ユーザーの声 with CrystalDiskMark

NVMe vs SATA SSD: What to expect

Many users run CrystalDiskMark on both an NVMe and a SATA SSD to see the difference. Typical findings:

• Sequential: NVMe (Gen3) often 3–5× higher than SATA (e.g. 3000 vs 550 MB/s). Gen4 NVMe can double Gen3.
• Random 4K: NVMe usually shows much higher IOPS and lower latency, which matters for boot and app load times.
• For games and general use, SATA SSD is often "good enough"; NVMe shines in large file work and heavy multitasking.

ヒント: Use the NVMe SSD profile in Settings when testing NVMe drives so queue/thread settings match the drive type.

"My benchmark failed" — and how it was fixed

A common experience: first run fails with an error or no result.

• Run as administrator: Several users reported that after right-click → Run as administrator, the test completed. Performance tests need elevated rights on many Windows setups.
• Antivirus: One user had the test file deleted or blocked by real-time protection; temporarily disabling or adding an exclusion fixed it.
• Test size: On a nearly full or slow drive, reducing test size (e.g. to 256 MiB) avoided timeouts or failures.

ネットワークドライブ not showing

Users often run CrystalDiskMark as admin out of habit. When testing a mapped network drive, the drive didn't appear in the list. Solution: run without administrator rights (UAC: No). The drive then shows up and can be benchmarked. Keep in mind network speed and latency will limit results compared to local drives.

Comparing results with others: why numbers differ

Users sometimes compare their CrystalDiskMark results to reviews or forum posts and see different numbers. Common reasons:

• Test data: Random vs 0-fill can change results on some SSDs. Check Settings → Test Data.
• Version: Major version changes (e.g. 7 → 8 → 9) change test defaults; only compare within the same major version.
• Test size and profile: 1 GiB vs 4 GiB, or デフォルト vs NVMe profile, can change scores. Note the settings when sharing results.

Before you benchmark: tips that helped users

• Close browsers, cloud sync (OneDrive, Dropbox), and other disk-heavy apps for more consistent numbers.
• Let the SSD "settle" after a long idle or after boot — some drives need a moment to reach peak performance.
• Use the same test size and profile when comparing before/after (e.g. driver or firmware update).
• Don't benchmark too often on the same SSD; it writes a lot of data and can wear the drive over time.