[SOLVED] Does the free space for optimal SSD speed need to be allocated ?

nofanneeded

Respectable
Sep 29, 2019
1,541
248
2,090
Hello,

I understand that if the SSD is near full it slows down , and it is recommended to leave 20% of it empty all the time .

My Question is :

Does this free space need to be allocated ?

Because it would be better not to allocate 20% of the size to prevent users from filling it up no matter how.
 
Solution
It does not matter either way.

Allocated/Unallocated/partitions...don't really matter.
What you might see in Disk Management is only a visual representation. The drive firmware shuffles data around to ALL cells, as it sees fit.
It does not matter either way.

Allocated/Unallocated/partitions...don't really matter.
What you might see in Disk Management is only a visual representation. The drive firmware shuffles data around to ALL cells, as it sees fit.
 
Solution
It does not matter either way.

Allocated/Unallocated/partitions...don't really matter.
What you might see in Disk Management is only a visual representation. The drive firmware shuffles data around to ALL cells, as it sees fit.

Cool , then I will just leave 20% not allocated so no one by accident fills the SSD and it slows down !

Will tell customers about it as well.
 
If this is the case then SSD makers should add that 20% size and not include it in total SSD capacity.
That's what a lot of them do, not 20% because you don't need that much but a lot of drives come already overprovisioned and leaving even more space free on top of that won't do anything.

Also the ssd firmware doesn't make a difference between allocated or not when it comes to wear leveling so lifespan doesn't change either way.
 
Most ssd makes already do include more than you can use, they leave the drive space for error correction even if you don't leave that 20%. There will be memory on the sticks you can never use.

Do ssd makers call 1tb 1000gb like hdd makers do or do they give users the extra 24gb that should be on the sticks? as bigger sticks get the more they can keep to themselves.
 
no, they use better nand such as emlc and slc, whereas the drives you mostly get now are 3dnand or similar
the less levels, the more writes they can do.

.

most the drives we use now can sustain more writes than we are likely to do to them, before you replace them with bigger, so its not a major concern.
 
Last edited:
If this is the case then SSD makers should add that 20% size and not include it in total SSD capacity.
Most of them do. If the SSD is not advertised as having a capacity in a power of 2 (e.g., 128, 256, 512, etc), overprovisioning is already there at the firmware level. The "missing" capacity is the overprovision.

makes me wonder , are Enterprise SSDs with very high TBW Just an oversized capacity and labeled at 1/4th the capacity to give us more TBW ?

Like a 1TB with some 4000 TBW are in real 4TB with 1000 x4 TBW ?
Yes, for the most part enterprise SSDs have more spare flash memory for higher write endurance, and probably some additional controller magic for the use cases they're expected to run in.
 
  • Like
Reactions: nofanneeded
And that TBW endurance is the warranty number. Not the point when it will die from too many writes
(we've discussed this before)

These drives, even a 250GB 840 EVO from 6 years ago, have been tested to last loooong beyond that warranty number of write cycles. Even up to the petabyte range.

Lastly, outside of hard use in an enterprise situation, like a data center or database server, that TBW is decades away.
 
And that TBW endurance is the warranty number. Not the point when it will die from too many writes
(we've discussed this before)

These drives, even a 250GB 840 EVO from 6 years ago, have been tested to last loooong beyond that warranty number of write cycles. Even up to the petabyte range.

Lastly, outside of hard use in an enterprise situation, like a data center or database server, that TBW is decades away.

true but this is not what we are talking about here . we are talking about How enterprise TBW is more than eight times the consumer TBW while having the same TLC chips , which can only results from using larger capacity and label it lower.
 
Might also be better binned/quality chips.
Tighter quality control.

Note the use of the word "might".

TLC 3D NAND write cycles range from 1500-3000 , so you are limited to Double the TBW using the best Quality chips around ...

we are seeing more than 8x the TBW using the same TLC , and I know how to use the word "might" .. and there is no way outside using bigger capacity to get 8 times the endurance .

Even Firmware will not change the Write cycles ! there is a barrier.

comparing the SSD to HDD is false ...
 
Last edited:
TLC 3D NAND write cycles range from 1500-3000 , so you are limited to Double the TBW using the best Quality chips around ...

we are seeing more than 8x the TBW using the same TLC , and I know how to use the word "might" .. and there is no way outside using bigger capacity to get 8 times the endurance .
Again, that is the published warranty endurance number.
What the manufacturer wants to bet the return policy on.

What are the actual wear limits? When do these things actually wear out, regarding TBW? 1,500? 20,000?
Let's look at those numbers.

We can start with these, from several years ago:
http://www.anandtech.com/show/7173/...w-120gb-250gb-500gb-750gb-1tb-models-tested/3
 
Again, that is the published warranty endurance number.
What the manufacturer wants to bet the return policy on.

What are the actual wear limits? When do these things actually wear out, regarding TBW? 1,500? 20,000?
Let's look at those numbers.

We can start with these, from several years ago:
http://www.anandtech.com/show/7173/...w-120gb-250gb-500gb-750gb-1tb-models-tested/3

Show me enterprise SSD test not 840 Evo please . our point is , the Enterprise uses larger size to give More TBW (more than 8 times)
 

TRENDING THREADS