Old controllers have low lag?

[shmupsrocks]

I was looking at the input lag measurements here:

https://rpubs.com/misteraddons/inputlatency

and they all seem to be modern controllers. Do older console controllers all have similar and very low lag so that they aren't worth testing?

[bigbadboaz]

It only became an issue in more recent times with wireless and USB controllers. I can't speak to statistical measurements on classic controllers, but I'm reasonably sure that in general they didn't have extra protocols/processing getting in the way to a significant extent. Hopefully someone else can chime in with real details.

[gray117]

Nice list.

Old controllers are just like a electrical switch for the most part so pretty much instantaneous. There are some adapter tests in there that I assume is adapter plus legit pad.

5ms or less is fine. Most usb devices should really be doing 10 or under. It really shouldn't be a problem, but some not good fighting stick designs and (unpatched) adapters really let things down so now people are testing and reviewing this.

Most input lag will still be attributable to I) software with deferred and buffered renderers ii) screen choice... But yeah annoying if a bad usb device is adding 15+ ms to the chain....

[Sumez]

Classic controllers most often work via a parallel to serial converter handled on a single very simple IC (or a few of them chained).

If you wanna be technical there's a tiny bit of overhead in reading the button input as a serial value (ie. you'd read one bit at a time, instead of all at once), but we're talking nanoseconds at worst, not miliseconds. By all standards this would be considered 100% lag free, not just "low lag". It's not comparable to the overhead introduced via USB or wireless controllers at all.

[orange808]

Unless the manufacturer promises something better, you should expect standard USB polling (8ms) from every USB device. In some situations, you can set it manually. Standard USB polling time is half a frame of latency at 60fps. (It's a full frame at 120Hz) That's a scandal. It's awful. It's rubbish. Half a frame of lag isn't okay or good. "Under ten milliseconds" isn't a reassuring promise at all.

Four milliseconds is also completely unacceptable. One millisecond is the only acceptable polling time and that could be theoretically improved with a future revision of the protocol.

You don't want to stack more than one millisecond of controller lag with your display lag and miscellaneous latency. Depending on the hardware and software configuration, your PC may also add some lag.

Remember: when you miss a deadline, your input waits "on the bus" until the next time the software can poll the input. So, the worst case scenario is the sum of your controller lag and all the other latency in the chain. Assuming no additional hardware lag and software that polls every frame, that would be the controller lag plus one more frame of lag, but the additional overhead (beyond controller lag) can be more.

[bigbadboaz]

I'm totally with you in wanting control to be the way that it should be.. but USB/wireless have been the standards since the 360. Quality comtemporary fightsticks consistently test around 4ms, and 2.4G is generally expected to come in in the low 20s. What you're expecting simply isn't happening in the modern gaming paradigm.

[shmupsrocks]

This is interesting stuff. So the lowest lag general purpose USB controller is any classic controller connected to the new Reflex Adapt converter which has 1ms polling?

https://misteraddons.com/products/reflex-adapt

[bigbadboaz]

Hmm. I'd be curious about legit Saturn controller + that adapter. But the Saturn pads definitely had a more complex microcontroller inside them than pads from even a generation or two before. So I wonder if they themselves did have a touch of latency.. anyone have specs?

[orange808]

I'm totally with you in wanting control to be the way that it should be.. but USB/wireless have been the standards since the 360. Quality comtemporary fightsticks consistently test around 4ms, and 2.4G is generally expected to come in in the low 20s. What you're expecting simply isn't happening in the modern gaming paradigm.

The convenience of wireless comes with trade offs. ¯⁠\⁠_⁠(⁠ツ⁠)⁠_⁠/⁠¯

FWIW, my post isn't related to wireless. I was strictly chiming in on the death of serial ports and USB.

I've always understood wireless to include a certain amount of latency and there's usually some kind of crosstalk or interference that can also affect reliability.

You're quoting complete touch to glass response input lag, there. Good controller input lag is about getting the input sent to the game machine quickly. After that, there are a lot of variables.

Also, I prefer a different metric (besides a millisecond value) to convey complete touch to glass input lag for USB/wireless controllers versus original hardware OEM controllers (running on a real console). It's better when we use a lot of samples and measure the amount of "on time" frames. A percentage of "on time" frames (@~60Hz) gives you a better idea of overall performance for our needs with "retro" games, because (as I said before) the timing of inputs will vary; there is a best and worst case scenario, so we want an average. Of course, lots of samples will also sniff out wireless controllers that drop input signals, because they don't maintain good signal.

Obviously, new games are designed for new controllers, so that's a bit different; I always want the best lag, but new games shouldn't be asking for "twitch" responses. Developers (well, good ones) think about lag (from user hardware) when they are designing and polishing new software.