CrystalDiskMark Guide: Benchmark Results Explained

Written and maintained by Joseph James Stahl · DriveMetric Guide

Learn what sequential and random tests, queue depth, MB/s and IOPS mean.

What a storage benchmark measures

CrystalDiskMark issues controlled read and write workloads to a selected storage target. The result describes performance under those test conditions; it is not a universal score for overall PC speed.

Sequential vs random performance

Sequential workloads move large contiguous blocks and are useful for large-file throughput. Random workloads access smaller blocks in scattered locations and can better represent many operating-system and application patterns.

A drive may lead in peak sequential throughput while showing smaller gains in low-queue-depth random work. The useful metric depends on what you actually do with the drive.

Queue depth and thread count

Queue depth is the number of outstanding I/O requests. Higher queue depths can expose parallel capability; queue depth 1 can be more representative of light desktop activity. Thread count changes how many workers generate requests.

MB/s, IOPS and latency

Why results change between runs

Background activity, antivirus scanning, temperature, free space, SLC cache state, power settings, drivers, filesystem state and test parameters can all change results. Sustained writes can look very different from short burst tests.

How to compare results fairly

1. Confirm exact drive model and capacity.
2. Use the same benchmark profile and units.
3. Keep the same test size and number of runs.
4. Note free space and temperature.
5. Let background activity settle.
6. Compare several runs rather than one outlier.