Input Lag - How can "CHECK" if present in Retroarch core / Standalone emulator?

DjDiabolik

@sleve_mcdichael said in Input Lag - How can "CHECK" if present in Retroarch core / Standalone emulator?:

@DjDiabolik there's an app you can get on your phone called is it snappy, it'll record video in high-speed mode and you can mark two points and it will tell you the time and number of frames in between. so you can use this to measure the time for ex. in between depressing the button and when the character reacts on-screen, and then do the same with original hardware and compare.

mmmmmmmm... in i understand i need to install this ?
https://play.google.com/store/apps/details?id=io.gosnappy.snappy&hl=en_US

i thinks it's not the correct app :)

If I understood the concept correctly anyway. You tell me I have to shoot the entire monitor plus arcade stick scene and that way I can visually calculate if there is a delay from pressing the button to my character's movement in the game?

mmmmmm... But then how can I compare everything with the original hardware? I should have a street fighter cabinet at home that runs on original hardware and not on an emulator.

At the "Desktop Level" it's can be installed an app "jstest-gtk" and visually I can see that, at least at the OS desktop, there appears to be no delay from key press to visual output on screen.
but if this happens when emulationstation or, even worse, retroarch cores are running, it is not easy to understand.

barbudreadmon

@DjDiabolik said in Input Lag - How can "CHECK" if present in Retroarch core / Standalone emulator?:

I can visually calculate if there is a delay from pressing the button to my character's movement in the game?

You seem to have some kind of major misunderstanding about input lag : there is a delay, even if you were playing on original hardware, there would generally be a delay, input lag is part of gaming.

A commonly accepted answer is that emulation adds a minimum of 1 frame of input lag by design, but that's not always true though (iirc "Soukyugurentai" had 7 frames of input lag on original hardware, while it has significantly less input lag in MAME or Kronos).

If you want to have the same input lag as the original hardware, then you want fpga hardware, fpga hardware is the only thing that can reproduce original hardware input lag perfectly by design.

If you simply want to shave off frames of input lag, including frames of input lag existing on original hardware, you can use retroarch's runahead or pre-emptive frames features. Note that some cores are more compatible than others with those features, FBNeo is supposedly the only arcade emulator with full support for this.

sleve_mcdichael

i thinks it's not the correct app :)

You're right, that doesn't look right. Maybe the one I saw was just for iOS.

mmmmmm... But then how can I compare everything with the original hardware? I should have a street fighter cabinet at home that runs on original hardware and not on an emulator.

I mean, ideally you could find one in a bar or bowling alley or something, but yeah, you'd have to compare it against a real one to learn any useful.

DjDiabolik

@barbudreadmon wait. so delay and input lag aren't the same term?

There is something to explain. I have to explain why I took the Street Fighter game as an example.
There's simply to made an "Hadoken" (D,DF,F + P it's problably most famous move on game lol ) you know :) well I try and try several times and the game seems to ignore the command or the move happens 1 out of 3 times.

this leads me to think that on my current system there may be some type of delay or some input lag that may cause some movement to be "lost".
I don't know if I can explain the concept...

However, there is one thing to point out. which visually everything seems to work very well because if I press the punch button you can say that Ryu hits almost immediately but then everything gets confused when I try these "combos"
Obviously I'm trying player vs player with the p2 which remains still like a stockfish lol :)

it gives me the idea that the game doesn't take the commands well or that something is ignored but I don't think there's a way to test everything individually.

@sleve_mcdichael Exactly... but I would have to try the two things right away to have a real comparison. that is, maybe the gaming experience on real hardware and on a cabinet is completely different, maybe even using different brand arcade sticks.

barbudreadmon

@DjDiabolik said in Input Lag - How can "CHECK" if present in Retroarch core / Standalone emulator?:

wait. so delay and input lag aren't the same term?

I'm not sure about that, but at the very least they mean the same in my previous answer : the time between the game receiving your input signal and sending the result of that action to the display.

@DjDiabolik said in Input Lag - How can "CHECK" if present in Retroarch core / Standalone emulator?:

Street Fighter game

Are you refering to Street Fighter 1 ? I sort of remember reading somewhere that the hadoken works slightly differently in that one (F,P instead of F+P ?)

Lolonois

That might be quite a journey to measure the input lag.

Did you check https://inputlag.science ?

If you want to take a shortcut, maybe you are better off with connecting your sticks and buttons to a RP2040 [0] with GP2040-ce [1]. The Zero Delay Decoder [2] -despite its name- does not cut it in contrast to the GP2040-ce performance [3].

[0] https://en.wikipedia.org/wiki/RP2040
[1] https://gp2040-ce.info
[2] https://inputlag.science/controller/results/zero-delay-usb-encoder
[3] https://gp2040-ce.info/#performance

DjDiabolik

@barbudreadmon yes.. Street Fighet 1. But it's just one example.... let's say that generally with all systems I don't notice visual delays or inputlag. but when I tried this game it was particularly unplayable so I got the suspicion that there are some.

@Lolonois
Very interesting site....... look your link here :
https://inputlag.science/controller/results/zero-delay-usb-encoder

Apparently it's the same exact hardware i have... anyway in general it doesn't seem easy to replace my current zero delays with two raspberry pico.

barbudreadmon

@DjDiabolik SF1 is known for having horrible controls, and its specials are almost impossible to do reliably, you are actually very lucky if you succeed once out of 3.
Also, i confirm the move is apparently D,DF,F,P and not D,DF,F+P, and it might also be worth mentioning that inputs are registered at release, not at press.

DjDiabolik

@barbudreadmon Lol.. so in the end we can say that, as an example and as a test, i have choice the really BAD games ever :) ?

Anyway, I should have understood the concept and the difference between "input lag" and "delay" even better:

Input lag. In my case about "zero delay encoder" it's 16ms . If we take as theoretical and as average the values calculated by the above mentioned in thesite inputlag.science 16 thousandths of a second is pratically "zero".
What could be affected by mine "ecosistema retropie" is the presence of a possible "command delay":
The time span from the physical pressing of a key to the action of the "player" or "character"on the screen.....

Upper here we take as example 16ms... the correct order of things could be:
Press of a button -> Input Lag of hardware -> Delay of the games/character/games/emulation -> action on screen

anyway for my next arcade kit I was considering buying something similar:
RP2040 Advanced Breakout Board on Aliexpress

I was also considering buying a totally original Sanwa kit at least for around €100 but that's a whole other story.

Lolonois

@DjDiabolik said in Input Lag - How can "CHECK" if present in Retroarch core / Standalone emulator?:

anyway for my next arcade kit I was considering buying something similar:
RP2040 Advanced Breakout Board on Aliexpress

Nice. That creates added value to a plain RP2040, pinheaders and Dupont wires. FWIW, here is the non-commercial link with all the tech specs.

Great you have some takeaways/learnings, that's the Raspberry/RetroPie spirit.

PS: The Sanwa's do have a high price tag but I never regret my invest: Mechanical still precise after ~5yrs, no maintainance needed, colors fresh as of the first day (my setup is partially exposed to sunlight during the day).