GGXrd Input Lag Tests / Results (PC, PS4, PS3)

[Teyah]

I heard that the PC version of Xrd was performing amazingly fast, and after I tried playing on it I could definitely notice the difference.  I just finished testing the input delay of the PC version of Xrd, and thought I'd compile my results here.

First of all I'd like to mention the methods I used. I used the red LED light on a TE Kitty to time the input of a button press to the exact frame, then counted the number of frames until the first frame of animation, using a 60 fps camera to capture video. I used Faust's 5D as the button input, since it has a huge first-frame animation (the huge tornado spin). I took several passes of video for all versions of each port and confirmed that the results shown below were, by far, the most common result. Almost every time, the results were as shown below, rarely they were one half or one frame off - this is likely due to variances in input timings (whether the input arrives at the beginning or end of the frame buffer). After taking several (~20-25 tests) on each setting, I can say with confidence that these input delay results are accurate.

 

Here's a list of the equipment that I used:

 

Canon PowerShot SD800IS - used for taking 60 fps video

Panasonic HD CRT TV - Used to test GGXrd PS4/PS3.  Panasonic is known (and tested by me) to have the lowest lag CRTs on the market, especially when it comes to HD CRTs. This CRT was confirmed to be 100% lagless - measured at 0.0 ms delay for the top bar, using the Leo Bodnar tester.

Samsung SyncMaster 955 DF - a zero-delay CRT computer monitor, used to test GGXrd PC

PC running dual GeForce GTX 680s, 32 GB DDR3 RAM, 5th gen i7 processor, Windows 7.  GGXrd PC runs at a constant 60 fps.

PS4, PS3 consoles - for testing GGXrd ports

PS360+ PCB for testing on PS4 - confirmed to have the lowest lag (+0.0 ms) of any PS4 arcade stick, self-tested

TE Kitty PCB with LED for testing on PC/PS3 - confirmed to have near-zero additional lag (+0.27 ms), LED is used to time the zero frame of input

Arcade stick lag testing methods & results can be viewed here: http://teyah.net/sticklag/overview.html & http://teyah.net/sticklag/results.html

 

Now for the results. Watch the red LED light to see when the zero frame of input is.

PC Xrd: 1F Delay (FXAA, no V-sync, no post effects), 2F Delay (FXAA, post effects on), 3F+ Delay (MLAA or SMAA, V-sync on, post effects on)

Tested image is with V-sync off, smoothing turned off, standard FXAA.  The amount of delay you experience depends on which Display options you tick, roughly as follows:

• Standard delay, FXAA: 1F
• MLAA/SMAA: ~+0.7F
• V-sync ON: +1F
• Post effects ON: +1F
• Processing -> Quality mode: No effect

PC Xrd runs at a native delay of about 1F, which is even faster than the PC port of GGXX #Reload which had a delay of 2F (as tested in my other thread) and considerably faster than the PS3/PS4 ports of Xrd which have 4F of delay.

So what does this mean, now that we know that the PC version is about 3 frames faster than the console standard?  Assuming a stable, smooth connection, you should be able to play up to Delay 3 on PC Xrd and have it feel like "offline".

The PC port being much faster than the console port is what also lead to #Reload being so great online back in the PS2 era.  We were all used to Delay 4, so being able to hop on PC and play in roughly the same delay, but online, was a great experience (most of the time).
 

PS4 & PS3 Xrd: 4F Delay

Posted via twitter awhile back:

 

Picture shows PS4, but I'm fairly sure I tested PS3 to be the same speed.  I'll re-test this one since I can't find the videos I took to confirm.  

No LED on this one, since that stick does not function on PS4; however if you compare the "-1" and "0" frames, you'll see I'm being very generous with the timing of the zero frame here.

[heavymetalmixer]

Well, I misunderstood the FXAA one, never mind.

[Nyaa]

You might get different results regarding vsync if you turn vsync off ingame but force it through nvidia control panel / catalyst.

[Jakestation]

Hoh and twitch mikado chat said my tv is so laggy if i feel difference between ps4 and pc :D (yep they were serious)

[Teyah]

7 hours ago, Nyaa said:

You might get different results regarding vsync if you turn vsync off ingame but force it through nvidia control panel / catalyst.

V-sync adds a minimum of 1F of delay in order to synchronize frames, so tweaking could only result in more lag and not less. So I think we're okay on that end, since in game V-sync is already down to 1F of extra delay.

[Tong]

Very interesting!

Ever since my first day playing the PC version, I felt that the timings were more "tight". And no wonder I've been blocking Dusts on netplay very often!

[Nyaa]

13 hours ago, Teyah said:

V-sync adds a minimum of 1F of delay in order to synchronize frames, so tweaking could only result in more lag and not less. So I think we're okay on that end, since in game V-sync is already down to 1F of extra delay.

Yeah which still sets it faster than console. If anything tweaking such settings is necessary for the PC version to feel like console. That being said, different GPU's can give different results too.

Not to criticize the testing but usually such tests are done with a LED-lamp or similar modded to the controller being used directly wired to the button being used to test to be absolutely sure that there's no variation. I would assume the poster tested the same test multiple times as well to find an average, but this isn't mentioned. Especially when it comes to PC there can be a lot of variance depending on what's going on in the background.

http://www.displaylag.com/reduce-input-lag-in-pc-games-the-definitive-guide/ This is a pretty comprehensive article regarding this.

