piratethingy
Self professed bad raider
Real talk I don't want to read this topic because I know I'm living in gleeful denial
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
NX Gamer Wired vs Wirless input latency.
- Thread starter Yautja_Warrior
- Start date
Not to knock on NXGamer, but I don't know how anyone can believe these here results. 125-200ms input lag on Xbox? Lol, going from 60fps to 30fps adds 16ms, and that already feels pretty significant. Imagine what 10 times that value would feel like.
So, one game?
Not surprised I suppose.
https://youtu.be/L4rlLYE-e20?t=1m25s
PS4 version of USF4 will show ~4 frames of latency (without the converter, at 60fps) on this guy's setup so that's 16*4=64ms of lag.
So yeah, I don't know what the fuck NXGamer is doing.
Anyway, his videos are impossible to watch without sound. I couldn't get info from it properly until I got home... I wish he would actually show that this is total lag measured, and not merely input lag. He isn't accounting for display lag etc.
NXGamer
Member
2 things here,
1) Fighting games by nessacity have the lowest input lag, I never tested any in this video but I have games that hit the 60's-80's. These are not comparable to other games and more so 30 V-Synced titles that will have the highest and more than just 16ms.
2) Streetfighter V on PS4 launched with twice the latency of the PC release, around 124ms. This was later improved to half that 64ms.
This video also has 80ms counts on this game and it all depends where in the input poll, blanking stage etc you are, hence my mean point of the video as being a good base.
Other factors can also play a part, I will have my Web article up soon and I am still working on refining and improving the process, the 30-50ms gap would be very hard for most to spot and with some tv configs making this 2X worse than that.
The numbers this video is giving don't look right, some more documentation of how they were measured and the equipment used would be good.
Here are a few numbers from Digital Foundry last gen.
- Most 60FPS games have a 66.67ms latency - Ridge Racer 7, for example.
- Citing GTAIV as an example, West suggests that a 166ms response is where gamers notice controller lag, which could also explain the Killzone 2 furore too.
- The lowest latencies a video game can have is 50ms (three frames) - the PS3 XMB runs at this rate, but few games reach it.
- Halo 3 -100ms
- Guitar Hero 67ms
- Burnout Paradise 67ms
- BioShock 2 Frame-rate Locked 133-150ms
- BioShock 2 Frame-rate Unlocked 100-150ms
- Call of Duty: World at War 66ms-100ms
- Dante's Inferno 100ms
- Killzone 2 150-183ms
- LittleBigPlanet 100ms
- Mirror's Edge 133ms
- MotorStorm: Pacific Rift 116ms-133ms
- Resident Evil 5 100-150ms
- Ridge Racer 7 66ms
- Street Fighter IV 66ms
- Unreal Tournament III 100-133ms
- WipEout HD 84ms
But this video saying
PS4 XMB is 75ms
DOOM 104ms
Battlefield 75ms
Halo 5 165ms
All 60FPS games, but all them them significantly higher than either the theoretical lowest point or the response rates of games from last gen.
Forza Horizon 3 -171ms again.
So all really odd numbers.
Tripolygon
Banned
PS4 Dash Latency Test = 75ms https://www.youtube.com/watch?v=_7kcLTJTQK8
BigEmil
Junior Member
Similar results, the description of that video also says:
" This input lag test was recorded at 120 frames per second. Latency is obtained by counting frames between first keyboard light and first in-game action. All measurements are rounded up to the nearest millisecond. Display lag is included in these measurements; subtract 10ms from final averaged result to receive an approximation of game engine latency. "
The point of my post was to show that whatever you're using to measure the latency doesn't seem right. What method ARE you using? Frame by frame like this person here? (he's looking at the inputs that are updating, not the animation).
Street Fighter V? The game in the video is USF4.
And I don't know about that, I think SFV still has 7 frames of input lag (which is 112ms) and that's an example of a game with too much.
NXGamer
Member
What are you basing this on?
I have seen even In this thread other figures that match mine, in addition to other sources.
At this point I have some other things to iron out but the tests were done multiple times and the deviation was all similar and within expected levels.
My point on SFV is your example has no connection, looking at a completely diffirent game on completely different controllers is exactly apples to oranges, it makes no sense. Like I say I never tested fighting games in this video as they are nearly always the lowest level (but SFV shows not always)
I tested the same game across various machines with the standard controllers, but I will have more to come.
Halo 5 was 125Ms, with a lowest count of 108.
Again, comparing last gen titles in various machines is not the same as here, it needs to be done on a game by machine basis.
Interesting find on that, thanks I never saw this so at least 2 of us got the same results.
