My main work laptop sits on my desk, with my peripherals all conveniently plugged into the USB ports on the left-hand side. That frees up the other side for USB file transfers, but, for some unknown reason, those ports have always been significantly slower.

Up until now, I never bothered to prove it, but I’ve finally decided to make it official. I used the same USB flash drive on both sides and found sequential writes were over 70 MB/s on the left and only 24 MB/s on the right.

This was a relatively new laptop with USB 3.0 ports, so USB 2.0 was spared the blame. The culprit turned out to be one Windows setting that quietly became the default when Microsoft assumed users weren’t ejecting flash drives safely.

I thought USB 2.0 was the obvious explanation

An old laptop proved me wrong

I used CrystalDiskMark to assign some numbers to the speed difference. It measures storage read and write speeds, and I kept its settings the same across all tests with three passes and a 1 GiB test file. I also used the same USB drive and moved it between ports.

An older HP 250 G7 I had lying around gave me a clear example. Its USB 3.0 ports all carried the SuperSpeed (SS) marking, while its USB 2.0 port that topped out at 480 Mbps was unmarked.

On its USB 3.0 ports, my first flash drive managed around the 131 MB/s mark for sequential reads, and 78 MB/s writes. Moving the same drive to USB 2.0 dropped those figures to a predictable 37 MB/s and 22 MB/s.

Back on the USB 3.0 ports, the faster speeds returned. Nothing is surprising about that, but it shows how accidentally picking the wrong port can really limit a drive’s speed.

But on my main HP Victis laptop, the issues with transfer speeds were harder to explain. To test, I used a different, known and reliable USB 3.0 drive — a 32 GB Kingston DataTraveler. Both connections reported SuperSpeed, yet sequential writes reached 74 MB/s on the left, and crawled at 24 MB/s on the right. Both sides stayed around 97 MB/s for reads.

These results didn’t match what I’d seen from the old laptop’s USB 2.0 ports.

The results were repeatable, and the right side always brought the slow write speeds back up. Only this time, the USB standard alone just couldn’t explain it.

OS

Windows

Developer

Crystal Dew World

CrystalDiskMark is a simple and popular open-source benchmarking tool used to test the read and write speeds of storage drives like SSDs, hard drives, and USBs.

The USB connection wasn’t telling the whole story

I went searching for the USB controller to narrow down the search

For this investigation, I used the tool USB Device Tree Viewer to see how each port connected together internally. I’d initially suspected that the slower side might use a different controller or even share a connection through an extra hub.

The tree soon made short work of my theories:

• Both connections reported SuperSpeed.
• Both sat beneath the same Intel USB Controller and root hub.
• Neither showed an additional hub between the drive and that root hub.

Sharing the same controller didn’t guarantee identical performance, but it put a massive dent in my potential explanations. The right-side connection wasn’t falling back to USB 2.0, and I could no longer blame a different controller.

The biggest clue then came unexpectedly when I tried to eject the USB drive from the left-side ports. Out of habit, I always eject external drives before physically removing them. This time, however, the drive refused to eject cleanly.

That warranted further investigation, so I opened the Kingston drive’s properties in Device Manager and checked the Policies tab. Better performance was selected. I hadn’t checked that setting before, so I compared it with the right-side connection, which turned out to be set to Quick removal.

That was strange. It was the same flash drive, but Windows wasn’t applying the same removal policy in both locations. As it turned out, my friend who previously owned the laptop remembered choosing better performance to make copying movies faster. That finally explained the speed differences and why the setting had been changed in the first place.

In both cases, Enable write caching on the device wasn’t ticked, but I’d still found a difference in the removal policy and now had something specific to test.

OS

Windows

Developer

Uwe Sieber

USB Device Tree Viewer is a Windows utility that maps connected USB devices, hubs, and controllers in a detailed tree view. It shows connection speeds and device information to help troubleshoot USB issues.

One Windows setting brought the write speeds back

Microsoft assumed no one had been safely ejecting USB drives since 2018

With the Kingston USB drive plugged back in on the right, I selected Better performance and ran the same benchmark. Like with the left-side ports, I left the device write-caching checkbox untouched.

Sequential writes jumped to 71.96 MB/s, extremely close to the 74.29 MB/s I’d measured originally on the left. Reads still stuck around 97 MB/s, but the supposedly slow port suddenly looked much more capable.

To make sure the setting was responsible for the speed differences, I switched it back to Quick removal and repeated the test. Write speeds immediately dropped back to almost where they’d started:

Changing the removal policy from the Windows 11 default of Quick removal to Better performance explained the slowdown, and better still, it was repeatable.

Microsoft actually made Quick removal the default policy for USB drives in Windows 10 version 1809, trading the speed of write caching for the convenience of unplugging an idle drive without having to use Safely Remove Hardware. The two policies make a big difference in write speeds, but come with their own set of drawbacks.

Quick removal prevents Windows from caching disk writes and allows physical devices to be pulled from the USB port without safely ejecting them.

Better performance lets Windows cache writes by temporarily holding data before it reaches the drive. For this setting, the ritual of safely ejecting the drive actually allows those outstanding operations time to finish up.

Since old IT habits die hard, and I already safely eject all my drives anyway, that little bit of extra responsibility wasn’t a change for me. Quick removal, though, still makes sense if you’re used to just unplugging the drive as soon as the transfer hits 100%.

I’ll check Windows before blaming the port again

The connection speed was only a small part of a larger explanation

I’d started all this off expecting to uncover a slower USB connection on my laptop. Instead, matching the removal policies brought the transfer speeds on both sides into alignment.

The older HP showed me what speeds I could expect from USB 2.0, and why the USB standards are still something that needs checking during any PC USB troubleshooting.

Next time a drive starts feeling slow, or transfer speeds become unpredictable, I’ll check its negotiated connection speed and Windows removal policy before deciding the poor USB port itself is to blame.