SSD and Hard Drive Benchmarks: Best Free Tools

The large number at the top of the tests—those thousands of megabytes per second in sequential read—is actually what matters least in everyday use. Your computer’s responsiveness depends on random 4K values with a queue depth of one, which you’ll find at the bottom of the screen; these are hundreds or even a thousand times smaller. Once you understand this line, you’ll also understand why a “slow on paper” drive can feel snappier than an incredibly fast one.

Beyond speed, there are two other things to check on a drive: whether it is healthy and, in the case of a USB stick or SD card bought online, whether it actually has the capacity it claims to have. These are different checks, and each requires a different program.

The tools #

Program	Purpose
CrystalDiskMark	Measure speed: this is the industry standard everyone uses
CrystalDiskInfo	Read drive health and temperature
ATTO Disk Benchmark	See how speed changes with different block sizes
AS SSD Benchmark	Tests designed for SSDs, featuring real-world file copy tests
HD Tune	Speed graph across the entire surface, useful for mechanical hard drives
H2testw, ValiDrive, F3	Detect USB sticks and SD cards that claim more space than they actually have

CrystalDiskMark #

This is the industry standard: it’s free, lightweight, and because everyone uses it, results are easy to compare. Select your drive, the number of passes, and the test file size, then press All.

Two things to know before running it:

The test actually writes to the drive. Running it dozens of times out of curiosity consumes write cycles. It’s not a major issue, but it is unnecessary.
Different major versions yield different results. If you are comparing your numbers with those found online, make sure they were generated using the same version.

CrystalDiskInfo #

This does not measure speed. It reads SMART parameters—the drive’s own self-diagnostics—and shows you an overall health status, power-on hours, total data written during its lifetime, and temperature.

This is the first program you should install because it tells you if your drive is about to fail. Recent versions also flag counterfeit drives by comparing firmware and IDs with what is expected for that specific model.

CrystalDiskInfo screen showing health status, temperature, power-on hours, and SMART parameters — Health status, temperature, and total data written during the drive’s life

AS SSD and HD Tune #

AS SSD Benchmark is free and portable, designed specifically for SSDs: in addition to sequential and random tests, it measures access times and simulates real folder copies (like an ISO image, a program, or a game), providing an overall score at the end. HD Tune is a classic for mechanical hard drives: it plots a speed graph from the beginning to the end of the disk and includes a surface error scan. The free version is old but still works; the Pro version is paid.

How to read the results #

The test rows have somewhat confusing names and mean very different things.

SEQ1M refers to sequential large-block read and write speeds—the number manufacturers print on the box. This matters when you are copying a massive file, such as a video, a disk image, or a folder of photos. For everything else in your daily routine, it matters very little.

RND4K Q32T1 measures random access to small blocks with many requests queued up at once. This represents a server or a computer performing ten tasks simultaneously.

RND4K Q1T1 measures random access to small blocks with one request at a time. For typical PC users, this is the most important row: launching Windows, opening an app, or loading a game level involves reading thousands of tiny files one after another.

Look at the last line: between an SSD that does 7,000 MB/s sequential and one that does 3,500 MB/s, you’ll barely notice the difference in daily use. But between one that does 60 MB/s in RND4K Q1T1 and one that does 90 MB/s, you will feel it every time you open something. This is also why switching from a mechanical drive to an SSD is life-changing, whereas the difference between two modern SSDs is barely noticeable.

To give you an idea of the scale: a mechanical hard drive performs a few hundred random operations per second; an SSD performs tens of thousands. That jump is where all the performance gain lies.

Why speed drops after a few minutes #

You might measure 5,000 MB/s in write speed and then see a 100 GB copy drop to one-tenth of that. The drive isn’t broken: many SSDs use part of the memory in a faster mode—a sort of transit area (the SLC cache)—and once it fills up, they continue writing at the actual speed of the cells.

Therefore, a short test measures the cache, not the drive itself. To see how it performs under a sustained load, you need a test with a file much larger than the cache.

Another factor on faster drives is temperature: once they hit a certain threshold, they slow down to protect themselves. If your numbers drop with every repetition, check the temperature in CrystalDiskInfo and, if necessary, install an SSD heatsink.

Graph of a prolonged copy where write speed suddenly drops after the first few gigabytes — Once the transit area is full, writing drops to real-world speeds

Verifying a USB stick or SD card #

This should be done on every storage device bought online, especially if the price seems too good to be true. Counterfeit drives are modified to report more space to the system than they actually have: your computer sees a 1 TB flash drive, the first files copy without any issues, but anything written beyond the actual capacity will disappear or become corrupted without warning. You might only realize this months later when you actually need those files.

The check always works the same way: fill the drive with known data and then read it all back, verifying that everything is identical. On Windows, you can use the classic H2testw or GRC’s ValiDrive, which is faster because it checks random sample points across the drive. For Linux and macOS, there is F3 (Fight Flash Fraud), available via terminal.

Results of a capacity check on a USB stick where sectors written beyond actual capacity are marked in red — A counterfeit flash drive: data does not return correctly once you exceed the real capacity

Do this on an empty drive. H2testw and F3 fill up all the free space. ValiDrive only checks existing data, but if you suspect a drive might be fake, it’s better not to take risks: perform the check as soon as you buy the memory, before putting anything important on it. On a large SD card, this can take hours, but it is the only way to catch the problem before you lose your data.

How to run a reliable test #

Close everything else. Cloud syncs, antivirus scans, and ongoing downloads will skew your results.
Leave some free space. An SSD that is more than 80% full will run slower: this is normal.
Choose the right drive. On PCs with multiple drives, it’s easy to accidentally test the wrong one.
Repeat at least three times and take the average, discarding the first run.
Compare with your specific model. Benchmark values should be compared against what is rated for your unit, not against the fastest drive on the market.

With mechanical hard drives, there is one extra thing to keep in mind: they are faster on the outer edge of the platter, so results change depending on where the test file ends (the HD Tune graph shows this clearly). If your drive is slower than it was a few months ago or starts making new noises, check its health status immediately and back up your data.

FAQ #

Do benchmarks damage an SSD? #

If you do them occasionally, no: they only write a few gigabytes, while modern drives can handle hundreds of terabytes over their lifespan. Running them constantly is just useless, as the numbers won’t improve.

Why is my SSD slower than advertised? #

There are three common causes: the M.2 slot operates at a lower speed than the drive (for example, PCIe 3.0 instead of 4.0), the drive is nearly full, or the test has exceeded the high-speed cache. Additionally, manufacturer ratings are measured under ideal conditions.

The drive is healthy but the computer is slow: what should I do? #

If random read/write values are normal and health status is good, the problem probably isn’t the drive. Check your available RAM, startup programs, and CPU temperature: slowness is often caused by several factors combined.

How often should I check my drive’s health? #

A quick look every few months is enough, plus whenever you notice something strange: sudden freezes, read errors, or new noises. The key metrics to watch are reallocated sectors for mechanical drives and the remaining life percentage for SSDs. CrystalDiskInfo can also stay in your notification area and alert you automatically if anything worsens.

Is an external drive slower? #

It depends on the connection. A good SSD in a USB 3.2 or Thunderbolt enclosure will approach internal performance, whereas an external mechanical drive is naturally slow. The cable and port matter as much as the drive itself: plugging it into a USB 2.0 port is the most common way to accidentally slash your speed to one-tenth of its potential.