I'm basing it entirely off the fact that the numbers seem very high.
But regardless, as constructive criticism, you should detail your methods used in the video. Regardless of the validity of the results, this reaction is to be expected if you're highlighting that they are surprisingly high.
Yeah, something was screwy with that game. I'm not super sensitive to input lag, but that one was terrible
LukasTaves
Member
The Ps4 version had it at first too, but eventually got patched.
Wait you are saying Halo 5 with a perfect 60fps has changing input lag?
Perhaps I should find other videos
LukasTaves
Member
I found this other (older test):
https://displaylag.com/console-latency-exploring-video-game-input-lag/
I wonder if the differences on xbone measurements are due the OS? Sometimes I notice worse performance across many games and a restart fixes it (used to happen more at first iterations of the dash now it's almost gone), and curiously, my brother just called me to tell he thinks Cortana was giving performance issues on UMvC3, he said that even offline game felt a lot snappier after disabling it.
https://displaylag.com/console-latency-exploring-video-game-input-lag/
I wonder if the differences on xbone measurements are due the OS? Sometimes I notice worse performance across many games and a restart fixes it (used to happen more at first iterations of the dash now it's almost gone), and curiously, my brother just called me to tell he thinks Cortana was giving performance issues on UMvC3, he said that even offline game felt a lot snappier after disabling it.
If NX gamer is to be believed there's not issue at all with XB1 latency on RoTR. Despite the fact that it's as clear as day to anybody who even understands the concept.
Something is wrong with his measurements.
AgentP
Thinks mods influence posters politics. Promoted to QAnon Editor.
Tomb Raider?
SapientWolf
Trucker Sexologist
Engine input lag is not a constant. It's a range.
copelandmaster
Member
Did he even outline his testing methodology? These results are kind of all over the place.
A setup like this is the ideal.
A setup like this is the ideal.
Yautja_Warrior
Member
Maybe there will be more info on the setup from the website article NX is doing about all this.
NXGamer
Member
I agree on that, and I wanted to take time to demonstrate and show my methodologies, which I will very soon.
Do you really think a game is a constant at all times and on each input method/action?
Sorry you got banned but, nothing is "wrong" with my methods, even if I have a variation of 10-20ms it would be relative to each other. RoTR was a big game that was played by many across 4 devices, the fact the x360 had less lag than X1 and PS4 should at least explain to some what is going on here.If NX gamer is to be believed there's not issue at all with XB1 latency on RoTR. Despite the fact that it's as clear as day to anybody who even understands the concept.
Something is wrong with his measurements.
Exactly correct, like a game engine itself.
I have a article coming and more videos that explain it and how I aim to improve.
Indeed I will have that, want to make sure I spend time doing it clearly and concise so everyone is clear on my methods.
Could people be underestimating just how quick (for example) 100ms is? Maybe being used to having TV's that have sub 30ms of input lag and being under the impression that this is the only input lag going on? So hearing talk of 100 and 200ms seems so alien. Simple test to do is to start and stop a stopwatch as quick as you can, you will probably find the fastest you can do is about 70 to 80ms, which seems almost instant, might help put things in to perspective?
eso76
Member
I remember someone (digital foundry?) did the same test years ago and the results were surprising for different reasons: basically wireless didn't seem to introduce any meaningful lag or at all in some cases.
In many cases lag was as low as the framerate allowed.
Although iirc it was suggested that in a few instances controllers could still be communicating wirelessly​ even while connected.
Not sure who to believe but these values seem
way too high
In many cases lag was as low as the framerate allowed.
Although iirc it was suggested that in a few instances controllers could still be communicating wirelessly​ even while connected.
Not sure who to believe but these values seem
way too high
J_ToSaveTheDay
Member
Yeah, I am wondering if NXGamer's methodologies are actually relatively accurate and we all have some misperception or time-scaling issues that need to be explained/addressed.
For example, I've been shopping around for a new arcade stick to go with some upcoming fighting game purchases I'll be making, and I started thinking about the real impact of having a "laggier" stick that's maybe 8-10ms slower than another one... And it got me thinking that perhaps the arcade stick input lag probably wouldn't really be perceived on top of my display's input lag in a linear, stacked manner, as even the worst, "F-tier" sticks (according to SRK user teyah's popularized testing) generally have input lag that rates lower than most TV's display lag. There would be a small stacking effect, but it'd work more like a cascade/ripple rather than a straight stack-up of perceived input lag, and from that perspective and my inability to really detect extremely minute differences in milliseconds, I kinda figure even an "F-tier" stick @ 13ms would feel fine versus a 3ms stick, considering my display itself still has a 20ms input lag anyway, which might really resolve all-together as perhaps a 25-30ms total input lag perception (to clarify, I'm talking about my 20ms display + a 13ms arcade stick, not the 3ms one) -- still within very tolerable bounds as far as I'm concerned.
