DRAM vs DRAM-less SSD
A DRAM cache keeps the drive's FTL (Flash Translation Layer) map in fast memory — without it, the controller dips into the NAND itself, adding latency.
1. What the DRAM cache actually does
Every SSD must maintain an FTL map: a lookup table translating logical block addresses (what the OS sees) to physical NAND pages. With DRAM, this map lives in low-latency cache memory (~10 ns access time). Without DRAM, the map is stored in the NAND itself (~100 µs read time). The difference appears most in 4K random reads and random writes, especially when the drive's SLC write cache is exhausted.
2. When you feel the difference
Under light sequential workloads (gaming load screens, document opening), DRAM vs DRAM-less is imperceptible for most users. The gap opens under: sustained random write (large file copies to many small files, game installs, database writes), heavy multitasking that randomises access patterns, and after the SLC write cache is saturated. DRAM-less drives can drop from 3,000 MB/s peak to 400–800 MB/s sustained write once SLC cache fills.
3. HMB: the middle ground
Host Memory Buffer (HMB) allows a DRAM-less controller to use a small slice of system RAM (typically 64–256 MB) for FTL caching. HMB drives perform significantly better than pure DRAM-less designs and are sometimes marketed as "DRAM-free but with HMB support." Most modern NVMe-capable systems support HMB. If the spec sheet says "no DRAM, HMB supported," it is meaningfully better than a drive with no cache path at all.
4. QLC DRAM-less: the worst combination
QLC NAND combined with no DRAM is the most performance-limited configuration. These drives are sold on large capacity at low per-GB cost. They work for sequential read-heavy workloads (media archives, surveillance ingest) but degrade severely under mixed random I/O. If you see a 4 TB QLC drive at a notably low price, verify it has DRAM — if not, it is a dedicated bulk storage drive, not a primary workload drive.
5. Should you buy a DRAM-less drive?
DRAM-less drives are acceptable for: secondary media archive drives with mostly sequential access, budget secondary game library drives where occasional latency spikes are tolerable, and OS drives in builds where cost is the primary constraint and the workload is light. They are not appropriate for: primary gaming drives in competitive play (shader cache writes, map streaming), content creation scratch disks, or any write-heavy server-adjacent workload.
Bottom line
For any primary workload drive: get DRAM cache. The price delta between DRAM and DRAM-less at the same capacity tier has narrowed — it is rarely worth sacrificing. If you must save money, use a DRAM-less drive only as a secondary archive or bulk library, and keep the DRAM-cached drive as your primary.