PS/2 devices are not polled or sampled, they push their data and on each byte it triggers an interrupt. Higher interrupt rates - higher congestion.
But I'd imagine keyboard's IRQ1 would have higher priority than mouse's IRQ12, so the issue should boil down to slow handling of mouse interrupts, or interrupts in general. Or a bad/suboptimal mouse driver, although I can't advise on what to check, it was way too long ago...

I think I saw you somewhere.
Have a good evening!

Armacadia wrote on 2026-03-14, 14:00:

Some more testing, if I move the mouse very fast in various directions while also pressing keys (two at once are enough), Windows beeps the PC speaker, several times in rapid succession. Like how it does when I press "too many" keys at once (for the matter, I don't understand how this works either - for example, I can hold down asdfh at once and have the last key repeat, but asdfg will make it beep instead; likewise, I can hold ftg but not gyh, etc.)

I think this happens because of ghosting. Some key combinations are not possible on generic keyboards, unless they're NKRO (N-key rollover) capable. PS/2 or USB does not matter.

A notable example from my experience would be trying to press both UP and RIGHT while holding SPACE for firing in some games. On generic, non-NKRO keyboards, I would be moving only towards the first direction key I held, with the second direction key causing the system to beep.

So, I found out this was happening on my windows xp box while playing doom 3, the character would keep moving sometimes after I let go of the movement keys. Turns out the keyboard would not register a keydown/keyup event if the mouse was actively moving - or at least it would be delayed until the mouse stopped. I could easily replicate it in notepad by holding down a key so it repeatedly types a letter, moving the mouse would stop new letters from being typed.

It went away after I changed the input buffer length in the mouse advanced settings from the default 100 up to the maximum of 300. The PS/2 mouse port is an extension of the keyboard controller so it’s possible for there to be some interaction between the two. My guess is the higher polling rates fill up the packet buffer in the driver and the keyboard has to wait.

Interestingly on my older 98SE box, it works at 200hz without any extra adjustments.

vstrakh wrote on 2026-03-15, 10:25:

Or a bad/suboptimal mouse driver, although I can't advise on what to check, it was way too long ago...

I don't believe this mouse came with any special driver, or needed one at all. I've got it without its original packaging; I did try installing ioCentre, which adds its own "filter" driver, in turn allowing to remap the wheel click function (as long as the software is running) and seemingly does nothing else. (Also having googled around a bit more, seems like those were still being released well into 10s, specifically the PS/2 ones as there's also the USB variant. Interesting.)

LSS10999 wrote on Today, 06:37:

I think this happens because of ghosting. Some key combinations are not possible on generic keyboards, unless they're NKRO (N-key rollover) capable. PS/2 or USB does not matter.

Ah, didn't know this phenomenon has a proper name. Having taken apart quite a few keyboard over the years, I somehow never thought of that.

NeoG_ wrote on Today, 10:36:

It went away after I changed the input buffer length in the mouse advanced settings from the default 100 up to the maximum of 300. My guess is the higher polling rates fill up the packet buffer in the driver and the keyboard has to wait.

I did try this and for my case I see no difference at all, it must be something else. Now that I think about it, the PS/2 keyboard alone (three different keyboards), USB mouse connected instead, never felt particularly responsive on this system, which I used to discount on not-100%-stable framerates in games (that still were reasonably high), then on suboptimal input systems in game engines, then on my own "brain lag", until I wasn't sure about anything anymore. Could as well be the DVI to VGA converter lagging it further (I don't use it anymore, direct VGA to VGA now), though I think that would be too minimal to be noticeable at all, or pretty much anything else. I think I now can see the mouse being kind of erratic at 100 Hz too when I shake it too much, just with a way higher threshold (and I'm not sure it's not the sensor malfunctioning).

animacrxss wrote on Yesterday, 13:12:

I think I saw you somewhere.
Have a good evening!

Uh... Hi?