Xhci may be the culprit.

[Vocaleyes]

Hi all.

Very quickly, i believe this all started when echi was phased out and xhci became the norm. Ehci used to handle usb 2.0 ports but now xhci handles all ports. The timeframe fits, and the feeling of current 2.0 ports being floaty points to this change.

When my mouse worked, i could only use 2.0 ports, as the 3.0 ports feel like.. well, how they do today.
This is the only plausible explanation i have left, luckily it fits.

Feel free to take this and run with it using AI to try to poke holes in it.

But as it stands, i believe xhci is the reason.


For anyone that can’t be arsed to look it up, ill paste it here;

The xHCI mouse feel is noticeably worse than EHCI because xHCI prioritises high-bandwidth data handling over raw polling timing, introducing scheduling structures that alter the physical response of a mouse. EHCI (legacy USB 2.0) was built strictly around hardware-driven interrupts, while xHCI (USB 3.0+) is driven by software-managed packet scheduling. This shift introduces fundamental performance differences that alter cursor control

1. Packet Batching vs. Immediate Reporting
EHCI: When your mouse sends a coordinate update, the legacy EHCI chip fires an instant hardware interrupt. The data moves immediately to the CPU with zero mid-route handling.
xHCI: To maximize data transfer efficiency for heavy storage drives, the xHCI controller groups smaller data packets together into "bursts" before processing them. Because your mouse sends tiny packets, it sits in an xHCI memory buffer waiting for the burst threshold to clear. This creates micro-variations in packet arrival times, resulting in a heavy, disconnected, or floaty feeling.

2. The Initial Motion Delay (Idle Latency)
EHCI: The data lanes remain wide open and continuously active regardless of whether you are moving your mouse or holding still.
xHCI: xHCI introduces native Link Power Management (LPM) deep in the hardware layer to reduce power consumption. If you pause your mouse for even a split second (such as holding an angle in a game), the xHCI controller shifts into a micro-sleep state. The very first millimeter of your next hand movement suffers a brief, sluggish processing lag as the controller wakes the port back up.

3. Bus Contention and Endpoint Sharing
EHCI: Legacy systems featured separate, isolated controllers for different port groups. Your mouse had its own clear line to the processor.
xHCI: Every single port on a modern motherboard routes into a singular, massive xHCI hub topology. Your mouse data packets are forced to share the same controller queue with high-bandwidth devices like audio interfaces, webcams, Wi-Fi 7 chips, and RGB lighting. This shared congestion creates tracking jitter because your mouse updates are constantly fighting for execution priority.

4. Dynamic Micro-Scheduling (Jitter)
Under EHCI, the time gap between mouse reports is incredibly rigid (e.g., exactly 1.0ms apart for a 1000Hz mouse).Under xHCI, software-level drivers control the polling cycle. This causes the reporting interval to fluctuate dynamically (e.g., 0.8ms, then 1.2ms, then 1.0ms). While the average polling rate still calculates to 1000Hz, this micro-fluctuation causes the cursor to track unevenly across your display pixels, leading to a noticeable loss of crisp, 1:1 tactile precision.

[ablemor]

So this could be tested with a PCIe EHCI expansion card, couldn't it? I guess a USB hub wouldn't help because the data would still end up going into the xHCI anyway, but I don't know.

[Vocaleyes]

So this could be tested with a PCIe EHCI expansion card, couldn't it? I guess a USB hub wouldn't help because the data would still end up going into the xHCI anyway, but I don't know.

Exactly what im about to do. However… not standard pcie cards of which ive tried 3, but specifically a pcie usb 2.0 card with onboard ehci handling. This way will bypass the xhci layer like the other pcie cards never did.

Literally looking at them now, but this one looks promising.
MEO PCI-E to 4 Ports USB 2.0 Converter Card (MCS9990 Chipset)

(Haven’t tested, buy at your own risk or wait for me to and i’ll let you know)

[ablemor]

It would be really great if you could try it out, I'm curious!

[Vocaleyes]

It would be really great if you could try it out, I'm curious!

Ordered. 4-5 days. I am for the first time in a long time, quite hopeful.

[ablemor]

I found one for $3 right here in my country, I can't even believe it! Already ordered it a big bag of chips costs more than this. Curiosity is killing me, hopefully it arrives by Friday!

Might be a waste of money, might not we'll see.
btw check this blurbuster post:
viewtopic.php?t=11135

[agendarsky]

i own both xHCI and true eHCI usb 2.0 card powered from psu. Don't bother trying that
that one https://www.kaufland.sk/product/556383540/

[Vocaleyes]

i own both xHCI and true eHCI usb 2.0 card powered from psu. Don't bother trying that
that one https://www.kaufland.sk/product/556383540/

That looks like the same chip, still gonna give it a test for testing sakes. What was your experience with it, did you feel any difference between mobo 3.0 and that card's 2.0?
Also what were you trying to diagnose specifically?