Adding to the notion that these are wireless controllers and my preconceived notion that they inevitably add a fair amount of input lag on that anyway, I'm perfectly willing to accept that games typically result in ~100ms of perceived input lag, and that it's honestly still something of a minute timeframe to perceive -- even if you are incredibly sensitive to that sort of thing. I'm not denying that a difference can be felt, but I also think that the scale by which people make that judgment in their head is perhaps a bit exaggerated when it comes to standard gameplay scenarios that most of us encounter. I do believe that hyper-competitive players feel and examine this very closely to good effect, but I think even those types of players might not really be keeping track or needing to keep track as closely when they play a single player game against AI which can be learned or manipulated in a way that a dynamic human being can't -- that is to say, I think a player can compensate by becoming intimately familiar with constructed behavior patterns that exist in game AI, even if it falls short of standards that are examined and desirable in more dynamic situations such as player-v-player.
All of this is conjecture, of course, I haven't run any measurements myself. But given my own somewhat limited ability to detect input lag within a scale of ~100ms (or really, a lack of verification of my ability to accurately assess input lag on a scale of ~100ms), I can buy these numbers and also still believe that games feel responsive enough to not be a bother within these conditions. I felt Rise of the Tomb Raider on XB1 and Dishonored 2 really bad on my PS4, and the results seem to back up exactly why, but most other games tested still seem fine and they all seem to also float around 100ms.
Unless the game decouples it's game logic from the framerate (which as far as I'm aware, is super rare) going from 60fps to 30fps will actually add something like 50-67ms additional lag. This is because the way most game engines work, there is inherently 3 to 4 (or sometimes more) frames of input lag; completely independent of the length of the frames; so halving the framerate naturally doubles the lag. That's why the jump from 60 to 30 is so easily perceptible (if it were only 16ms, it wouldn't be).
See the reason I find these numbers odd is because of this following list :
https://displaylag.com/console-latency-exploring-video-game-input-lag/
According to this, Destiny on Xbox One has an input lag of 86ms (not counting display lag of 10ms in their testing) while being 30FPS. So Halo 5 a locked 60FPS title having an average of 125ms of lag seems off to me because Halo 5 certainly feels more responsive than Destiny.
The website I've linked to also details how they did the tests so maybe you could look at them and compare your methods to figure out if and why there are any discrepancies.
NXGamer
Member
108-125 is not high at all, I doubt many would notice the gap between 90-130ms at all.
What do you mean by "proper" display and measure equipment for these tests?
I think you have a very valid point and what I tried to cover in the video, some can notice a delay but a wrong tv setting, link through Hdmi or such would add far more than we are seeing here and the expectation of the gap seems to high from some I feel as you suggest.
It is not as much about "who to believe" as what your expectations are. A 30-50ms gap is small, much smaller than many think and it all depends on a game by game basis.
Agree, some are high but many are within my expectations, as was said here already human response rates (average) is around the 100Ms mark.
I am confident in my methods and quality here, but am always open to discussion and I hope this has achieved that. I also agree with your statement that expectations and the 100-200ms is what is getting people thinking they are high. Many engines have far more integration and cross polonation with online servers, o/s interupts/calls and such so the fact we are seeing some higher times with much,much more being added to these games is not a surprise for me.
That was my finding. I've retested today and narrated a video that will be up in about 10 minutes @ https://youtu.be/y69g5cQSC0s My finding remains that the DS4 in wired USB mode is markedly slower than wireless.
Looking forward to reading more about the methodology. I'm sure our methods aren't the same (my is reasonably unique) but am curious as to why our conclusions are different.
BigEmil
Junior Member
Is the new DS4 wired communication mode done on PS4 Pro and Slim or standard PS4 for your test?
Caayn
Member
I mean a display with 1ms input lag and doesn't have a varying input lag, some displays have this. To lower the error margin introduced by the display.
As for the equipment that'd be a high speed camera, 120fps isn't highspeed, and a press detection mechanism such as a wired LED to the circuitboard of the controller.
Also, absolutely no equipment between the source and the display, but that should be obvious I guess.
I have a G900 and haven't noticed a difference between attached and wireless. There will be a light difference of course, but I sure as hell don't feel its affecting any of my gameplay and this is coming from someone who notices 9-11ms differences in response times between my monitor and my TV.
