Download CrystalDiskMark for Windows

Developer: Crystal Dew WorldOfficial website ↗

Measure storage performance with selected read and write workloads on a Windows computer.

The ZIP contains a shared installer, not a standalone copy of CrystalDiskMark. Check the available software before installing.

Free, open sourceUtilitiesWindows

Give a storage-speed question a measurable answer

A new SSD, an external enclosure or a different USB connection can change how storage behaves. CrystalDiskMark measures read and write performance with defined test workloads, giving you results that can be recorded and compared. It is useful when you want to investigate whether a particular setup performs as expected. The numbers describe that test under those conditions, rather than a universal rating for every application on the computer.

Sequential tests concern transfers through larger continuous areas, while random workloads exercise smaller accesses in different locations. These are different patterns, so the largest number on the screen is not automatically the best explanation for slow application startup. Queue and thread settings also affect the workload. Keep the test labels with the results instead of saving only one isolated figure.

Select the intended drive and a consistent setup

Check the selected drive before starting, especially when several internal and external volumes are available. The test uses storage space and performs writes; it is not a passive inspection of existing health information. Leave enough free space for the chosen test data and avoid benchmarking a drive that is already showing serious read errors or disappearing unexpectedly. Preserving important files takes priority in that situation.

For a useful comparison, keep the program's major version, profile, test size and repetition count consistent. Also note the connection type and whether other jobs were using the same device. A background copy or backup can compete for access. Changing several conditions at once makes it harder to understand why a later result differs from an earlier one.

Interpret the result in the context of the device

A drive attached through an external bridge may be limited by the enclosure, cable or port as well as the drive itself. Available space, controller behavior, caching and temperature-related conditions can influence measured performance. A short test and a longer sustained transfer therefore do not necessarily produce the same rate. Avoid turning a single run into a claim about every future workload.

The developer warns that results from different major versions are not directly comparable. When checking a result against another person's screenshot, compare the settings and version first. The same product name printed on two SSDs does not establish identical system conditions, capacity variants or firmware. Use measurements to narrow a question, then change one relevant condition if another test is actually needed.

Distinguish speed measurement from drive health

CrystalDiskInfo provides a different kind of information through supported drive status and SMART reporting. A high benchmark result does not prove that a drive is healthy, and a lower result does not identify a specific hardware fault by itself. CrystalDiskMark neither repairs a filesystem nor recovers deleted documents.

The application is free and open source, with Windows packages for supported x86, x64 and ARM64 environments. Its published compatibility list includes older Windows releases, but the installer and x64 version do not support Windows XP or Server 2003. Repeated write tests add writes to flash storage, so run enough tests to answer the practical question rather than looping them indefinitely. Keep your existing backup and record the settings beside each result.

For recording the wider hardware configuration alongside a benchmark, AIDA64 Extreme can provide a system overview. Keep that context separate from the measured transfer rates.

Questions and answers

Does a high score prove my SSD is healthy?

No. A benchmark measures performance under a selected workload. It does not replace drive-status information, error investigation or a backup, and cannot guarantee that the device will continue working without faults.

Why are two screenshots of the same SSD different?

Test settings, software major version, connection, available space and other activity can differ. Compare those conditions before drawing conclusions; the same drive model does not make two measurement environments identical.

Should I run the benchmark continuously?

Repeated write tests add writes to SSDs and USB flash storage. Use a deliberate test plan and stop when you have answered the question; indefinite repetition is not a routine health-maintenance procedure.

Softmesa is an independent catalog and is not affiliated with the developer. Product names and logos belong to their owners. Always check the file source before installing.

Similar programs

More in Utilities →