I was buying the fastest SD cards for the wrong reason

I've bought a huge number of SD cards over the years, and my method for doing so hasn't really changed. Find the biggest numbers on the packaging, and buy that; faster is always better, after all.

But then I noticed that some of my fast cards were making my device feel slow, and some cheaper cards were doing just fine. Which made me realize that the number I was defaulting to wasn't always doing what I thought — and you're probably making the same mistake.

What actually makes an SD card feel fast

Sequential read and write isn't everything

The big number you read on the box is typically the sequential read/write speed, stuck way up front because it uses the most marketable numbers.

That measures how fast the card can handle one long, continuous stream of data. It's not a lie, mind. Sequential read/write is important, but only in certain scenarios, such as recording video or a single large file copy. And the reality is that most of the time, that's not how we use memory cards.

Most of the time, we're adding small files en masse, such as installing and booting an operating system, installing and loading apps, recording short clips, burst photography, and so on. That's when you really want to check for random read/write speeds, measured in input/output operations per second (IOPS).

The difference between these two types of data processing is why even though your SD card looked like it had wonderfully fast storage, it still feels slow and underpowered; the numbers aren't entirely related.

The spec that measures this isn't always easy to find

You're looking for the Application Performance Class

I must admit that SD card manufacturers are much better at promoting the Application Performance Class these days. It's easier to find on packaging than ever before, and you'll typically find it on the SD card itself these days, shown as A1 or A2.

• A1: Minimum 1,500 random read IOPS / 500 random write IOPS
• A2: Minimum 4,000 random read IOPS / 2,000 random write IOPS

It gets more confusing because the Application Performance Class only guarantees a minimum sequential read/write speed of 10MB/s, which is on the lower end of the scale when it comes to storage cards.

Don't forget about command queuing though

A2's drastically higher IOPS also depend on a feature called command queuing, which basically relies on the host device. If the host device doesn't support command queuing, it'll never hit those speeds, and you may have wasted your money on an expensive card.

The problem with this is that most devices don't really make it clear if they support command queuing, which makes it even more difficult to figure out. It's not the card itself that needs to provide the support, either. That means you can't just throw it in a card reader and check out its specs on your PC or laptop.

There is one handy and slightly comforting caveat to all of this: a high-quality A2 card from a reputable memory card brand will usually outperform other A1 cards, especially if they're from a lesser-known or "bunch of letters" brand. In that, A1 and A2 are a performance floor rather than an outright guarantee.

Sequential speeds aren't a con

There are other memory card bottlenecks, too

That doesn't make sequential storage pointless. It's still a spec you need to pay attention to, and it's all made a bit easier because the top-performing memory cards mostly come equipped with class-leading specs across the board.

That's generalizing, and you'll find exceptions, which is why paying attention to the specs is so important.

For example, when it comes to Video Speed Class, you're looking for the V30, V60, or V90 label, which guarantees a minimum sustained write of 30MB/s, 60MB/s, and 90MB/s, respectively.

• V30 covers basic 4K and high frame-rate 1080p video recording
• V60 and V90 exist for burst RAW photography and high-bitrate or 8K video and similar high performance output

However, even a card with all the right labels can be held back by something else: the bus interface.

• UHS-I tops out at 104MB/s
• UHS-II can reach up to 312MB/s
• UHS-III theoretically can reach 624MB/s but has barely any level of consumer adoption

If you use a UHS-II V90 A2 card in a UHS-I slot, it's capped at UHS-I speeds, and there isn't anything you can do. It's not the card underperforming; it's a complete mismatch of tech.

So, what should you actually check before buying a new memory card?

Really? Check the slot of the device you're trying to buy for, then match to that. It's the "easiest" way to figure out what you really need from your memory card, especially if you're only using it in that one device.

Manufacturer specs are the best place to start, and they should list something like "SD/SDHC/SDXC UHS-II compatible," as it's an important spec and selling point. It gets a bit trickier with certain devices that are so forthcoming with specs; you'll often find smartphone specs listed as "microSD up to 2TB", and that's if there is support at all.

I'd hesitate to push you towards using AI models to search for this sort of information. On the one hand, AI models are excellent at parsing a huge volume of information for niche data like this. On the other, it's the exact sort of information an AI model might hallucinate just to "get the job done."

Failing both of those, the empirical route: put a known-fast card in, run a benchmark, and see where the number plateaus. If a UHS-II-rated card caps out around 100MB/s regardless of what you throw at it, you've found your ceiling.