CrystalDiskMark 9 Guide for Windows: Benchmark SSD Speed, Test Modes and 0 Fill Explained

For Windows enthusiasts and system administrators, CrystalDiskMark (CDM) is practically furniture. Developed by Noriyuki Miyazaki under the Crystal Dew World banner, the open-source utility has reached the CrystalDiskMark 9 series—with version 9.0.3 listed on the project's official download portal alongside the separate CrystalDiskInfo 9.9.3 health utility. Yet even veteran PC builders occasionally misread what its settings actually do, or confuse speed benchmarking with hardware health diagnostics. Breaking down how CrystalDiskMark works, where vendor utilities overreach, and what the official documentation says about its test modes will help you get real answers from your storage hardware.

Why Independent Benchmarking Beats Branded Drive Suites​

A major reason experienced reviewers and everyday tinkerers gravitate toward CrystalDiskMark is what the software doesn't do. As Clemons notes in ZDNET's report, proprietary management utilities from major SSD manufacturers often bundle drive monitoring with aggressive data collection. While tools like Samsung Magician provide legitimate utility for firmware updates and hardware-specific features, their license agreements routinely authorize the collection of drive models, connected hardware identifiers, and usage logs whenever the software connects to the internet.

CrystalDiskMark takes the opposite philosophical approach:

• Fully Local Operation: The benchmark runs entirely on your local machine without requiring an internet connection to test drives, outputting results only when you choose to save a text log or screenshot image.
• Open-Source Transparency: Released as open-source software with source code available for inspection, it avoids background services, system tray clutter, and upsell prompts.
• Broad Hardware and OS Compatibility: While ZDNET notes compatibility going back to Windows 8, Crystal Dew World's official system requirements actually go even further—supporting Windows XP, Vista, 7, 8, 8.1, 10, and Windows 11, alongside Windows Server 2003 through Windows Server 2025 across x86, x64, and ARM64 architectures (with the sole caveat that the installer and x64 builds do not run on Windows XP or Server 2003).
• Interface Agnosticism: Whether you are testing a legacy mechanical hard drive over SATA, an internal PCIe NVMe M.2 drive, a removable USB thumb drive, or a multi-drive RAID array, the program applies the same transparent read and write passes.

What if you also manage non-Windows machines in a mixed household or lab? While CrystalDiskMark itself is strictly a Windows application, ZDNET points out two capable open-source counterparts: AmorphousDiskMark for macOS (modeled after CDM with the original creator's blessing) and KDiskMark for Linux, which mirrors CDM's visual layout and testing behavior.

Section Summary: Proprietary SSD toolboxes are useful when you specifically need a firmware patch, but for day-to-day speed verification across mixed brands and interfaces, CrystalDiskMark delivers clean, local testing without background telemetry.

Speed Benchmark vs. Drive Health: Clearing Up two Common Myths​

Before diving into the menus, it is crucial to establish a distinction that frequently trips up PC users—and even appears in ZDNET's walkthrough. CrystalDiskMark is a performance benchmark, not a drive-health monitor.

Crystal Dew World maintains two separate flagship storage utilities on its download page:

1. CrystalDiskMark: Described officially as "a simple storage benchmarking software" that measures sequential and random read/write speeds.
2. CrystalDiskInfo: Described officially as "a storage health monitoring software" that reads S.M.A.R.T. (Self-Monitoring, Analysis, and Reporting Technology) attributes, drive temperatures, power-on hours, and wear-leveling percentages.

Keeping that boundary in mind also clears up a major technical misconception regarding CrystalDiskMark's All 0x00 (0 Fill) setting under Settings > Test Data. In ZDNET's guide, Clemons describes the all 0x00 option as a diagnostic test that checks each storage sector for a 0x00 code to determine drive health and whether you need a disk cleanup.

In reality, Crystal Dew World's official documentation specifies that Test Data simply selects the payload pattern written during the benchmark: either Random (incompressible pseudo-random data, the default) or All 0x00 (0 Fill) (highly compressible streams of zeroes). Why does that toggle exist? Historically and on certain modern budget controllers, some SSDs use real-time data compression. If you feed those drives a stream of all zeroes (0x00), the controller compresses the data on the fly and reports artificially sky-high transfer speeds, whereas random data forces the drive to show its true physical write performance. It does not scan sectors for corruption or tell you to run Windows Disk Cleanup. If a drive's speeds plummet over time, that can certainly be a symptom of a nearly full drive, thermal throttling, or hardware degradation—but you should pair CrystalDiskMark with CrystalDiskInfo if you want actual health telemetry.

Section Summary: Use CrystalDiskMark to measure how fast your drive moves data, and rely on S.M.A.R.T. tools like CrystalDiskInfo to check physical drive health. Never confuse the All 0x00 zero-fill compression test with a sector repair or health scan.

