How to test the speed of an SD card on Windows 10

SD cards are tiny. The drama they cause is not. One day they’re “170 MB/s blazing fast,” and the next day your camera dump
is crawling along like it’s carrying each photo one-by-one up a staircase. The good news: you can test an SD card’s speed
on Windows 10 in a few minutes and figure out whether the card is slow, the reader is slow, or reality is simply refusing
to match marketing.

This guide walks you through reliable ways to run an SD card speed test on Windows 10 (beginner-friendly to power-user),
explains what the numbers actually mean, and shows you how to diagnose common bottleneckswithout turning your storage
drawer into a graveyard of “mystery slow” cards.

Before you benchmark: make sure you’re testing the SD card (not the bottleneck)

1) Use a decent card reader (this matters more than most people want to admit)

If you plug a fast UHS card into a bargain-bin USB 2.0 reader, your results will look like the SD card is bad when it’s
actually being handcuffed by the reader and port. For best results:

• Use a USB 3.x card reader and plug it into a USB 3.x port (usually blue or labeled “SS”).
• If your card is UHS-II, you need a UHS-II reader to get UHS-II speeds. Otherwise it’ll fall back to UHS-I behavior.
• Avoid USB hubs during testing, especially unpowered ones.

2) Check capacity type and file system (SDHC vs SDXC is not just alphabet soup)

SD cards come in capacity families. In practice, that affects formatting defaults and compatibility:

• SDHC (commonly 4–32GB) is typically FAT32.
• SDXC (commonly 64GB and up) is typically exFAT.

For Windows 10 speed testing, exFAT is usually the sensible default on SDXC cards because it handles large files better
than FAT32 (which has a 4GB single-file limit). If you’re using the card across cameras/phones, stick to what your device
expects.

3) Create a “clean-ish” test environment

• Close large downloads, cloud sync apps, and video editors that might be reading/writing in the background.
• Run benchmarks with the SD card mostly empty for a baseline, then again at real-life fill levels if you want.
• Back up anything important first. Some tools write test files (that’s the point).

Method 1: The easiest and most popular option (CrystalDiskMark)

If you want a straightforward SD card speed test on Windows 10, CrystalDiskMark is the “microwave popcorn” of benchmarking:
quick, consistent, and usually satisfyingunless you picked the wrong bowl (reader).

Step-by-step: running CrystalDiskMark on an SD card

1. Insert the SD card, then open File Explorer and confirm it shows up with a drive letter (for example, E:).
2. Open CrystalDiskMark.
3. Select your SD card drive letter from the drop-down (double-check this so you don’t benchmark the wrong drive).
4. Choose a test size. For SD cards, 1 GiB is a good quick baseline; 4 GiB or 8 GiB can better reflect sustained performance.
5. Set the number of runs (3–5 is fine).
6. Click All and wait for results.

How to read the results (without pretending you’re a storage engineer)

CrystalDiskMark typically reports a few types of tests. The key ones:

• Sequential Read/Write (SEQ): Best-case “big file” performance, like copying a large video file.
• Random 4K (RND4K): Small-file performance, like opening lots of photos or app data.
• Queue/Thread numbers: Higher queues can benefit SSDs, but SD cards often don’t scale the same way.

For most SD card use cases (camera footage, drone videos, game installs), sequential write is the star. If sequential
write is low, burst shooting buffers fill faster and long video writes can stutter. If random 4K is low, managing
tons of small files feels sluggish.

A realistic example (what “good” might look like)

Let’s say your results show:

• Sequential Read: 90 MB/s
• Sequential Write: 40 MB/s
• Random 4K Read/Write: Single digits MB/s (or less)

That’s not automatically “bad.” Many mainstream SD cards have modest writes compared to reads, and random 4K is often
unimpressive on removable flash. What matters is whether it matches your needs and whether it’s wildly below the card’s
class or advertised capability (more on that later).

Method 2: ATTO Disk Benchmark (great for seeing speed by file/block size)

ATTO is useful when you want to see how performance changes with different transfer sizes. SD cards can look decent in
big transfers but fall apart with smaller blocksATTO makes that pattern obvious.

How to use ATTO for an SD card speed test

1. Insert the SD card and confirm the drive letter.
2. Open ATTO Disk Benchmark.
3. Select the SD card drive.
4. Set a test file size (1–4GB is typical).
5. Run the benchmark and review read/write curves.

If your speeds only look “fast” at huge block sizes, that explains why small-file workflows feel slow. For most people,
ATTO is an optional “nice to have,” not a requirement.

