CrystalDiskMark is free benchmark software for Windows that measures the read and write speeds of any storage device - including USB flash drives, SSDs, HDDs, NVMe drives, and network drives. It is one of the most widely used drive speed testing tools available, trusted by hardware enthusiasts, IT professionals, and everyday users worldwide.
Quick answer: Run CrystalDiskMark, select your target drive from the dropdown, and click All to run the full benchmark suite. Focus on the 4K random read and write scores to understand day-to-day system responsiveness, and on sequential read and write scores for large file transfer performance. Results are displayed in MB/s. The current version is 9.0.3 (May 2026).
You can save benchmark results as a text file or image. A portable ZIP version is available with standalone executables for both 32-bit and 64-bit systems, making it ideal for storing on and running from a USB flash drive without any installation.
About CrystalDiskMark Free Benchmark Software
CrystalDiskMark is widely regarded as one of the best free benchmark tools for testing USB flash drive speed and overall storage performance. The screenshot below shows it running a quick benchmark on a SanDisk Extreme Pro flash drive.
CrystalDiskMark USB Flash Drive benchmarking made easy:
🎙️ Transcript (click to expand)
Speaker 2: Welcome back. Today we are diving deep into, well, a really essential tool, I think. It's free, and it helps answer that question. Is my storage device actually fast? You know, you get that slick new NVMe drive, or maybe you're wondering about an old USB stick. How do you actually measure the performance? We're evaluating the benchmarking utility known as CrystalDiskMark. Go, let's unpack this.
Speaker 1: Right. Crystal Disk Mark, or CDM. It's pretty much the go-to utility for this. And its genius is really its simplicity, but also its capability. It's freeware. Totally free. And it's designed to test the maximum read and write speeds of basically every type of storage you can think of. You know, old hard drives, SATA SSDs, the fast NVMe ones, even external USB drives. It's been kept up to date for years by its author, Hayao Hiyo.
Speaker 2: And I think for anyone listening who maybe troubleshoots PCs or build systems, that accessibility is just fantastic. Freeware, OK, great. But the portable app part, that's a huge convenience factor.
Speaker 1: Oh, definitely. That portability is extremely useful. You can just download the zip file, right? Put it on a USB stick. then you can run the actual program, the executables, directly from that stick. Test storage on pretty much any modern Windows machine it supports, Windows 7 right up through Windows 11. No installation needed. Perfect if you're quickly checking a friend's PC or testing parts on a bench.
Speaker 2: Okay, so you run it, and the report shows these, well, these different kinds of speeds. Sequential and random. And I think this is where people sometimes get a bit confused. Why do we need both? I mean, if I just want my computer to feel snappy, why care about, say, moving one huge file versus lots of tiny little data chunks?
Speaker 1: Ah, yes. That's really the critical insight here. So the sequential test that measures how fast data moves when it's all in one big continuous block. Think about transferring, like, a massive video file, maybe 10 gigs. A high sequential number is good for that, for bulk storage tasks. But the real measure of how fast your system feels, that often comes down to the random tests, specifically the 4K random...
Speaker 2: test. 4K random, okay.
Speaker 1: Yeah, because system responsiveness, loading apps, browsing the web, multitasking, that stuff doesn't usually involve moving one giant file. It involves reading and writing thousands of tiny little four kilobyte data blocks all over the drive, often at the same time. So if your 4K or random score is low, your whole operating system can feel sluggish, even if the sequential speed looks amazing on paper.
Speaker 2: Right, okay, that makes sense. Now, CDM also gets a bit more advanced. It lets you test with multiple cues and threads. For someone just learning about this stuff, why bother with those settings? Is that just for, like, benchmark bragging rights?
Speaker 1: No, not at all, actually. It's really vital for properly testing modern SSDs, especially NVMe drives. See, a traditional SATA drive is kind of like a single-lane road. It handles requests mostly one after the other. But NVMe drives, they're built for parallel processing. Think of a superhighway, multiple lanes, which are like the threads, and they have areas where multiple requests can stack up waiting to be processed. Those are the queues. If CDM didn't test with multiple queues and threads, it just wouldn't show you the true potential of these modern drives. It'd bottleneck the test itself.
