How much does your arcade stick lag? Arcade stick input lag testing & results thread


Hey guys, I wanted to share something I’ve been working on for the past half year or so. I’ve performed input lag tests for a large number of PS3, and XBox 360 sticks/PCBs and figured that I should post up what I’ve got done before these consoles become obsolete, lol. I’m now working on testing PS4 sticks.

All of the information on the tests can be found on my website including the methodology used, controls in place, and support for my testing method.

The results can be found here:

And finally, the raw testing data can be viewed & downloaded from here:

Sticks/PCBs tested as of last update:
PS4: 17
PS3: 43
360: 22

Here are some quick results charts I made up for quick reference:

PS4 results:

PS3 results:

XBox 360 results:

I’ll be updating it as our group comes across more sticks. The PS3 results were certainly interesting, I didn’t expect the unlicensed PCBs to have such a huge advantage over licensed PS3 sticks. Anyway, hope this helps out those of you who consider input lag to be an important factor in choosing which stick to use.

3rd Strike Lag/Speed Tests
Brook PS3/PS4 "Fightboard" PCB Thread
Official USFIV/GGXrd/P4U TE2 Thread

Hey, I’m a bit confused about some of your results. Instead of just listing arcade stick models would you also be able to list the PCB?

Example: AFAIK the Q4RAF uses the same pcb for their white/red & blue versions.

Also I think that the Joytron Exchanger might use the same pcb as the Q4RAF. Can you clarify this please.

UD-CPS2: Fully Consolized Capcom Play System 2

One issue I have this data is contrary to Toodles test data he did a while back with the same methodology.


I discussed the issue regarding the Q4RAF Blue / White / Q2 Pro PCB on the overview page, it’s towards the bottom. Also the EXChanger does have a different PCB than these 3 sticks, just had a look again now.


Cool, but I still think it’s important to list which pcb each stick is using. For instance the original TE Fightsticks had three different motherboard revisions for the PS3 version, and at least two for the 360. And the last factory batch of the TE Round2 sticks used TE-S motherboards (the same found in the Fightstick Pro and V.S).

I believe the Etokki Omni uses the ‘paewang’ pcb which is the same used in the Datel ArcadePro and a few other sticks.

If you want to identify the TE model in the 360 table then post a pic up here.

The PS360+ 360 results seem a bit shocking… are you still testing sticks out or is the project over for now? It would be interesting to see how the newer MadCatz & Hori sticks compare…

Oh - and you spelt ‘Cthulhu’ wrong in your PS3 table above :stuck_out_tongue:


Are you referring to this article?

It’s too bad that there’s confusion out there regarding the Q4RAF. There are two PCB models for the Q4:

  1. The original PCB on the initial runs of the stick, which is a two tier PCB (looks like two PCBs stuck together). The Eightarc Fusion, or at the very least the early models of it, also used this same PCB. This PCB is in fact faster than the Toodles Cthulhu PCB as shown in the SRKlabs testing and my testing.

Pics of this PCB are here:

  1. The new PCB, which is a single wider board, and is used on all of the new Q4RAF coloured models and also in the Q2 Pro line. This PCB was redesigned and is considerably slower than the Cthulhu, as shown by Toodles tests in the above link, as well as in my testing results.


Where’s the Razer Atrox? :slight_smile:


Unfortunately none of our players are sponsored by Razer so we don’t have access to that. Or did they finally put it out for us peasants?

Either way, the sticks tested are based on what I’ve got available. A friend actually just got a new Paewang installed, so I may test that to see if it performs the same as the model in his Omni.




That’s what I’ve heard over the grapevine anyway, that the firmware has the potential to be patched to be PS4 compatible. No idea if that’ll end up being the case though.


Seriously doubt it.


Based on new information, I doubt it. But that’s getting off topic, I just thought it was interesting as I hadn’t read anything about that as a planned thing for those sticks in the short term.


how does the latency of these sticks compare with official licensed gamepads?


I like what you’re doing here. This could be very useful information. But I wonder about the method.

When you press the button you are opening a circuit between the two PCBs. The button is not sending a “button activation” to two distinct PCBs, it is temporarily connecting the two PCBs as if they were wired directly together. So the behavior is not necessarily the same as a normal button activation.

Two years ago when the Qanba vs MC Cthulhu lag issue was first brought to light, Toodles did a similar test but used a signal separator and posted the schematic. I think if you were to do your tests this way, the results would be more easily accepted.


Are you suggesting that having both PCBs wired together will skew the results towards one PCB over the other, in meaningful amounts (+1 ms) on a consistent basis over 1,000 trials? If so, can you explain the logic and your research behind this?

If not, then what we are left with is simply additional variance. Which I’m willing to accept and expect to be ironed out after 1,000 trials using the same method. Regardless of whether an input splitter is used, you aren’t going to see any stick beating out the Hori VX-SA on 360 - you can even test this manually to confirm. Whether or not my results are accurate to within +/- 0.2 ms or +/- 0.5 ms is not my main concern - as long as the results are reasonably consistent and comparable, my goal has been met.


Toodles himself actually performed both the signal splitting test and the dual wiring test and came up with similar results under both methods. So I don’t see an issue with the way I have done things, unless you have something else I should consider.


Do newer black Q4s still contain the fast style PCB?


I don’t know if the amount will be meaningful or what will happen at all if I wire two controller PCB signals together, and I definitely don’t know if the output is identical to normal usage. That is not normal usage of the PCB. I believe Toodles had a point in doing his test with the circuits separated. Hopefully someone with better depth of electronics knowledge will chime in.

I don’t mean to downplay the amount of time you have put in. To be honest I have done the same type of test and it was my first instinct to just wire in a button, but I can see why that wouldn’t be the best idea. I like the work you are doing here, I just don’t know if the results are good or not.

Only if the test is accurate. Combining two PCB signals together could have unintended effects. We also have reports of odd results when certain combinations of PCBs are plugged into the same console, one affecting the other, so an X vs Y test doesn’t necessarily reflect on other combinations of X, or perhaps even Y vs X on different ports. You can see where I am going with this.

I hope you won’t accuse me of picking nits in a thread about millisecond differences in head-to-head controller input results.


I wish I knew! If anyone has a recently purchased Q4RAF Black and can snap a picture of the PCB, it would be really helpful. But my hunch is that since all of their other sticks are using the laggy PCB, the newly produced Q4RAF Blacks would use the same one.


So it sounds like your concern is with “overridden inputs” or “normal usage of the PCB” or something of that nature. Let’s consider what could happen if one PCB adversely affects the other in “normal” situations:

  1. Input is dropped by one side. Based on the roughly 20,000 trials I performed among sticks with low lag where there were no dropped inputs (every throw was broken so all inputs arrived), this is not a concern in my method as it didn’t happen even once.

  2. Input is delayed by one side (the same side) consistently. This would require some serious extra resistance by the extra equipment in between the button and PCBs (in my case, extra wire…) as well as requiring the signal to hit one PCB before it hits the other, where in reality it hits both at the same time.

Additionally, if one side imposed lag on the other, then lag would be understated for the laggier stick, which doesn’t seem to be the case.
Given how I performed alternate testing to confirm that the real world results of simultaneous button presses are within 0.5 ms of my dual wiring test, I can discount this as a significant issue. It also doesn’t make very much sense.

Toodles’ results also confirm that there are no significant differences.

Finally, I don’t mind you nitpicking about milliseconds but I do have an issue with you nitpicking about fractions of milliseconds. The effect of additional variance, the existence of which I’m seriously questioning at this point, would be closer to 0.01 ms than it would be to 1 ms.