SSD Caching and Why Copy Speeds Drop

Why an SSD's fast write cache exists, why copy speeds drop when it fills, how a fuller drive shrinks the cache, what the native speed is, and how to see it.

File transfer progress, photographed for a technology article.

Copy a large file to an SSD and you may notice something odd: it starts fast, then partway through the speed suddenly drops, sometimes dramatically. This is not a fault or a failing drive, but a designed behaviour of how SSDs manage writing, involving a fast cache that fills up. Understanding this speed cliff explains a common and puzzling experience, and why the drive is working exactly as intended even as the copy slows.

The fast cache that fills

Most consumer SSDs use a portion of their flash as a fast write cache, treating some cells in a faster mode to accept writes quickly, which is why the start of a large copy is fast. But this cache is limited in size, and once it fills during a large write, the drive must write the rest directly to the slower main flash, which is why the speed drops partway through: the fast cache is exhausted and the drive falls back to its slower native write speed. This is the speed cliff, and it is entirely by design, a way to make most writes fast while accepting that very large writes will slow once the cache fills, the cache-behaviour the guide on how SSDs manage writing touches on.

Why the cache exists

The fast cache exists because it makes the drive feel fast for typical use while keeping costs down. Most writes people do are small enough to fit within the cache, so they complete at the fast cached speed, giving the drive its advertised high write performance for everyday use. The slower native speed only appears for writes large enough to exhaust the cache, which are uncommon in normal use, so the cache delivers fast performance where it matters most while the slower fallback affects only the occasional very large transfer. This is a sensible design that gives fast everyday performance affordably, at the cost of a speed drop on large writes, the design-tradeoff the guide on how drives balance speed and cost reflects.

Phase of a large copy Speed
Into the fast cache Fast, the advertised speed
After the cache fills Drops to slower native speed
On a fuller drive Smaller cache, earlier drop

How fill level shrinks the cache

An important detail is that the size of the fast cache depends on how full the drive is, shrinking as the drive fills. On many drives the cache uses free space, so a nearly full drive has a smaller cache and hits the speed cliff sooner, sometimes much sooner, than an empty one. This is why a full SSD can feel notably slower at large writes than the same drive when empty, since the cache that hid the slower native speed has shrunk. It is another reason to keep some free space on an SSD, since a fuller drive not only manages wear less well but also loses more of its fast write cache, exposing the slower native speed earlier, the fill-level effect the guide on what affects a drive real performance reflects.

What the native speed is

The speed the drive drops to after the cache fills is its true native write speed, which varies by drive type and can be much slower than the cached speed, especially on cheaper, denser drives. This native speed is what the drive can sustain indefinitely, so it is the honest measure of sustained write performance, unlike the cached speed which only lasts until the cache fills. This is why sustained write performance, the native speed, matters for anyone regularly writing very large amounts of data, since they will spend time at the slower native speed once the cache is exhausted, while the cached speed suffices for everyone whose writes fit the cache. Knowing the native speed matters for write-heavy uses, the sustained-versus-burst point the guide on judging real performance reflects.

Measuring it yourself

You can see this behaviour yourself by copying a very large file and watching the speed, which will start fast and drop when the cache fills, the drop revealing where the cache ended and the native speed began. This is a normal, expected behaviour, not a sign of a problem, so seeing it should reassure rather than alarm. For most people it is irrelevant, since their writes fit the cache and never hit the cliff, but for those doing large writes, understanding it explains the slowdown and points to the value of a drive with a large cache or a fast native speed. Either way, the speed cliff is the drive working as designed, and recognising it as normal removes the worry that a slowing copy signals a failing drive, the reassurance the guide on understanding normal drive behaviour reflects.

Why it matters, and for whom

For most people, the speed cliff is a curiosity rather than a problem, since their writes fit within the cache and never reach the slower native speed, so the drive always feels fast. Understanding the behaviour matters mainly so that seeing a large copy slow down does not cause needless worry about a failing drive, when it is simply the cache filling as designed. Knowing this turns an alarming-seeming slowdown into an understood, expected behaviour, which is reassurance enough for the majority whose everyday use never seriously tests the drive sustained write speed.

For those who regularly write very large amounts of data, though, the native sustained speed genuinely matters, and understanding the cliff informs their drive choice. Someone frequently copying large video files, disk images, or datasets will spend real time at the slower native speed once the cache fills, so for them a drive with a large cache or a fast native write speed is worth seeking, unlike for everyone else. Recognising whether you are in that category, by considering whether you routinely write more than a cache-worth of data at once, tells you whether sustained write performance matters for your use or whether the cached speed, which covers everyday writing, is all you need, which is the practical value of understanding how the cache and the cliff work.

Frequently asked questions

Why does my file copy slow down halfway?

Because the SSD’s fast write cache filled up. Most consumer SSDs use a portion of flash as a fast cache to accept writes quickly, so a large copy starts fast, but once the cache fills, the drive falls back to its slower native write speed, causing the drop partway through. This is designed behaviour, not a fault, and it affects only writes large enough to exhaust the cache, which are uncommon in normal use.

What is SLC cache?

It is the fast write cache most consumer SSDs use, treating some flash cells in a faster single-bit mode to accept writes quickly, which gives the drive its high write speed for typical use. The cache is limited in size, so large writes that exhaust it fall back to the slower native speed. The cache makes everyday writing fast affordably, with the slower speed appearing only on very large transfers.

Does a full SSD write slower?

For large writes, often yes, because the fast write cache on many drives uses free space and shrinks as the drive fills, so a nearly full drive has a smaller cache and hits the speed cliff sooner. The everyday small writes that fit the cache are unaffected, but large writes expose the slower native speed earlier on a full drive. This is another reason to keep some free space on an SSD.