Method 3: Built-in Windows testing with WinSAT (no extra apps)

Windows includes a command-line tool called WinSAT that can measure disk performance. It’s not as pretty as a GUI
benchmark, but it’s handy when you want a quick sanity check on a machine where you don’t want to install anything.

How to run WinSAT disk on your SD card

1. Type cmd in the Start menu.
2. Right-click Command Prompt and choose Run as administrator.
3. Run a test (replace E with your SD card’s drive letter):

WinSAT will output results that include read/write assessments. Treat it as a quick benchmark, not a full lab report.

Method 4: The power-user route (Microsoft DiskSpd)

DiskSpd is a serious storage benchmarking tool from Microsoft. If you want to simulate specific patternssequential
writes, random reads, different block sizes, different queue depthsDiskSpd can do it. It’s also how you avoid the
“Windows file copy cache made this look weird” problem, because you can request unbuffered I/O patterns.

Two practical DiskSpd tests for SD cards

Tip: Run these against a test file on the SD card (not your system drive). Close apps that might use the card.

1) Sequential write (big-file behavior)

2) Random 4K read/write mix (small-file-ish stress)

Don’t worry if those switches look like alphabet soup. The gist:
you choose a file size (-c), duration (-d), read vs write (-w),
block size (-b), and concurrency (-t/-o).
DiskSpd then reports throughput and latency.

Method 5: A real-world speed test (copy files and time it)

Benchmarks are controlled experiments. Real life is… not. A real-world test answers a practical question:
“How fast can I move my stuff?”

Option A: PowerShell timing (simple)

Copy a large file (like a 5–20GB video) to or from the SD card and time it:

Then calculate speed: Speed (MB/s) ≈ file size (MB) ÷ seconds. Run it twice; the first run can be affected by caching.

Option B: Robocopy for folders (more realistic)

If you’re moving camera folders with many files, Robocopy is a better mirror of reality:

Robocopy reports statistics and timing. Just remember: copying lots of tiny files is naturally slower than one giant file,
even on a “fast” card.

How to interpret SD card speeds (and why the box can be misleading)

Read vs write: your workflow picks the winner

• Read speed matters when you’re importing footage to your PC.
• Write speed matters when your device records video or bursts photos onto the card.

Sequential vs random: big files are easy, tiny files are chaos

SD card marketing often highlights sequential read speed because it’s the flattering angle. Real workloads can involve
lots of small writes (thumbnails, app data, metadata). That’s why a card can “benchmark great” yet feel slow in practice.

Speed classes are minimum guarantees (not your top speed)

Speed class symbols (like C10, U1/U3, V30/V60) are about minimum sustained write requirements for certain use cases
(especially video). They do not promise peak performance. A card can exceed its class by a lotor barely meet it.

Bus limits and readers: the invisible ceiling

Even if your SD card can theoretically go faster, your reader and port can cap it. UHS-II cards can’t show their full
potential without UHS-II hardware. And yes, that’s annoying. Welcome to the glamorous world of “standards.”

Why your SD card speed test on Windows 10 looks slow (common causes)

• USB 2.0 reader/port: you’re capped before you begin.
• UHS-II card in a UHS-I reader: it will fall back to slower mode.
• Quick Removal policy: Windows may reduce write caching for safer unplugging, which can lower write speeds.
• Card nearly full: many flash devices slow down as free space shrinks.
• Thermal throttling: long writes can heat up and slow down.
• Counterfeit or low-quality card: “big label number” can be fiction.
• Wear-out: older cards can develop errors and performance drops.

Practical fixes to improve SD card performance (or at least stop the bleeding)

1) Upgrade the reader first

If you only change one thing, change the reader. A good USB 3.x reader can turn a “slow card” into a “perfectly fine card”
in about five seconds.

2) Consider Windows “Better performance” mode (carefully)

Windows 10 typically defaults removable drives to a policy that prioritizes quick removal. Switching to “Better performance”
can improve write performance because it enables write cachingbut you must use “Safely remove hardware” before unplugging.
If you frequently yank the card out mid-copy, don’t switch.

3) Format appropriately (and don’t mix compatibility goals)

• For SDXC, exFAT is usually the best balance for Windows 10 and large files.
• If a camera demands a specific format, follow the camera’s guidance for reliability.
• Use Quick Format for routine resets; use a full overwrite format only when troubleshooting (it takes longer).

4) Verify authenticity if results are suspiciously bad

If your “fast” card benchmarks at something like 8–12 MB/s writes, especially from an unknown seller, it’s worth checking
for counterfeit capacity behavior. Tools like H2testw (Windows) can validate whether the card truly holds what it claims.

