SSD Performance Test: How to Measure Speed

An SSD performance test measures a solid-state drive’s real-world speed by benchmarking its sequential and random read/write throughput, latency, and IOPS. You run a benchmarking tool, interpret the results, and compare them against the drive’s rated specifications or other drives to see if it is performing as expected.

Formula

The calculator above estimates average sequential transfer speed from a simple formula:

Average speed (MB/s) = (File size in MB × 1024) / Test duration in seconds

Where:

• File size – the amount of data written or read during the test, in megabytes (MB).
• Test duration – the time it takes to complete that transfer, in seconds.
• Average speed – the resulting throughput, in megabytes per second (MB/s).

This is the same calculation benchmarking tools use internally to report sequential speed. Real benchmarks also measure random I/O, latency, and queue depth, which this simple formula does not capture.

Worked Example

Using the default values from the calculator above:

1. File size = 1024 MB (1 GB)
2. Test duration = 10 seconds
3. Convert file size to KB: 1024 × 1024 = 1,048,576 KB (optional, but keeps units consistent)
4. Apply the formula: (1024 × 1024) / 10 = 104,857.6 KB/s
5. Convert to MB/s: 104,857.6 / 1024 ≈ 102.4 MB/s

The calculator outputs 102.4 MB/s, matching the step-by-step result.

Why SSD Performance Tests Matter

SSDs are not all equal. A budget SATA SSD might hit 550 MB/s sequential read, while a high-end PCIe 4.0 NVMe drive can exceed 7,000 MB/s. That difference affects how fast your PC boots, how quickly games load, and how snappy file transfers feel. But sequential speed is only half the story. Random read/write performance (small files at random locations) matters more for everyday tasks like opening apps, loading game levels, and multitasking.

Manufacturers often quote only the maximum sequential numbers, which are achieved under ideal conditions with large files and deep queues. Real-world workloads are messier. That’s why independent benchmarking is useful: it tells you what the drive actually delivers on your system, with your motherboard, your driver, and your workload.

How SSD Performance Tests Work

Benchmarking tools like CrystalDiskMark, AS SSD, and ATTO Disk Benchmark create test files of a set size (e.g., 1 GB) and perform read/write operations in different patterns:

• Sequential read/write – large contiguous blocks, simulating big file transfers (e.g., copying a movie).
• Random read/write (4K) – small 4 KB blocks at random positions, simulating OS and app operations.
• Queue depth (QD) – how many commands the drive handles simultaneously. QD1 is typical for a single user; QD32 is used to stress the drive’s controller.
• Mixed workloads – some tools test simultaneous read/write, which is common in database servers.

The tool measures how long each operation takes and calculates throughput (MB/s) and IOPS (input/output operations per second). Latency (the time for a single operation) is often reported as average, max, and sometimes 99th percentile.

How to Choose an SSD Benchmark Tool

There are several reputable free tools. Each has strengths:

• CrystalDiskMark – the most common, easy to read, and available for Windows. It tests sequential and random 4K with customizable queue depths.
• AS SSD Benchmark – similar to CrystalDiskMark but adds a compression test and access time measurement. It’s also Windows-only.
• ATTO Disk Benchmark – tests a range of transfer sizes (from 0.5 KB to 64 MB) and shows a graph. Useful for seeing how performance scales with file size.
• Blackmagic Disk Speed Test – designed for video editors, works on macOS and Windows, and uses real-world video file sizes.
• fio – a command-line tool for Linux that offers deep control over test parameters. Not for beginners.

For most users, CrystalDiskMark with default settings is sufficient. Run it a few times and take the median result, because SSDs can vary run-to-run due to background garbage collection and thermal throttling.

Common Problems and Misconceptions

One of the most common mistakes is comparing results from different tools or different test settings. A 1 GB sequential test on one tool might not match a 1 GB test on another because of cache behavior, queue depth, or file alignment. Always compare apples to apples: same tool, same settings, same drive capacity.

