despite the drive sporting a Phison controller. Also, one spot says three year warranty and another says five year.

Your drive does appear to be burning through endurance quickly but isn't showing as using any spare blocks. What kind of data are you writing to it and how full are you keeping it? As someone else mentioned, it could be a very high Write Amplification but it would have to be obscenely high to degrade this quickly. It's possible there's something else going on that we can't see. I'm not sure what to tell you.

Edge Memory 1400TBW. Also CrystalDiskMark test consumes 400 GB of Total Host Writes.

Not from what I see.

This is a 1TB SanDisk SATA III drive.
Before and after a full CDM test.

That's because you tested a SATA drive. The faster a drive absorbs writes, the more CrystalDiskMark will write to it, assuming you're using the same settings.

 

[USAFRet]

That's because you tested a SATA drive. The faster a drive absorbs writes, the more CrystalDiskMark will write to it, assuming you're using the same settings.

And later, above, I also did 2 different NVMe drives.

 

[kornsoud]

What kind of data are you writing to it and how full are you keeping it?

Just some staff from failed HDD (1tb) and SSD (300Gb). I run CrystalDiskMark 3-4 times (400 GB each) so most of the GBs are from it. It had 65% used space for a while, now it's only 20%.

it could be a very high Write Amplification

I'm looking for a way to check NAND Writes.

one spot says three year warranty and another says five year

Yeah, Edge Memory is not the best brand, no official firmware update tool, nothing. I got it for free, so, I want to fix it if I can.

And later, above, I also did 2 different NVMe drives.

It could be slower than mine, maybe: Up to 7175MB/s (read) / 6800MB/s (write)

 

[kornsoud]

Should I try another slot? I use a CPU slot (the only one). I have 3 chipset slots.

 

[MWink64]

Just some staff from failed HDD (1tb) and SSD (300Gb). I run CrystalDiskMark 3-4 times (400 GB each) so most of the GBs are from it. It had 65% used space for a while, now it's only 20%.

I'm looking for a way to check NAND Writes.

Yeah, Edge Memory is not the best brand, no official firmware update tool, nothing. I got it for free, so, I want to fix it if I can.

Unfortunately, I don't believe any of the SMART attributes reported by that drive convey NAND writes. This will probably be of limited practical value but you could try running the appropriate (Phison) Flash ID on the drive, so we can get a bit more information about it.

This is a very long shot and probably not worth the trouble but you could try performing a Secure Erase on the drive and starting over. I wouldn't expect this to make a difference but there's a small chance it could wipe out some gremlins.