FAQ

Will benchmarking damage my SD card?

Flash memory has finite write endurance, but normal benchmarking now and then is unlikely to “kill” a healthy SD card.
That said, avoid running heavy write tests repeatedly for fun. There are better hobbies.

How many times should I run a test?

For a quick check: 3 runs. For more confidence: 5 runs and compare consistency. Large swings can hint at throttling,
background activity, or a flaky reader.

What speed should I expect?

It depends on the card class, bus type, reader, and the test pattern. Many everyday cards write far slower than they read.
Use your benchmark to confirm whether performance is in the expected neighborhoodnot necessarily the “best case” printed on the packaging.

Conclusion

Testing SD card speed on Windows 10 isn’t complicatedyou just need the right tool and the right expectations.
Start with CrystalDiskMark for a quick benchmark, use WinSAT if you want built-in simplicity, and step up to DiskSpd
when you need serious, controlled results. Most importantly: don’t blame the SD card until you’ve ruled out the reader.
It’s usually the reader. The reader is the plot twist in 80% of these stories.

Experiences and real-world lessons from SD card speed testing (the “stuff nobody tells you” section)

Here’s what tends to happen when people actually start running SD card speed tests on Windows 10especially when they’re
trying to solve a real problem (slow imports, choppy video recording, or a card that suddenly feels like it aged 40 years
overnight).

Experience #1: “My card is slow.” (It wasn’t.)
Someone plugs a premium microSD card into the first reader they can findoften a freebie dongle from a drawerand runs a
benchmark. The results are disappointing: maybe 25 MB/s read and 12 MB/s write. Panic begins. Then they try a different
reader (USB 3.x, reputable brand) and suddenly the numbers jump dramatically. The SD card didn’t change. The bottleneck
did. This is why “test with a known-good reader” is not a fussy suggestionit’s the entire game.

Experience #2: The “advertised speed” argument with reality
SD card packaging often features the most flattering number (usually maximum sequential read). But real workflows are
a mix: sustained write matters for video, and random performance matters for small-file workloads. People are often shocked
when a card advertised as “up to 170 MB/s” writes at 30–60 MB/s. That can be perfectly normal, especially if the card’s
video speed class only guarantees a minimum sustained write level (like V30 = 30 MB/s sustained write for video scenarios).
The lesson: compare the right benchmark metric to the right promise.

Experience #3: Copy tests can lie (or at least exaggerate)
A simple “drag-and-drop” copy feels real, but Windows caching can make the first few seconds look amazing, then performance
falls off a cliff. People see a copy dialog spike to a huge number and assume the card is flyinguntil the average speed
settles much lower. That’s why running a controlled benchmark (CrystalDiskMark or DiskSpd) plus a real-world copy test is
the best combo: one tells you the potential, the other tells you the lived experience.

Experience #4: Many small files are the SD card’s natural enemy
Photographers and drone users often copy folders with thousands of files (RAW photos, thumbnails, sidecar files, previews).
Even on a fast card, copying many small files can feel slow because the overhead isn’t just “speed”it’s file system
operations, metadata updates, and lots of tiny I/O requests. The “fix” is sometimes procedural: zip a folder before moving,
copy fewer at once, or accept that small-file performance is where SD cards tend to be merely okay.

Experience #5: The “sudden slowdown” that’s really “nearly full”
A card that benchmarks fine at 10% used can look worse at 85–95% used. Some flash storage slows down as it runs out of
easy-to-manage free space. The user experience is often: “It used to be fast, then it got slow.” Running two benchmarks
one empty-ish and one at normal usagecan explain the mystery and help you decide whether you need a larger card or a
different model designed for sustained performance.

Experience #6: Counterfeits don’t always scream; sometimes they whisper
Counterfeit cards aren’t always obviously broken on day one. Some “work” until you exceed their real capacity, at which
point files corrupt or disappear. Others are simply built from slower memory. A benchmark that’s wildly below expectation,
especially paired with inconsistent behavior, is a strong hint to validate the card with a capacity test tool before you
trust it with important footage. People who learn this lesson once become evangelists forever (and also slightly suspicious
of too-good-to-be-true deals, as they should).

Experience #7: The best outcome is not a big numberit’s confidence
The point of testing isn’t bragging rights. It’s knowing whether your SD card is fit for your job. If you need reliable
4K recording, you care about sustained writes. If you just want quick photo imports, you care about reads. A good test
routine gives you confidence: this card, this reader, this Windows 10 PCtogetherbehave the way you need them to.
That’s the real win (and it’s way more satisfying than arguing with a marketing label).