How fast do modern SSDs die: Upcoming test
- Thread starter Patrick
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
May have to pause periodically to make sure this is a valid test, as @T_Minus mentions above; otherwise the controller may not have any idle cycles available to it to run maintenance tasks. Although at this time it's probably too late to change the approach, so might as well see how far they get. I suppose they could be slowing down because all the maintenance activity that normally occurs in the background is being forced out into the open.
Also wondering if Samsung will exchange your drives under warranty...
Also wondering if Samsung will exchange your drives under warranty...
At 200TB they are resting every 100TBW for 30 min.May have to pause periodically to make sure this is a valid test, as @T_Minus mentions above; otherwise the controller may not have any idle cycles available to it to run maintenance tasks. Although at this time it's probably too late to change the approach, so might as well see how far they get. I suppose they could be slowing down because all the maintenance activity that normally occurs in the background is being forced out into the open.
Also wondering if Samsung will exchange your drives under warranty...
No worries on the warranty. I cannot even think what I might use something like this for.
let's get a go-fund-me started for this, our money is good in your hands and besides it's for science
unlikely, as that would mean the company ends up losing money. plus the 35tbw is kind of a lower conservative figure so I doubt it will happen before that number.
Maybe if we expand the project significantly but hopefully unnecessary. $110 is fine for something I am personally curious about.
The delta between the 0GB pre-fill and 90GB pre-fill drives has grown.
0GB pre-fill is now at 600GB lifetime writes and 93% health.
90GB pre-fill is now at just under 400GB lifetime writes. 98% health.
Health figures based on the wear leveling count but both have 0 reallocated sectors.
So the empty drive is 50% faster than the partially-filled drive. Do they both get the same IO pattern?
Edit: I realise you're writing the same data, reading the same data etc. What I mean is - are the performance graphs the same shape, with the originally empty drive just double the filled drive? Or does one or the other have significantly more consistent performance?
Last edited:
So that's 10TBW timetime for the zeru prefilled one? pretty weak conservative number.
You're still seeing higher write speeds with the 0GB pre-fill drive, right? If that is case then I'd say this makes sense. This drive can churn through NAND much faster. The other one is slower and can't kill NAND as fast. The slower drive will also last longer, but since speeds are drastically different that comparison would not be fair. Have to judge by host writes.
so the zero filled drive should last basically only for about 50 or so more days at current rate?
Have their speeds equalized or do you wait for the slower drive to catch up? Trying to understand how they can be at the same level of writes if the write speeds are considerably different.
More that I take snapshots at write intervals. So at 1TB, 2TB and etc. They are not the same speed to reach those marks.
Just posted a 2TBW update for the two drives. Sorry it took so long!
Project Kenko 01 Update: Samsung 750 EVOs pass 2TB written
Certainly interesting results. I did add a performance anecdote from when I tried using the third "test setup" drive as a ZIL device. Please do not do that!
Project Kenko 01 Update: Samsung 750 EVOs pass 2TB written
Certainly interesting results. I did add a performance anecdote from when I tried using the third "test setup" drive as a ZIL device. Please do not do that!