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

Tech Talk
101

@Phreakazoid‌, though I am ignoring bInterval value, Toodles has 2 different firmware versions, 1 of which reports 1ms bInterval and the other 10ms during enumeration. @Teyah‌, please download this firmware set and test with the 1ms version:

http://www.godlikecontrols.com/download/cthulhu/MCCthulhu23.zip

I’m betting you will see a significant difference in your results. Also, it would be great if you would indicate for the firmware updateable PCBs exactly which firmware revision you tested with so that people who view your results can compare apples with apples.
-ud

102

Is there actually another way to dual mod besides the use of the PS360+? If there is, would it be less laggy on Xbox 360? I’m really concerned after reading this. I have a custom stick being made right now and its going to be using the PS360+ to drive it. I play SF4 on Xbox 360 currently with a TE-S. I don’t want to have a beautiful custom stick that actually performs worse than what I already have.

Does the use of an LED controller like the Kaimina (specifically the kaimana actually) add to / cause any lag?

103

I already have customers asking me about this chart, and that’s making me angry.

Why angry? Because the average person has no idea how to truly interpret this data, and especially as to what it means to them.

To the average player, this chart means JACK SQUAT. If someone loses in a match, it’s 99.99% them being outplayed and 0.01% input lag. I really wish this would have included real-world applications, but instead most end-users interpret this as “any lag = bad” therefore “dual mods cause lag, so dual mods are bad.”

Unfortunately, bad news travels faster than good news, so no matter how many times I explain to people how useless these test results are for them, people are now going to think dual mods are bad.

104

From what I gather they’re saying that the ps360+ lag on 360 equates to a little more than 1 frame of lag. If I go into training on SF4 and change the network simulation from 0 to 1 I actually notice a difference. It isn’t huge, but I sure as heck prefer it at 0. With that 1 frame of lag added it doesn’t feel ‘right’. Things feel much tighter with 0 frames of lag added. If that’s the real world application of the increased PS360+ lag than its a problem. Also, I see people lose pro matches all of the time because they missed the last hit on their combo. People shouldn’t have to go ‘well my stick has lag so I need to play around that’. Especially when they have the option to purchase a stick without that problem.

I’m really concerned now. I would be interested to know if there’s another way to dual mod that doesn’t have these problems, or if there’s just a 360 / PC lag free PCB that can be purchased.

105

i could probably loan my EX-SE for testing sometime

106

There is no comparison between input lag inherent in online play and input lag (if any) in your arcade stick. None whatsoever. In fact, online play should NEVER come up in any matter concerning local match play, including (and especially) tournament play.

This is exactly what I’m talking about. What you’ve done here is misinterpreted the data and created a problem where none existed, and this is my problem with the tier list and the data itself.

I have personally witnessed several pro players win tournaments using the MadCatz TE, supposedly the worst stick on this list. I myself have worked on over a thousand arcade sticks, including either dual-modding or working on sticks for Chris G, Justin Wong, PR Balrog, Ryan Hart, Filipino Champ, and many others – sticks that these players have used to win majors. Point is – I kinda know what I’m talking about when it comes to arcade sticks. You can trust me when I tell you that the stuff I and other modders use is just fine. MC Cthulhu, Cerberus, ChImp, PS360+ – ALL JUST FINE.

I cannot overstress this enough: THE RESULTS OF THESE TESTS MEAN NOTHING TO THE AVERAGE PLAYER IN REAL-WORLD GAMEPLAY.

107

So I won’t feel any difference between a PS360+ and my TE? I just assumed that network simuation thing in training mode was just causing 1 frame of input lag, the same thing I would experience with the PS360+. I hoped I was wrong here. What I worry about is these people talking about how their hit boxes feel ‘off’ because of the PS360+. That’s what concerns me.

I don’t really give a crap about any game like guilty gear or anything. I only play SF4 and I’d like to pick up MvC3.

I’m just trying to interpret the data I see. I’m not proclaiming I’m an expert. I’m honestly new to all of this and am just trying to figure this out. I just invested a lot of money into a custom stick that’s going to be driven by a PS360+ so you can imagine I was very disheartened when I saw this thread.

