CrystalDiskMark Results Explained
Learn what sequential and random results mean, why SSD benchmark numbers vary, and how to compare drives without mixing incompatible test conditions.
How to read CrystalDiskMark results
CrystalDiskMark runs controlled storage workloads. Sequential tests emphasize large contiguous data transfers, while random tests work with smaller scattered blocks. Queue depth describes how many requests are outstanding, and thread count describes the workers issuing them.
| Metric | What it tells you |
|---|---|
| Sequential MB/s | Large-file throughput under the selected test conditions. |
| Random MB/s or IOPS | Performance on small scattered operations. |
| Queue depth | How much parallel I/O is outstanding. |
| Latency | Time required to complete operations; useful for responsiveness. |
Why a result may look too low
Before assuming the SSD is defective, check the negotiated interface, exact drive capacity, test profile, temperature, free space, Windows power state and background activity. NVMe drives can be limited by PCIe generation or lane width, while SATA SSDs are constrained by the SATA interface itself.
Write performance can also change when an SSD’s fast SLC cache fills. A short benchmark and a sustained file transfer can therefore produce different but valid answers.
How to compare two results fairly
- Use the same CrystalDiskMark profile and units.
- Match test size and number of passes.
- Confirm exact drive model and capacity.
- Record free space and temperature.
- Run several tests under similar idle conditions.
- Compare representative results instead of selecting one peak screenshot.
What this guide does not do
This page does not provide CrystalDiskMark and does not act as the official developer site. Its purpose is to explain benchmark interpretation and troubleshooting.