[heavymetalmixer]

7 hours ago, Nyaa said:

Yeah which still sets it faster than console. If anything tweaking such settings is necessary for the PC version to feel like console. That being said, different GPU's can give different results too.

Not to criticize the testing but usually such tests are done with a LED-lamp or similar modded to the controller being used directly wired to the button being used to test to be absolutely sure that there's no variation. I would assume the poster tested the same test multiple times as well to find an average, but this isn't mentioned. Especially when it comes to PC there can be a lot of variance depending on what's going on in the background.

http://www.displaylag.com/reduce-input-lag-in-pc-games-the-definitive-guide/ This is a pretty comprehensive article regarding this.

I always play without V-Sync at all and 75Hz as refresh rate for the monitor (PC), i guess that's why I can play with more lag (online) than most of the people.

[Teyah]

3 hours ago, Nyaa said:

Yeah which still sets it faster than console. If anything tweaking such settings is necessary for the PC version to feel like console. That being said, different GPU's can give different results too.

Not to criticize the testing but usually such tests are done with a LED-lamp or similar modded to the controller being used directly wired to the button being used to test to be absolutely sure that there's no variation. I would assume the poster tested the same test multiple times as well to find an average, but this isn't mentioned. Especially when it comes to PC there can be a lot of variance depending on what's going on in the background.

http://www.displaylag.com/reduce-input-lag-in-pc-games-the-definitive-guide/ This is a pretty comprehensive article regarding this.

You can turn on all the effects and increase aliasing to achieve near parity with the console versions.

Also.. there is an LED lamp wired to the button, you can see it just above the stick on the left side.  It glows upon button press and is used for timing the zero frame.

My testing methods are detailed in my other thread on input delay, I'll copy that over. I did test multiple times to arrive at this result, what you see above is the most common result.

[RoBoBOBR]

Now all we need to know is arcade input deay.

[Nyaa]

8 hours ago, Teyah said:

You can turn on all the effects and increase aliasing to achieve near parity with the console versions.

Also.. there is an LED lamp wired to the button, you can see it just above the stick on the left side.  It glows upon button press and is used for timing the zero frame.

My testing methods are detailed in my other thread on input delay, I'll copy that over. I did test multiple times to arrive at this result, what you see above is the most common result.

Oh ok, I didn't notice the LED at all. But, ok then. Good testing!

7 hours ago, RoBoBOBR said:

Now all we need to know is arcade input deay.

Should be the same but the arcades have different monitors, I think I recall Fab saying it's 2-3f added delay on some arcades because of the monitors. Explains why japanese are so bad at blocking 20f overheads that should be pretty reactable at their level.

tldr; pc online is like console offline, arcade offline is like console online.

[Donsklz123]

The frame delay difference needs to be resolved. The ps4 one is so bad, that 4fps drop causes severe input lag. It makes arrogant people spam moves, abusing net play to obtain free attacks....like greed sever even though I am KY playerror. I know greed severe is cheating on netplay. Thank God I bought this on PC, way better but the community is dead. I need my refund for Revelator on ps4. I will not play a plagued input lag game. I have 120 down and 25 up, plus hardwired...no reason for my to experience any form of lag. Sony fix your hardware and ark do better.

[zankoku]

This is a measurement of input delay, not loss of framerate.

[Magaki]

If you check this website: http://www.displaylag.com/video-game-input-lag-database/

You will notice that actually, 5f of input lag is pretty good compared to other games.
As for your issues online, you might want to either play against people closer to you, fix your connection or lab it up in order to better recognize when you can identify a move you want to block on reaction.

[Mulbruk]

Are you going to test Xrd R's input lag Teyah? Some seem to think it's lower than SIGN's!

[Kain_Dizengarth]

It seems that's not only me that find Xrd R being faster that Xrd S. It would be nice if we can see if its actually that or simply a thing of perception

 

[Kyosuke Kagami]

I think it's definitely faster. At least it's easier for me to play on reaction compared to Sign.

[B.Z.B]

Has someone tested this yet on a ps4 copy of Xrd:

Using a Dual modded PS3/PS4 arcade stick check input lag difference between its PS4 mode vs PS3 mode using Xrd's w/ legacy drivers? I recall SKD was saying there's a 2 frame delay.

Seems that is the case.  They make it sound even worse here though.

 

[Teyah]

5 hours ago, B.Z.B said:

Has someone tested this yet on a ps4 copy of Xrd:

Using a Dual modded PS3/PS4 arcade stick check input lag difference between its PS4 mode vs PS3 mode using Xrd's w/ legacy drivers? I recall SKD was saying there's a 2 frame delay.

Seems that is the case.  They make it sound even worse here though.

 

Yeah, I did some quick / nonrigorous tests when I noticed sluggish response when using PS3 sticks in PS4 XrdR.  It's somewhere between 1-2F of additional lag when using PS3 sticks.  For simplicity's sake I'll call it 1.5F for now, though I will run some tests on this when I'm able.

[Clamper]

@Teyah do you plan on testing PC Revelator? I'm under the impression it's a bit more laggy than Sign...