Wowfunhappy
Member
Just for the record, there's only actually three fully-isolated channels in the 2.4 Ghz band. The others overlap each other.
But yeah, you generally won't have problems with more devices.
NXGamer
Member
I will detail my test methods, but again Destiny is a cross-gen title and as such will be designed with that hardware in mind. Thanks for the link.
I will watch your video and I agree the older pad could sometimes be slower but in the OS test the difference was very, very minimal to be with 2-4ms.
We can chat about methods once I have completed my article, thanks.
A 1Ms monitor is not required, but I do use a persistent screen with low input lag for base tests in addition to standard screens for "real world" results that I also want to cover.
Of course nothing to interupt the signal and a faster camera than 120hz, input device vary depending on test, these were designed for real consoles and controllers but other tests will go for improvements. Also a few other parts on top.
Wow, that is impressive as I love fast paced games and screens but even I would struggle at 10ms to notice.
Video is on standard PS4. I have done the same test on PS4Pro in boost mode with the same result.
I did my own test on the Doom demo for PS4 and X1. My standard methods as explained here
https://www.youtube.com/watch?v=07nKZBXed7M
and here
https://www.youtube.com/watch?v=jNSWB6DwpgQ
X1 (response on 6th frame)
https://twitter.com/noodalls/status/862407771219648516
https://twitter.com/noodalls/status/862407836579397634
PS4 (response on 6th frame)
https://twitter.com/noodalls/status/862407917512695808
https://twitter.com/noodalls/status/862407993131909120
On the sixth frame of each one you can see a tiny bit of flash on the barrel and also the shot coming out.
Edit - to add, I used Brook Ultimate Fighting Board on these tests, updated to FW 1.8
https://www.youtube.com/watch?v=07nKZBXed7M
and here
https://www.youtube.com/watch?v=jNSWB6DwpgQ
X1 (response on 6th frame)
https://twitter.com/noodalls/status/862407771219648516
https://twitter.com/noodalls/status/862407836579397634
PS4 (response on 6th frame)
https://twitter.com/noodalls/status/862407917512695808
https://twitter.com/noodalls/status/862407993131909120
On the sixth frame of each one you can see a tiny bit of flash on the barrel and also the shot coming out.
Edit - to add, I used Brook Ultimate Fighting Board on these tests, updated to FW 1.8
FutureLarking
Member
As someone who has worked on Xbox One input on system level applications - it's entirely down to the OS / not particularly efficient focus and input handling at the XAML layer, so it' a bad place to do comparisons from. Games are fair play tho.
Prey demo X1 (reponse on 9th frame)
https://twitter.com/noodalls/status/862756623672000512
https://twitter.com/noodalls/status/862756731637583872
https://twitter.com/noodalls/status/862756828848955392
Prey demo PS4 (response on 13th frame)
https://twitter.com/noodalls/status/862756946780106752
https://twitter.com/noodalls/status/862757060814884864
https://twitter.com/noodalls/status/862757136182398976
https://twitter.com/noodalls/status/862757185775783937
Brook UFB FW 1.8
Note, I used the light because it is available from the beginning, and very easy to see when it changes. The UI does indeed change two frames earlier. I have also tested jumping and this responds at the same time.
I played up until I got the first gun, but unfortunately the way my current setup is designed I can alter the time a button is pressed between 0-33.33ms. Prey's guns will not shoot with inputs 2F (33.33ms) long. I tested this manually on my setup where you can physically push the button very quickly and not have the guns shoot. While this is a shortcoming of hardware, it is also interesting as it will be adding to the input lag (no response until 3/4/? frames of button push).
https://twitter.com/noodalls/status/862756623672000512
https://twitter.com/noodalls/status/862756731637583872
https://twitter.com/noodalls/status/862756828848955392
Prey demo PS4 (response on 13th frame)
https://twitter.com/noodalls/status/862756946780106752
https://twitter.com/noodalls/status/862757060814884864
https://twitter.com/noodalls/status/862757136182398976
https://twitter.com/noodalls/status/862757185775783937
Brook UFB FW 1.8
Note, I used the light because it is available from the beginning, and very easy to see when it changes. The UI does indeed change two frames earlier. I have also tested jumping and this responds at the same time.
I played up until I got the first gun, but unfortunately the way my current setup is designed I can alter the time a button is pressed between 0-33.33ms. Prey's guns will not shoot with inputs 2F (33.33ms) long. I tested this manually on my setup where you can physically push the button very quickly and not have the guns shoot. While this is a shortcoming of hardware, it is also interesting as it will be adding to the input lag (no response until 3/4/? frames of button push).