Speaker 2: Yeah. So the NVMe drive is like this multi-lane highway, and you need a test that can actually fill those lanes to see how fast it can really go. And then we get these four main numbers, right? Megabytes per second for sequential read-write and 4K random read-write. But let's step back, connect this to the bigger picture. We see a huge number, maybe 7,000 MB/s writes. Sequential read looks incredible. But you mentioned these are synthetic benchmarks. Yeah. So does that mean they're not, like, real? How much difference is there usually between the test number and, you know, my actual daily use?
Speaker 1: Well, synthetic means it's a simulation. It's designed to push the drive to its theoretical maximum under specific controlled conditions. It's standardized, repeatable, which is great for comparing different drives side by side, but it might not perfectly match what happens when you are, say, rendering a video while also having 50 browser tabs open and downloading a game. Your specific workload might be different. So yeah, the real world performance can vary. That discrepancy is the main thing to keep in mind.
Speaker 2: Okay. And it's worth pointing out, right, that CDM is laser focused on speed benchmarks. It's not checking for drive errors or monitoring its health in real time. You'd need other tools for that complete picture. And yeah, there are alternatives out there like HDBench or Speedout, but they're also mostly synthetic tests too, generally.
Speaker 1: Exactly. They serve a similar purpose, comparative speed testing.
Speaker 2: So what does this all mean for you, the listener? CrystalDiskMark is definitely powerful. It's free, it's pretty customizable if you want it to be. Great for a quick assessment of any drive. And using it is straightforward. Pick your drive, maybe tweak settings if you know what you're doing, or just hit the all button. Then look at those results, those MB/s numbers, especially focusing on the 4K random read and write if you want to know how fast the system will likely feel day to day.
Speaker 1: Right. And that brings up a final thought, maybe something for you to chew on. When you are looking at those four key numbers, sequential read, sequential write, 4K random read, 4K random write, and you're thinking about a specific use case, like should this drive be my main OS drive for maximum responsiveness, or is it better as just a big vault for storing media files? Which one of those four metrics do you think is the single most important speed to look at if your main goal is minimizing that feeling of system sluggishness and just maximizing your daily productivity?
CrystalDiskMark is used to assess the read and write speeds of storage devices, providing a standardized, repeatable benchmark that is useful for comparing drives, verifying advertised speeds, and diagnosing performance issues.
Understanding Sequential vs. Random Speed Tests
CrystalDiskMark reports four key metrics for every drive it tests. Knowing what each one means helps you interpret your results correctly:
- Sequential Read/Write - measures speed when reading or writing large, contiguous blocks of data in order. Relevant for tasks like copying large video files, ISO transfers, or backups. A fast sequential score is important for bulk storage drives.
- 4K Random Read/Write - measures speed when reading or writing many small, 4-kilobyte blocks scattered randomly across the drive. This is the most important metric for everyday system feel - app loading, boot time, browser performance, and multitasking all depend on it. A slow 4K random score will make a system feel sluggish even if sequential numbers are excellent.
For NVMe SSDs, CrystalDiskMark also tests with multiple queues and threads (Q32T16 and similar settings) to reflect the parallel processing capability these drives are designed for. A traditional SATA drive processes requests mostly in sequence, while NVMe drives handle many requests simultaneously - testing with queues and threads reveals the true performance difference.
Features
- Sequential and random read/write tests across multiple queue and thread configurations
- Support for HDD, SATA SSD, NVMe SSD, USB flash drives, and network drives
- Flash Memory profile (added in version 9.0) optimized for USB and SD card testing
- Mix read/write test mode alongside Peak and Real World profiles
- Customizable test size, measurement count, and test types
- Results displayed in MB/s for all four key metrics
- Save results as a text file or image
- Selectable themes including Standard, Shizuku, and Aoi editions
- Dark Mode support (Windows 10 and 11)
- Portable ZIP version - runs without installation from any folder or USB drive
- Supports 32-bit and 64-bit Windows systems
CrystalDiskMark, much like CrystalDiskInfo, is widely used by hardware enthusiasts, IT professionals, and anyone who needs to evaluate storage performance. It is useful for verifying that a new drive meets its advertised speeds, comparing drives before a purchase decision, and identifying performance degradation in older drives.
