Why Is My SSD Slower Than When It Was New?
Short answer
Check free space first: an SSD above roughly 80% full loses a large part of its write speed, because the fast SLC cache shrinks and the controller has to erase blocks before writing. Free up space to leave 15–20% empty and most of the lost performance returns. If the drive is not full, the remaining causes are heat throttling and drive health, both of which are measurable.
On this page
SSDs do slow down, and three of the reasons are real and measurable. None of them is the vague “wear” that gets blamed in forum threads, and in a large majority of cases the answer is the simplest one: the drive is too full.
Worth separating one thing out first. If the whole machine feels slower and you are assuming storage, confirm storage is actually the bottleneck before going any further — Task Manager’s Performance tab will tell you in thirty seconds, and finding what is using 100% CPU or disk covers reading it properly. A machine short of RAM presents almost exactly like a slow drive, because it is paging.
Check free space first
Open Settings → System → Storage, or just look at the drive in File Explorer.
| Free space | Expected effect |
|---|---|
| Above 25% | None |
| 10–20% | Writes noticeably slower, occasional stutter |
| Under 10% | Writes sharply slower, system feels laggy |
The mechanism is worth understanding, because it explains why the effect is so abrupt.
Flash memory cannot overwrite in place. A cell must be erased before it is written, and erasure happens in large blocks while writing happens in much smaller pages. The controller handles this by keeping a pool of pre-erased blocks ready. On a drive with plenty of free space, writes go straight into that pool at full speed.
On a nearly full drive, the pool is small. Every write turns into a read-modify-erase-write cycle across a block the controller first has to assemble. That is several times slower, and it is why the drop feels like falling off a cliff rather than a gentle decline.
The second half of the same problem is the SLC cache. Consumer drives write a region of flash one bit per cell — fast — then consolidate it into denser three- or four-bit storage in the background. That cache is carved out of free space. Fill the drive and the cache shrinks, sometimes to nothing, and sustained write speed falls to the drive’s native rate.
The fix is to free space and leave it free. 15–20% empty is the target. If space keeps disappearing, a C drive that fills up again after cleaning covers finding what is writing continuously, and freeing up disk space on Windows covers what is safe to remove.
Confirm TRIM is working
TRIM tells the drive which blocks no longer hold useful data, so the controller can erase them in advance rather than during your write. Without it, the pre-erased pool runs dry and performance degrades permanently, even with free space showing.
Windows enables it automatically on SSDs. To verify, open Terminal (Admin):
fsutil behavior query DisableDeleteNotifyA result of 0 means TRIM is enabled. 1 means it is disabled, which is almost
always the work of an old tuning guide or a “performance optimiser” utility.
Windows also runs a scheduled retrim as part of drive optimisation. Search Start for Defragment and Optimise Drives. An SSD is listed with media type Solid state drive, and the operation it performs is TRIM, not defragmenting — the shared button label confuses people into thinking Windows is defragmenting their SSD. It is not. Leave the weekly schedule enabled.
Check for heat throttling
NVMe drives, particularly in thin laptops and in motherboard slots tucked under a graphics card, get hot. Above roughly 70°C most controllers reduce speed to protect the flash, and above 80°C they reduce it sharply.
The symptom is specific and recognisable: the drive is fast for the first ten or twenty seconds of a large file copy, then falls off and stays slow until it cools. A full drive produces consistent slowness; heat produces slowness that arrives partway through.
Read the temperature with CrystalDiskInfo, which reports it from the drive’s own SMART data. Idle should sit in the 30s or 40s; under sustained load, up to around 70°C is acceptable.
If it throttles:
- Fit the heatsink that came with the motherboard, or an inexpensive one. On a desktop this routinely drops temperatures by 15–20°C.
- Move the drive to an M.2 slot away from the graphics card, if the board has one with equivalent lanes.
- On a laptop, make sure the intake vents are not blocked, which is also the most common cause of a laptop fan running constantly.
Check the drive’s health
Flash cells wear out, but consumer workloads rarely approach the limit. Check rather than assume.
CrystalDiskInfo reports health as a percentage, along with Total Host Writes and the Percentage Used attribute on NVMe drives. A typical 1TB consumer drive is rated for several hundred terabytes written. Most home machines write 10–30TB over five years.
What matters more than the percentage is the trend in reallocated or uncorrectable sector counts. A health figure dropping steadily month to month, or a sudden appearance of errors, means replace the drive — and SSDs tend to fail by abruptly becoming read-only rather than by degrading gracefully, so there is less warning than with a mechanical drive.
The less common causes
Firmware. Several drive models shipped with firmware bugs that caused performance degradation or premature wear, fixed in later releases. Check the manufacturer’s utility — Samsung Magician, WD Dashboard, Crucial Storage Executive. Back up before flashing firmware; it is a brief but genuinely risky operation.
The wrong slot or mode. An NVMe drive in a slot wired for fewer PCIe lanes, or a SATA drive on a controller running in IDE compatibility mode, runs at a fraction of its potential. This is usually a problem from day one rather than a degradation, but it surfaces after a motherboard BIOS reset.
Background writes. A drive that is busy is a drive that is slow for you. Windows Search indexing, OneDrive syncing, a backup running, or an antivirus scan all compete for the same controller. Task Manager’s Processes tab with the Disk column sorted highest first names the process.
Power settings. Windows’ aggressive link power management can add latency to NVMe drives on battery. Settings → System → Power & battery → Power mode set to Best performance while plugged in is the simplest check.
What does not help
- Defragmenting. No benefit on an SSD, consumes write endurance.
- “SSD optimiser” utilities. They toggle TRIM and prefetch settings Windows already configures correctly, and some disable things that should be on.
- Disabling the page file to reduce writes. The write volume is trivial relative to endurance, and removing it causes out-of-memory failures.
- Secure Erase as a routine fix. It genuinely restores factory performance and it destroys all data on the drive. Reserve it for a drive you are about to reuse.
- Replacing the drive before checking free space and temperature. Most “worn out” SSDs are 95% full.
Realistic expectations
Freeing a full drive back to 20% free restores most of the lost write performance within a day, once the controller has had idle time to erase blocks in the background. Read speeds are affected far less by fullness, which is why a drive can feel fine opening files and terrible saving them.
Fitting a heatsink to a throttling NVMe drive removes the mid-transfer slowdown entirely and is the cheapest fix in this guide.
What will not come back is the box figure. Those numbers are sequential transfers on an empty drive with a cold controller, and ordinary use is mostly small random operations where real throughput is far lower by design. A drive at a third of its advertised speed during everyday work is behaving normally.