How to Use Portable CrystalDiskMark to Benchmark External SSDs
What you’ll need
- Windows PC (Windows 7–11)
- External SSD connected (USB 3.⁄3.2, USB-C, or Thunderbolt recommended)
- Portable CrystalDiskMark ZIP or PortableApps build (no install)
- USB cable and an idle system (close other heavy I/O apps)
Safety & notes (brief)
- Benchmarks can shorten SSD lifespan if run repeatedly—use sparingly.
- Results vary by cable, enclosure, controller, port, OS power settings, and test file size.
- Do not run on a drive with important unsaved work open.
Download & launch (portable)
- Download the portable ZIP from the official CrystalDiskMark site or PortableApps.
- Extract the ZIP to a folder on your local drive or the USB stick you’ll carry.
- Run DiskMark64.exe (or the PortableApps launcher) — no installation required.
Configure CrystalDiskMark for external SSDs
- In the main window, select the target drive (carefully pick the external SSD letter).
- Set Test Size: choose at least 1G (1,024 MB) for SSDs to avoid caching effects; 4G or 8G for more accurate sustained speed.
- Set Count/Number of Runs: 3 is a good default; increase for stability.
- Choose Test Type:
- Seq Q32T1 (sequential, queue depth 32, thread 1) for large-file throughput.
- 4K Q1T1 and 4K Q32T1 for small-random I/O (real-world app responsiveness).
- Choose Test Data: Random is typical; 0Fill may produce optimistic results on some SSDs—use Random for realistic numbers.
- Optional: enable Mix or other modes for specialized testing.
Run the benchmark
- Ensure the external SSD isn’t busy (no file copies, no antivirus scans).
- Click All (runs the full preset suite) or click specific test buttons.
- Wait for completion. Don’t disconnect the drive while testing.
Interpreting results
- Results show MB/s for read/write in each test.
- Sequential speeds reflect large-file transfers (gaming installs, backups).
- 4K random speeds reflect small-file operations (OS/app responsiveness).
- Compare peak values against the SSD’s advertised specs but expect lower numbers for USB-attached drives due to enclosure/controller limits.
- For NVMe in a Thunderbolt enclosure you may see near-native speeds; for USB-A/USB-C controllers, expect reduced throughput.
Troubleshooting & tips
- If speeds are much lower than expected:
- Use a different cable or port (prefer USB-C/3.2 Gen2 or Thunderbolt).
- Confirm the enclosure supports UASP and the SSD’s interface (NVMe vs SATA).
- Disable power-saving or USB selective suspend in Windows.
- Run tests with larger Test Size to avoid caching artifacts.
- For consistent comparisons, run tests multiple times and use the median/average.
Saving and sharing results
- Use File → Save Text or use the screenshot/note field in CrystalDiskMark to label the drive and test conditions.
- Record test date, Windows power profile, port/cable, enclosure model, and test size for repeatable comparisons.
Quick checklist (before running)
- External SSD connected to high-speed port
- Test Size ≥ 1G (prefer 4G)
- Test Count = 3
- Test Data = Random
- Close other I/O-heavy apps
- Note cable/enclosure model
Following this process will give you repeatable, meaningful performance numbers for external SSDs using the portable CrystalDiskMark build.