Real-world experiences with CrystalDiskMark: benchmarks, fixes, and tips from users.
Actual numbers depend on the exact drive, controller, and test settings. These ranges are only indicative.
| Drive type | Sequential read (approx.) | Sequential write (approx.) | Random 4K (approx.) |
|---|---|---|---|
| NVMe Gen4 SSD | 3,000–7,000+ MB/s | 2,000–6,000+ MB/s | High (tens of thousands IOPS) |
| NVMe Gen3 SSD | 2,000–3,500 MB/s | 1,500–3,000 MB/s | High |
| SATA SSD | 500–560 MB/s | 400–530 MB/s | Moderate to high |
| HDD (7200 rpm) | 100–200 MB/s | 100–200 MB/s | Low (tens to low hundreds IOPS) |
| USB 3.x flash / external | 50–400+ MB/s | 20–300+ MB/s | Varies a lot |
Use the same test (e.g. Default 1 GiB) when comparing. See Guide for recommended settings.
Many users run CrystalDiskMark to compare NVMe and SATA SSDs. Typical findings:
A common issue: "I run CrystalDiskMark as admin but my mapped network drive does not show." When you run with Administrator rights, Windows does not show network drives in the same session. Solution: Run CrystalDiskMark as a normal user (click No when UAC asks for admin). Then select the network drive and run the benchmark. This matches the official FAQ and fixes the problem for most users.
If the benchmark fails when writing the test file, it is often due to permissions. Solution: Right-click CrystalDiskMark and choose "Run as administrator" so it can create the test file on the selected drive. Use this for local drives; for network drives use the "no admin" approach above.
Users often notice CrystalDiskMark results differ from AS SSD Benchmark, ATTO, or manufacturer tools. This is normal: test data (random vs 0-fill), block size, queue depth, and test size all affect numbers. For fair comparison, use the same test size and similar settings, and prefer the same tool when comparing two drives. Do not compare results across different major versions of CrystalDiskMark.
For slow storage (USB flash, external HDD), use a small test size (e.g. 64 MiB or 128 MiB) so the test finishes in a reasonable time and does not fill the drive. Real World Performance or Demo profile can also be useful. Remember that benchmarking writes data and can wear USB flash; use it occasionally.
Users find File → Copy handy for pasting results into forums or support requests. Save as image (PNG/JPEG) is useful for reviews or social media. Save as text gives UTF-16LE output for logs. These options make it easy to document drive performance and get help when troubleshooting.
Some users report that a new SSD does not reach the advertised speed. Common causes: (1) drive connected to a SATA II port or wrong PCIe slot (e.g. x2 instead of x4 for NVMe), (2) power-saving or driver limiting performance, (3) testing on a full or nearly full drive, (4) comparing to a different benchmark or test size. Check the connection, use the NVMe profile for NVMe drives, and run the benchmark with minimal other activity. If sequential read is still far below spec, try another cable or port and update the storage driver.
Many users run both CrystalDiskMark and CrystalDiskInfo. CrystalDiskInfo shows health, temperature, and SMART data; CrystalDiskMark shows speed. Together they give a full picture: first check health and temperature with CrystalDiskInfo, then run a short benchmark with CrystalDiskMark. If the drive is already reporting warnings or high temperature, avoid long stress tests.
Users often run CrystalDiskMark before and after upgrading: for example, cloning from HDD to SSD, or moving from SATA SSD to NVMe. Running the same test (same size, same profile) before and after makes the difference clear. Keep a screenshot or copy of the result before the change so you can compare. Real-world feel (boot time, app launch) usually improves more than raw numbers suggest when moving to an SSD or faster NVMe.
Benchmark tools that access the disk at low level can sometimes be flagged by antivirus or Windows Defender. If you downloaded from the developer's official distribution or the Microsoft Store, the file is the genuine signed build. You can add an exclusion for the CrystalDiskMark folder or submit the file as a false positive to your security vendor. Do not disable antivirus entirely; only exclude the specific folder if you are sure of the source.