1. osu! forums
  2. osu!
  3. Gameplay & Rankings

posted
So I've been seeing people discussing about using ps/2 instead of usb and I decided to investigate this myself.
Upon testing with usb my results were 8ms of delay at a scan rate of 125hz.
Plugging into ps/2 via a usb converter has given me 16ms of delay at a scan rate of 62.5hz.
I don't have a native ps/2 keyboard to test with but I'm going to be sticking with using usb. I also measured the input delay through osu! using a high speed camera and have seen that it is also reporting roughly 8ms of input delay. The setup I used for that was binding k1 to num lock and recording the led turning on with k1 in game lighting up on the key overlay. I'm currently using windows 10 and I believe that usb overclocking is not compatible, so I haven't bothered to check and test that.

This whole thing is honestly whatever and you can take it however you'd like.

Edit: cleaned up some stuff (x2)
posted
osu! assumes you adjust to the input latenchies you play with.

uo is meant to adjust the hitsounds to the song timing points, audio offset between the wav hitsounds and song mp3s. if you use it for adjusting to the input lag of your keyboard, your hitsounds will be out of sync (doesn't matter at all if you play without hitsounds).
posted
You're thinking of local offset.
posted
so are usb -> ps/2 adapters worse than just using a usb port for keyboards, or better?

been trying to figure this out for years
posted
It's worse
posted
my life is a lie.
posted

Dexus wrote:

It's worse
Shit
posted

Dexus wrote:

It's worse
very strange
posted
Tbh this has always been my guess since my original keyboard I played osu! with used PS/2. Then I switched to an USB-keyboard and I found myself hitting significantly early because I was still accustomed to the apparently bigger delay of the old keyboard.
posted
apparently hitting shit 5ms too late with usb

back to ps/2
posted
Thank you very much for this even if I have this one

also plz don't adjust UO based on input... it's meant for syncing audio with the map, not for micro time travel input lag compensation.
That picture is perfectly accurate but only if you play without hitsounds (I think without music too but my brain is too small to process this)
posted
https://ahtn.github.io/scan-rate-estimator/

You can test it with this, it's not 100% accurate as it doesn't include all latency factors, but it can give you a general idea of how much latency is introduced. Mash all the keys on your keyboard as fast as you can until the shortest key press goes as low as it can.

Total latency from the keyboard by itself would be something like (polling rate) + (scan interval)

In my case I have 1000hz polling + 125hz scan interval which is 1ms + 8ms = 9ms of latency from the keyboard using usb alone.

I personally bought this keyboard: http://www.cherry.de/cid/RealKey_Techno ... witch=true
As it has a scan interval of less than 1ms, resulting in 1 to 2ms total latency from the keyboard through usb. I'm getting it because my current keyboard has black switches and I'd like to go back to red switches.

The reason why I was getting poor performance on the ps/2 port is I believe my ps/2 port is a fake one that converts the signal to usb. There's apparently real PS/2 ports and fake ones. As for how to check that, I don't know.

I tested using a DK2108SZ and FC210TP. I will test PS/2 again once I get my new keyboard to verify if it's simply my cheap keyboards causing the issue.
posted
thanks for this thread op, now that i've removed 8ms of latency from my keyboard i'm well on my way to becoming the #1 osu player of the world.
posted
As far as I'm aware, PS/2 has a maximum scan interval of 2.7 ms, but since it uses interrupts, I'm not sure if this only applies to key mashing or if it adds latency to single keypresses (and while I am curious, I'm not sure how this could be tested).
I'm also not really sure how you'd determine the debounce latency of a keyboard without using specialized equipment or directly comparing to other keyboards using something like Bloody's PK tester, but what I can say is that it plays more of a role in latency than the polling rate. Having a slower polling rate will simply create more variability and inconsistency in your key response.

And also: I have a cheap LK optical keyboard that happens to be native PS/2, and I can easily say it's waaay more responsive than any Cherry MX keyboard I've ever tried (USB or otherwise). It might be worth looking into optical keyboards (or similarly responsive keyswitches) if you're looking to reduce as much latency as possible.
Please sign in to reply.