Why Does a Full SSD Get Slow? SLC Cache and Garbage Collection
The Cause Is Not Fragmentation, It Is How Writing Works
Short answer: a full SSD slows down for two reasons. First, the drive's fast write area — the SLC cache — is proportional to free space, so it shrinks or disappears entirely as the disk fills. Second, garbage collection, which clears out deleted blocks, can no longer run in the background and has to work while you are writing.
This has nothing to do with fragmentation on hard drives — and defragmenting an SSD does not help; it creates pointless writes and shortens the drive's life. Read speeds are barely affected; what collapses is writing. This guide explains the mechanism and how much free space you actually need to leave.
Key Takeaways
- SSD slowdown is not caused by fragmentation; defragmenting an SSD does not help and creates needless writes.
- Modern SSDs run some TLC/QLC cells temporarily in single-bit (SLC) mode to form a fast write cache, and the size of that cache is PROPORTIONAL TO FREE SPACE.
- Once the cache is exhausted, write speed drops to the cell's native rate: several times slower on TLC, up to ten times slower on QLC.
- NAND cannot be overwritten; a block must be erased before it can be written. With no clean blocks left, garbage collection runs in the middle of your writes and adds latency.
- Read performance barely changes with fullness. It is writing that slows down.
- Practical rule: keep at least 10-15% of capacity free. On QLC drives, 20% is safer.
The SLC Cache: Why the First 50 GB Is Fast and the Rest Is Not
Most modern consumer SSDs use TLC (3 bits per cell) or QLC (4 bits per cell) memory. The more bits packed into a cell, the slower and more complex writing becomes. Manufacturers work around this by temporarily running some of the drive's empty cells in single-bit (SLC) mode and writing incoming data there first. That region is the SLC cache, and it is very fast.
The critical point is this: on most drives that cache is not fixed but dynamic. Its size is proportional to free space. On a half-empty 1 TB drive the cache can be tens of gigabytes; at 95% full, the same drive drops to a couple of gigabytes or loses the cache entirely.
In daily use you see it like this: copying a large file starts very fast, then suddenly drops and stays low. That is the moment the cache filled and writing fell back to the cell's native rate. On a full disk that fast start never happens — the drive is in slow mode from the first byte.
Garbage Collection and Write Amplification
The fundamental constraint of NAND is that you cannot overwrite data in place. Writing happens in pages, but erasing happens in much larger blocks. When you delete a file, the drive erases nothing at that moment; it only marks the relevant pages as invalid.
The real cleanup happens later, during garbage collection: the controller moves the still-valid pages out of a block, then erases the whole block so it can be reused. While the drive is idle this happens in the background and you never notice it.
On a full disk there are no clean blocks left. The controller has to move and erase at the same time as you are writing. This is write amplification: to store 1 MB of your data, the drive may shuffle several megabytes internally. Latency goes up and the drive wears faster.
How Much Free Space, and Why QLC Needs More
The practical rule is to keep at least 10-15% of capacity free. That figure is not arbitrary: it is the working room the controller needs to do garbage collection in the background and to keep the SLC cache at a meaningful size. Free space you leave as a user effectively acts as extra over-provisioning.
On QLC drives it pays to be more generous. A QLC cell holds four bits, so its native write speed is low and it depends more heavily on the SLC cache. When the cache runs out, the drop on QLC is far sharper than on TLC; on some models write speed can fall below that of a hard drive. Targeting 20% free on such a drive is sensible.
Then there is TRIM. TRIM is how the operating system tells the drive which blocks belong to deleted files; without it the controller cannot know which blocks are genuinely free. But even with TRIM enabled, its benefit is limited if there are no blocks to clean — TRIM does not replace free space, it works together with it.
What to Do When It Is Already Full
The only thing that genuinely works is freeing space; no setting or tool can change this physics. Make the target concrete: 15% of capacity. On a 500 GB drive that is 75 GB free; on a 1 TB drive, 150 GB.
The quickest wins usually come from four places: system and application caches, large files you have not opened in months, byte-identical copies scattered across folders, and old installers in the Downloads folder. Together those hold 30-60 GB on a typical machine, and none of it is personal data.
After freeing space, give the drive some idle time. The controller needs to work through its backlog of garbage collection; leave the computer on and avoid starting a huge copy right away, and you will see the old speed return. Disk Mop's Cache Cleaner, Large Files, Duplicates and Downloads modules target exactly those four categories, showing what will be deleted before anything happens, while the Disk Health module reports TRIM status and S.M.A.R.T. values on the same screen.
Frequently Asked Questions
Does defragmenting an SSD help?
No. Fragmentation is a problem on hard drives because of the physical movement of the read head; an SSD has no such movement. Defragmenting only creates needless writes and shortens life. Windows's optimize tool already runs TRIM rather than defragmentation on SSDs.
Does speed come back immediately after freeing space?
Usually not instantly, but shortly after. The controller has to finish its backlog of garbage collection, and it does that while idle. Leaving the computer on for a while is enough in most cases.
Do read speeds drop too?
Only slightly. Fullness mainly affects writing. If reading feels noticeably slower, the cause is probably elsewhere: not enough system memory, a background scan, or a drive that is genuinely worn.
Does creating a partition and leaving it empty help?
Yes, this is known as manual over-provisioning and it works — but it gives the same result as simply keeping the disk 15% free. The only advantage of a separate partition is that it stops you from filling that space by accident.
Verdict
A full SSD slowing down is not a fault, it is a natural consequence of the design: the fast write cache depends on free space, and garbage collection needs room to breathe. The fix fits in one sentence — keep 15% of capacity free, and aim for 20% on QLC drives.
The easiest way to free that space is to clear accumulated junk without touching your personal files. Disk Mop's Cache Cleaner, Large Files, Duplicates and Downloads modules do exactly that, and the Disk Health module shows TRIM status and the drive's wear indicators on one screen.
See what is really using space on your SSD
One-time payment for lifetime access to all features. No subscription, no hidden fees.
Download Disk Mop Now