I have a newish PC with a 960 GB SSD as hard disk no.1. Considering there's over 90% of free space on it, the total defragmentation indicated (4%) seems to suggest that the smallish amount of stuff on there is quite fragmented.
For years I assumed that you don't defrag an SSD, for two reasons: one, it creates unncessary writing to the disk and two, the software can "see" all the disk at once anyway, so the concept of physical defragmentation does not exist, unlike with a HDD.
However, I have been reading that this is not entirely true, and that an infrequent defrag (say, once every month or two) is a good idea. Modern SSDs can cope with more write processes than used to be the case and there is an improvement in performance.
I've been reading stuff about NAND and TRIM and don't really understand any of it, so I'd welcome guidance from the experts among you, if you would be so kind.
Last edited by: Observer on Tue 1 Sep 15 at 10:53
|
ssd drives no need to defrag, as it will reduce the life span of the drive,
|
SSDs and lifespan is a issue from that people wont let go of for no good reason. Its folklore
There is no point in deframenting an SSD drive. it does not get fragmented in the accepted sense.
|
And even if it did - I have a vague recollection tat they aren't written to sequentially like HDDs - why would you add write cycles unless speed was a problem?
|
In theory it doesn't fragment as all the blocks are accessible at the same speed.
I defrag mine using MyDefrag, which has an SSD specific function, about every 6 months to release some space which gets locked up and not accessible when deleting old restore points for example.
I can gain >1GB on a 64GB drive which is currently at 80% capacity.
Must look into getting a bigger drive or clean up what I've got.
|
To improve the life of an SSD, the drives use wear levelling algorithms to move data around.
Also, the SSD will appear to the operating system in a similar way to a hard drive with Logical Block Addresses (LBAs). These LBAs are mapped onto physical flash memory. So a contiguous set of LBAs might actual map onto any part of the physical memory.
|