Another misconception is that a drive’s rated speed is the speed you will always get. Real-world speeds are lower because of:

• Thermal throttling – NVMe drives can overheat and slow down to protect themselves.
• DRAM-less design – budget drives use host memory buffer or SLC cache, which slows after the cache fills.
• Full drive – performance degrades as the drive fills up, especially on TLC/QLC drives.
• Driver and firmware – outdated drivers or firmware can limit performance.

Also, note that the calculator at the top of this page gives a rough average; it does not account for caching or random I/O. It is meant for quick estimation, not as a substitute for a proper benchmark.

How to Read and Interpret Results

When you get your benchmark results, look at these numbers:

• Sequential read – the headline number. For SATA, expect ~550 MB/s; for PCIe 3.0 NVMe, ~3,500 MB/s; for PCIe 4.0, ~7,000 MB/s (theoretical max).
• Sequential write – usually lower than read, sometimes significantly so on budget drives.
• Random 4K read (QD1) – this is the most important for everyday feel. Values above 50 MB/s are good; above 80 MB/s is excellent.
• Random 4K write (QD1) – similar importance, but often lower than read.
• IOPS – higher is better, but only matters for heavy multitasking or server workloads.

If your sequential speeds are far below the rated spec (e.g., a PCIe 4.0 drive showing 2,000 MB/s), check if it’s plugged into a PCIe 3.0 slot, if the motherboard’s M.2 slot shares bandwidth with other devices, or if the drive is overheating.

Frequently Asked Questions

What is a good SSD speed for gaming?

For gaming, random read performance matters more than peak sequential. A SATA SSD with ~500 MB/s sequential and decent 4K random read (above 40 MB/s) is fine for most games. NVMe drives load games a bit faster, but the difference is often just a few seconds. The key is having an SSD at all; the jump from HDD to SSD is huge, while SATA to NVMe is modest.

Why does my SSD benchmark show lower speeds than advertised?

Advertised speeds are maximums under ideal conditions (often with high queue depth and large files). Real-world tests use different settings, and your system may have other bottlenecks like a slower CPU, outdated drivers, or the drive being nearly full. Also, some drives use a pseudo-SLC cache that speeds up writes but only for a limited amount of data; once the cache fills, speeds drop to the native TLC/QLC speed.

Should I run a benchmark on a new SSD?

Yes, it’s a good idea to verify the drive is working as expected. Run a quick CrystalDiskMark test after installation. If the numbers are drastically lower than the spec, check connections and firmware. However, note that benchmarking writes data to the drive, which consumes some of its lifespan, though modern SSDs handle this easily.

How often should I test my SSD?

There’s no need to test regularly. Run a benchmark when you first get the drive, and again if you notice slowdowns. Frequent benchmarking is not harmful, but it’s also not necessary. Instead, monitor drive health with tools like CrystalDiskInfo (which reads SMART data).

Does an SSD slow down over time?

Yes, but not because of age alone. The main causes are: the drive filling up (especially above 75-80% capacity), background garbage collection, and thermal throttling. Also, NAND cells wear out, but for typical use, a modern SSD will outlast the computer it’s in. Defragmenting an SSD is not needed and can reduce its lifespan.

What is the difference between MB/s and IOPS?

MB/s (megabytes per second) measures throughput – how much data is transferred per second. IOPS (input/output operations per second) measures how many individual read/write operations are completed per second. For large sequential transfers, MB/s is the key metric. For small random operations (like loading many small files), IOPS matters more. A drive can have high MB/s but low IOPS, or vice versa.

Related

• One-Time API Tokens: Replay Attack Defense & Trade-Offs
• Mass of Copper from Volume Calculator
• Compost Tons to Cubic Yards Converter

in t | Test duration | seconds | 10 | 1
in s | File size | MB | 1024 | 1
out speed | Average speed | MB/s | (s * 1024) / t | 2