GNOME Disk Utility Benchmark: How to Test Drives

GNOME Disk Utility (also called Disks) includes a built-in benchmark tool that measures a storage device’s read and write speeds. You select a drive, start the benchmark, and get average rates for reading and writing data, plus a graph showing how those rates change during the test. This article explains how to use that benchmark, what the results mean, and how it compares to other testing tools.

What GNOME Disk Utility’s Benchmark Actually Does

The benchmark in GNOME Disk Utility performs a sequential read and write test on a selected partition or disk. It writes a temporary file to the target location, measures how fast data can be written, then reads that file back to measure read speed. The results are shown as two numbers: read rate and write rate, both in MB/s (megabytes per second). The tool also plots a graph of transfer rate over time, which can reveal how a drive’s cache behaves or whether the speed drops after sustained use.

It is important to understand that this is a sequential test, not a random I/O test. Sequential tests are useful for large file transfers, but they do not represent typical everyday usage, which involves many small, random reads and writes. For a more complete picture, you would need a tool like fio or dd with specific parameters.

How to Run the Benchmark

1. Open GNOME Disk Utility (search for “Disks� in your application menu).
2. Select the disk you want to test from the left sidebar. The benchmark works on a whole disk or an individual partition.
3. Click the gear icon (or the hamburger menu) on the selected device and choose Benchmark Partition or Benchmark Disk.
4. In the dialog that appears, you can set the Number of samples (how many times to run the test) and the Sample size (the size of each read/write chunk). The defaults are usually fine for a quick check.
5. Click Start Benchmarking. You will see a warning that the test may temporarily affect system performance. The test writes and reads a file, so ensure you have enough free space on the target partition.
6. Wait for the test to finish. The results show average read and write rates, and the graph updates in real time.

The test writes a file named gnome-disk-benchmark.tmp in the target location and deletes it after completion. Make sure the partition is mounted and writable.

Interpreting the Results

The two main numbers are the average read rate and average write rate. For a typical SATA SSD, you might see 400–550 MB/s read and 300–500 MB/s write. For a NVMe SSD, speeds can reach 2000–7000 MB/s. A traditional hard drive (HDD) will usually show 80–200 MB/s. These numbers represent the maximum sustained throughput for large sequential transfers.

The graph is equally important. If the line stays flat, the drive has a consistent speed. If it drops sharply after a few seconds, that often indicates the drive’s cache (SLC cache on SSDs, or the DRAM buffer on HDDs) has been exhausted, and the sustained speed is lower than the burst speed. On HDDs, the graph also shows the expected decline as the read/write head moves from the outer to the inner platters.

Note that the benchmark uses the filesystem and the kernel’s page cache. If you are testing a partition that is mounted, the kernel may cache reads, inflating the read speed. To get a more accurate measurement, you can unmount the partition before testing, but then the tool cannot write to it. For a truly low-level test, you would need to run a tool like hdparm -t on the raw device.

Common Mistakes and Misconceptions

• Testing a mounted system partition – The benchmark writes to the partition, so if it is your root filesystem, the test will be affected by other system I/O and may not be reliable. It is better to test a secondary drive or a dedicated test partition.
• Confusing MB/s with Mb/s – The tool reports megabytes per second (MB/s). Network speeds are often in megabits per second (Mb/s). 1 MB/s = 8 Mb/s.
• Expecting random I/O results – This benchmark is sequential only. It will not show IOPS (input/output operations per second) or random read/write performance, which are more relevant for database workloads.
• Ignoring the cache effect – Many SSDs have a small fast cache. The benchmark may show a high initial speed that drops later. Always look at the graph to see the sustained speed.

How It Compares to Other Tools

GNOME Disk Utility’s benchmark is simple and built-in, but it is not the most comprehensive. For deeper analysis, Linux users often turn to fio, which can simulate various I/O patterns (random, sequential, mixed) and report IOPS and latency. Another common tool is dd, which can measure raw throughput but requires manual setup and does not produce a graph. For a GUI alternative, KDE Partition Manager has a similar benchmark, and GSmartControl focuses on S.M.A.R.T. health data rather than speed.

If you need to compare your drive’s performance against a known baseline, you can look up the drive model’s official specifications, but remember that real-world speeds depend on the interface, the filesystem, the kernel, and the workload. The GNOME tool is best for a quick sanity check, not for rigorous benchmarking.

When to Use the Benchmark

You might use this tool to:

• Verify that a new SSD or HDD is performing near its rated speed.
• Check if a drive is degrading over time (run the benchmark periodically and compare results).
• Compare two drives to decide which to use for a particular task.
• Test whether a USB drive or external enclosure is the bottleneck in your setup.

For these purposes, the built-in benchmark is perfectly adequate. It is not a replacement for professional benchmarking tools, but it is fast, free, and requires no installation.

Limitations and Caveats

One significant limitation is that the benchmark only tests sequential performance. If you are troubleshooting slow application loading or database performance, you need random I/O metrics. Also, the tool does not test the disk’s cache behavior under random access, nor does it measure latency. Finally, the benchmark can be affected by other processes running on the system. For best results, close other applications and avoid testing while the system is under load.

Another caveat: the benchmark writes a file to the filesystem, so it will wear out an SSD slightly. This is negligible for a single test, but do not run it thousands of times. Also, the test requires free space equal to the sample size; if you have less than that, the tool will fail.

Frequently Asked Questions

Can I benchmark a disk that is not mounted?

No. The benchmark writes a file to the filesystem, so the partition must be mounted and writable. If you want to test a raw device, you need to use a command-line tool like hdparm or fio on the device file (e.g., /dev/sdb), but that requires root privileges and is riskier.

Why is my read speed lower than the advertised speed?

Advertised speeds are usually the maximum possible under ideal conditions. Real-world speeds depend on the interface (SATA vs. NVMe), the filesystem overhead, the kernel version, and whether the drive is nearly full (which can slow down writes). Also, the benchmark uses a filesystem-level test, not a raw device test, so some overhead is expected.

Does the benchmark damage my drive?

No, not in a meaningful way. It writes a temporary file and deletes it. The write operation does cause some wear, but it is minimal and comparable to copying a large file. Running it once or twice is safe for any modern drive.

What is a good write speed for an SSD?

For a SATA SSD, anything above 300 MB/s is typical. For NVMe, expect 1000 MB/s or more. However, the

Related

• Alternative Proteins: Are Plant-Based and Cultured Foods Safe?
• Quota Circuit Breaker: What to Rerun After 15 Jobs
• Dying Light: The Beast Wiring and Electrical Parts

in t | Test duration | seconds | 10 | 1 | 300
out avg | Average rate | MB/s | 0 | 0