How to Use CrystalDiskMark
- Download and run the application: Use the portable ZIP version to run without installation - ideal for carrying on a USB drive for bench testing multiple machines.
- Select the target drive: Choose the storage device you want to test from the dropdown menu in the top right.
- Customize test settings (optional): Adjust test size, measurement count (default is 3 in version 9+), and test types based on your needs. For USB drives, use the Flash Memory profile introduced in v9.0.
- Start the benchmark: Click All to run the complete suite, or click a specific test button to run individual tests.
- Review results: Read and write speeds are shown in MB/s for each test type. Focus on 4K random scores for system responsiveness and sequential scores for file transfer performance.
- Save results (optional): Export results as a text file or image for future reference or comparison.
Note: Running CrystalDiskMark with Administrator rights disables network drive testing. If you want to benchmark a network drive, run the application without elevated privileges (select No at the UAC prompt).
Specifications and Download
- Author's Website: CDM Project Page
- Author: Created by Hiyohiyo (Crystal Dew World)
- Current Version: 9.0.3 (May 24, 2026)
- Extracted File Size: ~15-35 MB
- License: Freeware / MIT License
- Operating System: Windows 11, 10, 8, 7 (32-bit and 64-bit)
- Price: $0 - Free
- Application Category: Benchmarking Software, Storage Utilities
- Editor's Rating: 5 Star
- Download: CrystalDiskMark Download
CrystalDiskMark Pros and Cons
Pros:
- Ease of Use: Simple, intuitive interface makes it accessible for beginners and advanced users alike.
- Free to Use: Completely free to download and use with no paid tiers or feature restrictions.
- Comprehensive Testing: Covers sequential and random read/write tests, mix mode, and peak vs. real world profiles - everything needed to evaluate storage performance thoroughly.
- Flash Memory Profile: Version 9.0 added a dedicated Flash Memory profile optimized for USB drives and SD cards, making USB speed testing more accurate.
- Customizable: Test size, measurement count, queue depth, thread count, and test types are all adjustable.
- Wide Compatibility: Tests HDDs, SATA SSDs, NVMe SSDs, USB drives, and network drives on any modern Windows version.
- Actively Maintained: Regularly updated - version 9.0.3 released May 2026 with ongoing support for the latest storage technologies.
- Portable: Runs without installation from a ZIP archive, making it easy to carry on a USB toolkit.
Cons:
- Synthetic Benchmarks Only: Results reflect performance under controlled test conditions and may not precisely match real-world workloads - actual performance can vary.
- No Drive Health Monitoring: Does not check S.M.A.R.T. data or report on drive health. Use CrystalDiskInfo for health and diagnostics.
- No Real-Time Monitoring: Designed for one-time benchmarks, not continuous performance tracking.
- Resource-Intensive During Testing: Running large test sizes or multiple iterations can temporarily affect system performance.
- Results Not Cross-Version Compatible: Benchmark scores from different major versions (e.g. version 8 vs. version 9) are not directly comparable.
- Potential for Misinterpretation: Users unfamiliar with the metrics may draw incorrect conclusions - particularly when comparing sequential scores without considering 4K random performance.
Alternatives to CrystalDiskMark
If you need a different tool or want to verify results across multiple benchmarks, these alternatives cover similar ground:
- USB Flash Benchmark: A portable tool specifically designed for USB flash drive speed testing, offering quick read and write measurements without the additional complexity of full storage benchmarking.
- HDBench: An older but reliable benchmarking tool that tests multiple hardware components including storage, CPU, and GPU - useful for a broader system performance snapshot.