PS4 system (4th frame)
https://twitter.com/noodalls/status/862760367033798656
X1 RE7 (12th frame)
https://twitter.com/noodalls/status/862761444600815616
PS4 RE7 (13th frame)
https://twitter.com/noodalls/status/862761986693578752
Used the block animation, as nothing much else available early in RE7 demo.
https://twitter.com/noodalls/status/862760367033798656
X1 RE7 (12th frame)
https://twitter.com/noodalls/status/862761444600815616
PS4 RE7 (13th frame)
https://twitter.com/noodalls/status/862761986693578752
Used the block animation, as nothing much else available early in RE7 demo.
To be honest I'm probably really stupid but I cant understand anything your posting, a summary of some kind of what all this means might be nice.
SMOK3Y
Generous Member
Boost mode does nothing though does it? Have you turned on the cable input on pro?
Input occurs on the first frame. No visual response until 12/13 (X1/PS4) frames later. That's not to say that RE7 takes 12/13 to respond to everything, just that in the demo there's not much available to use, so I went with the blocking animation to have a clear visual indication when the game has responded, and it takes that long. So this is more just for X1 vs PS4 comparison.
Mentioning boost mode was more to pre-empt the next question, which would be "have you tested boost mode?"
USB cable was connected to new DS4, with USB communication mode on.
Spladam
Member
The only players/ gameplay this would really be relevant to are high level fighting game players, who sometimes time the games by a couple of frames, or pro esport players. Human reaction time is roughly 4/10ths of a second, or 400ms, it's what the "light trees" that drag racers use to start the race blink at. For reference:
10ms = 1/100th of a second
100ms = 1/10th of a second
Gameplay wise, most of us just playing our games are not going to be able to consistently perceive anything lower than 100ms -150ms of latency in normal games. Most of us will be comfortably oblivious to the difference.
10ms = 1/100th of a second
100ms = 1/10th of a second
Gameplay wise, most of us just playing our games are not going to be able to consistently perceive anything lower than 100ms -150ms of latency in normal games. Most of us will be comfortably oblivious to the difference.
Modified my setup, have retested. So this is prey, tested with the gun shooting. The input lasts for 50ms as this seems to be the minimum time required to get the game to respond. Flash light and jumping respond to shorter inputs (16.66ms). I have added an LED, and you can see that at the bottom of the screen. This lights up when a command is being input.
This is just shot with a standard 60fps camera, capturing 720p60 off-screen. You will see the first image in the twitter post has the LED off, and then it is lit on the subsequent frame. Counting from here (with the LED becoming lit as frame 0, and counting at 60fps even though the game is 30fps) you can see the xbox one responds on F10, while the PS4 responds on F14.
X1
https://twitter.com/noodalls/status/864442620407185408
PS4
https://twitter.com/noodalls/status/864443103003820032
So, these results demonstrate the same difference (X1 faster than PS4 by 66.66ms) between versions as my previous test. The controller is the same (Brook UFB FW 1.8). The screen is a BenQ RL series in game mode.
And yet in your tests "it's actually ever so slightly quicker on the PS4 than the Xbox one."
This is just shot with a standard 60fps camera, capturing 720p60 off-screen. You will see the first image in the twitter post has the LED off, and then it is lit on the subsequent frame. Counting from here (with the LED becoming lit as frame 0, and counting at 60fps even though the game is 30fps) you can see the xbox one responds on F10, while the PS4 responds on F14.
X1
https://twitter.com/noodalls/status/864442620407185408
PS4
https://twitter.com/noodalls/status/864443103003820032
So, these results demonstrate the same difference (X1 faster than PS4 by 66.66ms) between versions as my previous test. The controller is the same (Brook UFB FW 1.8). The screen is a BenQ RL series in game mode.
And yet in your tests "it's actually ever so slightly quicker on the PS4 than the Xbox one."
I have no idea what I'm looking at here... Were you doing the tests during movement? That's where he says that in the video.
NXGamer
Member
60FPS is not fast enough to be accurate in the tests, you will be out by 16.6ms at best, as such you wouldn't see the 4ms difference that can occur between them, UI to move split I mention Not the input delay as a whole.
And I tested movement not shooting, with the same results that the gap can be 33-50ms ( your capture speed is likely adding the extra 16Ms you get here), also the game(demo) varies in the game load and action in question.
So I am clear, you intercept the device feed,add in a pixel colour line edit when input is sent, show on screen and then count in 16Ms points when the screen updates?