Step-by-Step: How to Benchmark Your Drive in Windows​

Whether you just slotted a new M.2 drive into your motherboard or want to find out which unlabeled USB stick in your desk drawer is actually USB 3.2 versus a sluggish USB 2.0 relic, CrystalDiskMark scales from one-click simplicity to granular queue-depth tuning.

Level 1: The Simple Baseline Test​

For 90% of home users, the default main window is all you need to verify a drive is operating in the right ballpark:

1. Select the Number of Test Runs: Use the first drop-down menu at the top left to pick between 1 and 9 passes. Crystal Dew World sets the default to 5.
2. Choose the Test File Size: Use the second drop-down menu to choose a payload between 16 MiB and 64 GiB. The default 1 GiB setting works well for internal SSDs and modern external drives. However, Crystal Dew World's manual explicitly notes that 1 GiB can be painfully slow on low-speed storage like older USB flash drives—dropping down to 64 MiB or 256 MiB will save you from staring at a frozen progress bar.
3. Select Your Target Drive: Pick the drive letter (C:\, D:\, F:\, etc.) from the third drop-down menu.
   • Pro Admin Tip for Network Drives: According to Crystal Dew World's official documentation, if you run CrystalDiskMark with Administrator privileges (clicking "Yes" on the Windows User Account Control prompt), mapped network drives will not appear in the drive list because Windows isolates elevated sessions from standard user network mappings. To test a NAS or network share, launch CrystalDiskMark as a standard user without Administrator rights.
4. Pick Your Measurement Unit: The fourth drop-down defaults to MB/s (defined by the developer as 1,000,000 bytes/sec), or in the top-right unit selector you can view results in GB/s, IOPS (Input/Output Operations Per Second), or μs (Average Latency in microseconds).
5. Run the Suite: Click the green All button to run all four rows of tests sequentially, or click an individual row button (SEQ1M, RND4K, etc.) to test a single metric.

When the numbers populate, look at the two distinct halves of the chart. The top rows (Sequential) measure contiguous large-file transfers—like copying a 50 GB movie file—and usually match the big numbers advertised on the drive's box. The bottom rows (Random 4K) simulate the chaotic, small-block reads and writes of booting Windows, launching apps, and loading game assets, which matters far more for how snappy your PC actually feels.

Level 2: Profiles and Drive-Specific Presets​

If you want to tailor the benchmark to specific hardware or workloads, open the Settings and Profile menus at the top of the window:

Level 3: Customizing Queues, Threads, and Block Sizes​

If you are benchmarking enterprise storage or hunting for controller bottlenecks, clicking Settings > Settings opens a dedicated configuration dialog. Here you can manually alter block sizes, queue depths (Q), thread counts (T), measurement duration, and interval cool-down timers across the test tabs. Unless you are reproducing a specific workload or hardware review methodology, leaving these on the preset tabs ensures your numbers stay comparable to standard community benchmarks.

Section Summary: Stick to the 1 GiB default for SSDs, lower the test size for slow USB sticks, use Settings > NVMe SSD for modern M.2 drives, and launch without Admin rights if you need to test a mapped network share.

The Caveats: Drive Wear and Manual Record-Keeping​

No tool is without its quirks, and there are three practical limitations every Windows user should keep in mind when running CrystalDiskMark:

1. No Built-In Historical Graphing: As Clemons points out on ZDNET, CrystalDiskMark still lacks an internal database or history viewer to track a drive's speeds over months or years. While you can export results as clean text files or PNG screenshots via the File menu, tracking long-term performance requires keeping your own folder or spreadsheet.
2. Version and Environment Sensitivity: Crystal Dew World explicitly warns that benchmark results are not compatible across different major versions of CrystalDiskMark, and scores will fluctuate based on file size, drive fragmentation, how full the drive is, controller drivers, and even CPU speed. If you are logging results in a spreadsheet, always note the exact CDM version and profile used.
3. Mind the Write Endurance: Running CrystalDiskMark writes gigabytes of test files to your storage medium. Both Crystal Dew World and storage testing guides warn that running repeated, unnecessary benchmark loops can gradually eat into the finite Terabytes Written (TBW) endurance of SSDs and flash memory. A single test pass uses a tiny fraction of a modern SSD's lifespan, so you should never hesitate to test a new drive or troubleshoot a sluggish PC—just don't leave 64 GiB 9-pass loops running all afternoon for entertainment.

At the end of the day, CrystalDiskMark endures because it respects the user. It downloads in seconds, runs portably from a ZIP archive or standard installer, asks for zero personal information, and tells you the unvarnished truth about your storage hardware. Pair it with a disciplined testing routine—and a true S.M.A.R.T. reader when you need health diagnostics—and you will never have to take a storage manufacturer's marketing sticker on faith again.