Re: OT: slow USB drives

Apr 14, 2026 Last reply: 3 months ago 9 Replies


I have been using two USB3.2 memory sticks as secondary back-up for


> some of the files on my PC. I realise these are not recommended best
> practice for archival storage. However, two of the sticks are
> extremely slow. These are Kioxia. Two others (SanDisk) are by
> comparison very fast. Is SanDisk inherently a better product, or could
> my settings be at fault? Is there a reliable way to test a USB drive?
> I believe it is better to reformat than to delete. Is this correct? I
> would add that there is a very large number of small files, and I know
> this takes longer.

USB 3.2 is just the connection speed. It doesn't mean very much in real life, as you can get slow sticks and fast sticks.

How do I recognise a fast stick then?

Look for the read write speed numbers, as Theo suggested.

Another solution is to get an SSD and a caddy. That will be much faster than the memory sticks. The caddies are only a few £££ but the price of

256GB SSDs seems to have shot up to £40 or so. (My fault for using AI so much.)

You haven't said how much data you are backing up?

I meant in the shop.

I've actually got a portable SSD so maybe I should be using this.

About 5-6 GB so there is no need for a high capacity or expensive memory stick.

It's a specialist item, so if you buy it in a supermarket it's pot luck what you end up with. They are available on Amazon, and you should be able to see the stats for some of them.

Yes. :)

That's not a lot. What backup software do you use? You should only need to be copying any files that have changed.

A good trick is to scan Amazon review photos for CrystalDiskMark screenshots. The green bars are quite distinctive, and they tell you more than the official speeds do. Although nowadays 'IOPS' is the same as the 4K random speed (IOPS*4096 = 4K read/write speed) so that's a useful figure.

Although I'm not sure if they fudge it by quoting read-IOPS (which are easy) instead of write-IOPS (which requires decent hardware to achieve).

Bear in mind that small devices can be slower (perhaps use older generations of flash; if there are multiple flash chips to achieve larger size they can be driven in parallel whereas a smaller device might only use one flash chip). So it may be a false economy to buy the smallest size that will fit your data.

Theo

It's best to test USB drives before you first use them. The claimed 2Tb device I just bought actually has a capacity of about 30Gb. The manufacturer tells me that they are sending a real 2Tb device to replace it. I'm not holding my breath.

It's quite possible that your USB 3.2 devices are actually USB 2 and not full size either.

I used Gibson ValiDrive for testing.

IIRC Validrive is a random spot check -

formatting link

I'm sceptical about claimed specifications.

An example is that many USB2.x devices might have all the USB2.x protocols but not actually operate at High Speed.

I am not sure if this is also true of USB3.x where the number may be a protocol yet not SuperSpeed. Happily proven wrong as always.

Unless the device actually claims the relevant SuperSpeed I would assume it isn't!

The number is the version number of the specification. ie it's perfectly OK to make a USB 2.0 Full Speed (11Mbps) device - many keyboards are, because they don't need to go any faster.

But people use it as a shorthand for the maximum speed of the specification

- ie USB 2.0 at 480Mbps, USB 3.0 at 5Gbps. You could make an 11Mbps Full Speed keyboard using the USB 3.0 specification, but that's typically not what people mean by saying USB 3.0.

After 3.0 it all gets complicated (5/10 Gbps lanes - Gen1/Gen2/etc, x1/x2 pairs, Thunderbolt etc) so you end up with things like USB 3.2 Gen1x1 which is functionally the same as USB 3.0 Superspeed but built to match an updated specification document.

But that's a separate thing from the performance of the device. You can put a mechanical HDD behind a 5Gbps interface but you won't get near 5Gbps out of it - maybe it will for a small cache, but any time you hit the platters it'll be much slower.

Theo

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required