108

So none of the PS3 sticks are that great? Which stick should I look at buying if I want good PS3 and PC performance? I’m primarily a PC player, as the latency is better on my monitor and is usually better on the PC versions of the fighters.

Should I order a 360 stick if I’m going to play on PC primarily?

I own a PS3, I do not own a 360.

109

Get a PS3/PC stick that you think looks good and comfortable

110

So just because top players use laggy sticks and can win, lag shouldn’t matter? Sorry, but I don’t buy it. If those same players switched to S or A tier ranked sticks/PCBs then they would perform better. “Better” in the sense of shaving up to 21 ms off of their reaction time. Which would help them block random overheads more consistently, reaction punish whiffed normals more consistently, hit confirm from single hit pokes xx specials more consistently, win on hits that would have otherwise traded (eg. Twitch reaction air to airs), and so on. Basically anything requiring reaction time is improved to some degree.

These kinds of things do affect the average player, but obviously not to a high enough degree that it will let them start beating clearly more skilled players. That’s just a ridiculous assumption. Equally ridiculous as saying that THE RESULTS OF THESE TESTS MEAN NOTHING and implying that there’s no effect on gameplay at all.

111

Thank you for your work, very important numbers.

I am surprised to see some people downplaying them. Especially since some of these pcb’s add a significant amount of lag. And if they didn’t, it would be important to verify that they didn’t.
I also don’t understand the logic behind the top player argument. Fighting games are more than just proper timing, and we do see upsets happen. When there are bunch of time sensitive tasks where 1-2 frames can be the difference between failure or success i don’t see how you could downplay the importance of >1frame lag.
We have splitters, monitors, different versions of games and sticks possibly adding lag, even if each on is adding a small amount it could end up being significant.

Nice to see my old qanba q4 perform well, sadly i cant say the same for my ps3 madactz pro though.

I have a question, why do you assume that the qanba q4raf original on ps3 has the same amount of lag as on xbox360?

112

The apologetics are unnecessary.

What you feel is fine is moot in the face of hard evidence. Why settle for something fine when you can choose something you know to be definitively better? Winning tournaments with a laggy stick does not mean they could not have had superior performance throughout that same tournament using a more responsive input device.

Does that mean tournament results were largely impacted by the usage of certain sticks? Impossible to know. That requires data we don’t have. At the same time Madcatz products are so prolific that the playing field is practically even.

Does it call into question certain situations? Sure. DP FADC into forward dash is negative five on block. Been a month and people are still not accustomed to punishing this. Could it be because they realistically have four frames to punish due to input delay caused by a PCB? It’s absolutely possible.

113

I totally understand the position of wanting to let sleeping dogs lie (hardly anyone could detect differences < 1ms, so why even make the data available) and I also admire the desire to get some nitty gritty performance specs. (for science, not because I think people will play any better than they already do). I would just want the data to be as fleshed out as possible and dead nuts accurate, so though there may be arguments about the data being put out there, there should be no question of the accuracy of said data.
-ud

114

So it seems the etokki omni is all around really good. Couldn’t you just take the PCB from that and put it into a custom stick and it would be the same thing?

115

did you seriously bring up chris g, one of the whiniest top players in the fgc?

clearly you forgot when during ceo2013 he whined to jebailey about wanting to switch monitors during top8 ae2012, he wouldnt do it and when he lost, he threw another typical tantrum. seriously walk by him during any major and he’ll be saying something along the lines of “I LITERALLY HIT THE BUTTON AND NOTHING CAME OUT.” gee i really wonder why…?

pr balrog is up there too as well in terms of whining about lag. literally got up and took a picture of the tv he was playing on when he was getting bodied by nemo a year back in a ft10.

come to think of it, every single player you’ve brought up isn’t known for their execution at all…

and again, this point has been made before. you’ve brought up people who’ve won tournaments PLAYING MODERN GAMES WITH EXECUTION SHORTCUTS BUILT INTO THEM. you’ve never once accounted for people playing guilty gear, vampire savior, the other rest of the niche communities. I suppose you’re right that the average player would never touch any of those niche execution games but the point still stands.

