Two solid-state drives can share the same capacity, the same connector and almost the same price, yet one feels noticeably quicker when you open folders, launch programs or copy thousands of small files at once. The usual reason is a small pool of fast memory that the cheaper drive leaves out to save money. This piece explains what that memory does, how a drive copes without it, and the everyday situations where its absence becomes easy to feel. By the end you should be able to read a specification sheet and predict which drive will stay responsive in ordinary use.
The address book inside every SSD
A solid-state drive does not keep your files in fixed physical spots. Its controller constantly shuffles data between flash blocks to spread wear evenly and to reclaim space that deleted files leave behind, so the true location of any piece of data changes over its life. To track all that movement, the drive keeps a mapping table, often called the flash translation layer, that converts the logical addresses your computer asks for into the physical addresses where the data actually sits at that moment.
That table is not small. A widely quoted guideline is around one megabyte of mapping information for every gigabyte of capacity, so a one terabyte drive needs roughly a gigabyte of table simply to describe where everything lives. Every read and write consults this table first, so the speed of looking up an entry sets a floor under how quick the drive can feel.
Why the table wants fast memory
Since the mapping table is touched on every operation, where it physically resides carries a lot of weight. If the controller can hold the entire table in fast memory placed right beside it, a single lookup takes a tiny fraction of a microsecond and adds almost nothing to each request. Drives built for responsiveness therefore fit a dedicated DRAM chip, broadly the same family of memory your computer uses as its main system memory, set aside purely for this bookkeeping.
DRAM suits the job well. It is very quick to read and can be rewritten endlessly without degrading, which matters for a table that changes with nearly every write. Holding the map there lets the controller answer requests that jump around the drive without pausing to search slower storage for the right entry.
Borrowing memory from the host
To trim cost, many current drives leave out the DRAM chip but keep a middle option called the host memory buffer, usually shortened to HMB. The drive borrows a modest amount of your computer’s own RAM, commonly a few tens of megabytes, to cache the busiest portions of the mapping table. It cannot hold the whole table, but it keeps the most-used entries within easy reach, so routine work stays reasonably brisk.
That borrowed cache travels over the same PCIe link that carries your files, so the speed and generation of the connection has some bearing on how well the arrangement holds up, the interface question the guide on whether a faster storage interface earns its price also weighs. Host buffering helps most when your active data stays small and predictable.
How a cache-less drive behaves
A drive with neither onboard DRAM nor a host buffer must keep its full mapping table in the flash itself. Whenever it needs an entry that is not already in the controller’s tiny internal memory, it first has to read the right slice of the table from the NAND, and only then can it carry out the operation you requested. That amounts to two visits to the flash for one request, and the extra visit adds delay whenever the needed entry is not on hand.
For long sequential transfers this hardly shows, because the drive works through addresses that sit close together and one table entry serves many operations in a row. The additional lookups only accumulate under scattered access, and that repeated internal shuffling forms part of the ordinary write activity the guide on how everyday use gradually wears a drive describes.
| Factor | DRAM cache | Host memory buffer | No cache |
|---|---|---|---|
| Where the map lives | Dedicated chip on the drive | Small slice of system RAM | In the flash itself |
| Small random reads and writes | Consistently quick | Good for common data | Slows under scattered access |
| Large sequential transfers | Fast | Fast | Nearly as fast |
| Typical price tier | Higher | Middle | Lowest |
Where the slowdown shows up
The plainest symptom is weak small random performance, the measure of how fast a drive deals with many little reads and writes spread across its capacity. Starting an operating system, opening an application, installing updates and searching large folders all produce this pattern, so a drive that struggles here feels slow in the moments you tend to notice.
A benchmark that copies one very large file can make a cache-less drive look healthy, since that is the task it handles best. The gap widens under queued random requests, where a DRAM-equipped drive keeps replying steadily while the cache-less one keeps pausing to fetch table entries from flash. If you plan to run the drive inside a portable enclosure, the same responsiveness concerns feed into the broader trade-offs the guide on matching an external drive to your needs lays out.
When the difference matters to you
For a drive that mainly stores material you read in bulk, films, disk images, a photo archive you copy across in big batches, a DRAM-less model can be entirely satisfactory and saves real money. The workload is gentle on it, and you may never sense the missing cache during normal use.
The sums change for a system drive. When the SSD carries your operating system and the programs you open every day, its small random behaviour is on display from the moment you power on. There the extra few pounds for a drive with DRAM, or at least dependable host-buffer support, generally buy a computer that stays quick under everyday multitasking.
Choosing with this in mind
None of this brands cache-less drives a bad choice on their own. They exist to meet a genuine need, inexpensive and capacious storage for data that does not demand instant random access. The error lies in treating every SSD as interchangeable and then puzzling over why a cut-price drive feels slower than the one it replaced.
Read the specification for any mention of DRAM or a host memory buffer, and fit the drive to its intended task. Fast dedicated memory belongs where snappy small operations count, while plain capacity belongs where you chiefly move large files or store material for years, the archival angle the guide on whether optical discs still earn a place examines.
Frequently asked questions
What is a DRAM-less SSD?
It is a solid-state drive built without a dedicated DRAM chip for holding its mapping table. Instead it keeps the full table in flash and caches only a little of it, either in the controller’s small internal memory or in a borrowed slice of system RAM. The design lowers cost, and it changes how the drive behaves under scattered small operations.
Does an SSD need DRAM?
It does not strictly need it to work, and many dependable drives manage without. DRAM mainly helps with small random reads and writes by keeping the address map instantly available. For a drive that mostly stores large files you may never miss it, but for a busy system disk a DRAM cache or strong host-buffer support keeps everyday tasks feeling quick.
Why is my cheap SSD slow at small files?
Small-file work generates scattered requests, and a budget drive without DRAM often has to fetch mapping-table entries from flash before each one. That extra step adds latency a large sequential copy never exposes. Check that the drive is not nearly full, that TRIM is active, and that it runs on an adequate interface, then judge whether its cache design suits your workload.