- SpeedOut: A lightweight utility focused on quick USB pen drive speed tests - minimal setup, ideal for a fast read/write check without the full benchmark suite.
Comparing the Alternatives
| Feature | CrystalDiskMark | USB Flash Benchmark | HDBench | SpeedOut |
|---|---|---|---|---|
| Type of Tests | Sequential, Random, Mix | Sequential | Comprehensive (Storage, CPU, GPU) | Sequential |
| Device Compatibility | HDD, SSD, NVMe, USB, Network | USB | HDD, SSD, USB | USB |
| Flash Memory Profile | Yes (v9.0+) | N/A | No | No |
| Customizable Parameters | Yes | No | No | No |
| Portable (No Install) | Yes | Yes | Yes | Yes |
| Real-Time Monitoring | No | No | No | No |
| Actively Maintained | Yes (v9.0.3, 2026) | No | No | No |
Frequently Asked Questions
What does CrystalDiskMark measure?
CrystalDiskMark measures the read and write speeds of storage devices in megabytes per second (MB/s). It tests sequential performance (large, continuous data transfers) and random 4K performance (small, scattered data blocks). It also supports mix read/write tests and tests with multiple queues and threads for evaluating SSDs and NVMe drives.
What is a good CrystalDiskMark score?
It depends on the drive type. A typical USB 3.0 flash drive might read at 100-300 MB/s sequentially. A SATA SSD usually hits 500-560 MB/s sequential read. NVMe SSDs typically range from 3,000 to 7,000+ MB/s sequential read. For 4K random read, anything above 40-60 MB/s on a SATA SSD is normal; NVMe drives often exceed 500 MB/s. Compare your results against the manufacturer's advertised specifications as a starting point.
What is the most important CrystalDiskMark score to look at?
For everyday system performance and responsiveness - boot speed, app loading, multitasking - the 4K random read and write scores are the most important. Sequential scores matter most for large file transfers (video editing, backups, ISO copying). If your OS drive has a high sequential score but low 4K random score, your system will still feel slow.
Is CrystalDiskMark safe to run on SSDs?
CrystalDiskMark does write test data to the drive during benchmarking, which causes a small amount of wear. The developer notes this on the official site. Running it occasionally for benchmarking purposes is safe - running it continuously or with very large test sizes repeatedly would accelerate wear unnecessarily, particularly on lower-endurance USB flash memory.
How do I benchmark a network drive with CrystalDiskMark?
Run CrystalDiskMark without Administrator rights - select No at the UAC prompt when launching. With admin rights enabled, network drives are hidden from the drive selection list. Then select the mapped network drive letter from the dropdown.
What is the Flash Memory profile in CrystalDiskMark 9?
Added in version 9.0.0, the Flash Memory profile uses test parameters optimized for USB drives, SD cards, and other flash storage rather than the standard settings tuned for SSDs and HDDs. It provides more accurate results when benchmarking these device types.
Why are my CrystalDiskMark results different from advertised drive speeds?
Advertised speeds are often measured under ideal, controlled conditions using optimized test parameters. CrystalDiskMark uses standardized settings that may differ from those conditions. Test file size, drive temperature, available cache, fragmentation, and USB controller quality all affect results. Additionally, benchmark results from different major versions of CrystalDiskMark are not directly comparable.
Does CrystalDiskMark check drive health?
No. CrystalDiskMark only measures speed. For drive health monitoring, S.M.A.R.T. data, and diagnostics, use CrystalDiskInfo, which is made by the same developer and is also free.
Closing Thoughts
CrystalDiskMark is an excellent, actively maintained tool for quick and reliable benchmarking of any storage device on Windows. Its simple interface, meaningful metrics, portable design, and regular updates - now on version 9.0.3 with a dedicated Flash Memory profile - make it a go-to choice for USB speed testing, SSD performance verification, and drive comparisons. For health monitoring and S.M.A.R.T. diagnostics, pair it with CrystalDiskInfo for a complete picture of your storage device's condition and performance.