116

It’s just an estimate to use in case I find another stick that beats out the old Q4. Although really, I will probably end up reducing all numbers on PS3 by the least laggy stick, and set that to zero lag as the standard.

117

Hey, I’m an owner of a Cerberus with quick access to a PS3 MadCatz FightStick PRO to borrow. I was curious how the Cerberus would fare in a shorter version of the alternative test Teyah listed in his overview, since I don’t have the supplies to run his newer test. Then, of course, I checked this thread and saw that Phreakazoid is going to hook Teyah up with a Cerberus at Evo, and now I feel silly. But I figured I’d still share the results of my “test,” if it can even be called that.

I used the help of an assistant (the owner of the FSPRO) for this. I went into Guilty Gear XX Accent Core Plus R with the intention of running the same throw-break test as Teyah. My assistant held the joysticks so that both Sols were walking forward and watched the screen to keep count, while I would lower my head and close my eyes, pushing both Heavy Slash buttons. We did this fifty times, and then switched the positions of the joysticks so my other hand would be operating each stick. After another fifty of that, the two of us switched off so that I would keep count, and my assistant would push the buttons. We did another two sets of fifty like that.

Yes, I feel goofy describing this.

Here are the results:
Cerberus Wins 2+ - Cerberus Wins - Tie - FSPRO Wins - FSPRO Wins 2+
19 - 27 - 2 - 0 - 2 (Cerberus on left, Sloth on buttons)
2 - 31 - 13 - 4 - 0 (Cerberus on right, Sloth on buttons)
20 - 28 - 2 - 0 - 0 (Cerberus on left, assistant on buttons)
14 - 32 - 4 - 0 - 0 (Cerberus on right, assistant on buttons)
Total:
55 - 118 - 21 - 4 - 2
27.5% - 59% - 10.5% - 2% - 1%

The FSPRO obviously should not have won any of those exchanges, so that already says a ton about the validity of this “test.” But at least the numbers are a little more consistent when my assistant was pressing the buttons.

Anyway, if the PS3 FightStick PRO lags 23.5ms behind the baseline HRAP VX-SA, and if my math and logic are correct (that’s the part I’m most worried about; someone feel free to correct me), that would see the Cerberus sitting at 5.17ms.

Again, don’t take this test to mean much. I’m a tech-scrub, and this test is an unscientific mess with a small sample size. I’m sorry for polluting the thread with this.

Edit: For the sake of sharing the blame, my assistant in this test is known as Liberation Maiden.

118

Wondering if any more testing was done with the Cthulhu, as I’m thinking of modding the Namco PS1 stick with it for PC/PS3.

119

@ParryPerson I think the PC/PS3 Cthulhu is different than the MC version with the upgradable FW(I don’t think you can upgrade the FW on the PC/PS3 one, could be wrong though), so it may yield different results
Edit: Oh it seems he didn’t test the MC version

120

It’s not unscientific at all. It just has a very high variance due to human error and limited trials that makes it probably inaccurate to a high degree (+/- 1 ms, maybe more).

Your math is correct. If we remove the 2 outliers where the FS Pro won, since that isn’t normal behaviour (at least in every other stick I’ve tested) then we’re at 4.64 ms. Just for interest’s sake, I’ll keep this number in mind while I do the testing this weekend.

I think you’ve effectively ruled out that the Cerberus is E or F tier in terms of lag, with your simple test. Unless you and LM were both very clearly biased to hitting buttons early on the Cerb every time. :slight_smile:

Edit: Pretty sure I know where the MCZ Pro “winning by 2F” came from. In GGXX you can’t perform a throw on someone within 9F after they wake up. So if in the previous test, your Cerberus won by 2F and landed a throw, it’s very possible that you didn’t wait out the 9F of wakeup throw invulnerability and so the MCZ Pro Sol was throw immune. So he could have won the throw that way. This is consistent with your test only showing 2F wins (clean throws) for the MCZ Pro and never any 1F wins.