SSD question

Had an idle thought- on an HDD, if the size of a file isn't an exact multiple of the sector size then there's some unusable space in the final sector. Does the same apply to an SSD?

Put another way, are SDDs "low-level formatted" in the same way as HDDs?

It's of no relevance to any problem, I'm just curious.

If I had to guess, it would be that SSD sectors are virtual (firmware?), rather than physical, to work with PC architecture. Pretty sure I had to format my SSDs on installation, but no idea of what the setup was. Given the speed of SSDs, it would make sense to have clusters / sectors as small as possible and waste the least space.

All in my head of course. No idea of the real world.

In reply to a post by billford:

Had an idle thought- on an HDD, if the size of a file isn't an exact multiple of the sector size then there's some unusable space in the final sector. Does the same apply to an SSD?

Yes, think of that the same.

Put another way, are SDDs "low-level formatted" in the same way as HDDs?

no, and modern HDDs are much more complex than the old ST506 drives we had in the 80s. You can't low level format a modern drive, and if you could it would destroy it.

In reply to a post by camieabz:

If I had to guess, it would be that SSD sectors are virtual (firmware?), rather than physical, to work with PC architecture. Pretty sure I had to format my SSDs on installation, but no idea of what the setup was. Given the speed of SSDs, it would make sense to have clusters / sectors as small as possible and waste the least space.

Everything on a modern HDD or SSD is virtual, known as "Logical Block Addressing". The operating system makes a request for an LBA and gets the data returned. It has no connection to any physical arrangement. (for those that remember ST506 drives and reformatting to adjust the interleave factor).

The format you performed was at the logical level, for your operating system, not at the physical (LBA) level.

In reply to a post by jchamier:

Yes, think of that the same.

Thanks. I'm all too well aware that my knowledge of modern bulk storage techniques is somewhat less than up-to-date, so a nice simple answer is just what I needed

FAT was so much easier to understand

SSDs are divided into "blocks" rather than sectors. There would be no need to low-level format them as the purpose of this was to physically divide a hard-disk platter into logical sectors; by their construction this concept just doesn't exist for an SSD.

The sector on older hard disks were 512 bytes; more modern (Advanced Format) ones use 4K sectors. SSD blocks are, generally 4K. The block, or sector, is the smallest unit of information that can be accessed on a disk. So even if you save a file of only 1 byte the device driver writes 512 (or 4K) bytes to the device. This is further complicated by the fact that file systems use a minimum allocation size (called clusters by Windows, blocks by some other OSs) that is a multiple (normally more than x1) of the disks sector size.

Whatever storage device you use - hard disk or SSD - all modern file systems will waste any spare space in the last (OS) cluster or block allocated to the file. This is the minimum size that they can write/read on the disk. For all practical purposes you can take the size of this cluster to be 4K on a modern Windows PC.

It would, in theory, be possible to design a file system that used every last bit on a disk. No filesystem that I know is designed this way; I think the housekeeping involved would slow the disk down to an unacceptable degree. Nowadays, as far as disks are involved, speed is paramount; efficient use of space is just not necessary as the cost per GB is so low.

Yes, I should have used "sector/cluster" rather than just "sector".

In reply to a post by TinyMongomery:

Nowadays, as far as disks are involved, speed is paramount; efficient use of space is just not necessary as the cost per GB is so low.

A succinct justification, sounds fine to me 👍

In reply to a post by billford:

FAT was so much easier to understand

FAT is at the operating system / logical level, so you can run FAT (FAT12/FAT16/FAT32/exFAT) on any device from HDD, to SSD to SDcard or USB thumb drive. The physical side is quite different on all of those

This is pretty good though there are couple of features of filesystems you can lookup which cover this:

· Tail-merging (also known as tail-packing or block sub allocation): the ability to merge the final partial-blocks of multiple files into a single block

· Inlining: the ability to store very small files directly with the file metadata (e.g. the inode) rather than allocating an actual data block

These can be more relevant under specific circumstances, for example where using larger block sizes or else with a smaller disk (or volume/partition), or where efficient use of space was an overriding factor rather than say speed.

But yes for billford's original question as has been said elsewhere these are properties of the filesystem used rather than of the underlying storage technology such as SSD vs spinning disk.

i never even thought of that.
Good question