![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
elvn
Supreme [H]ardness
- Joined
- May 5, 2006
- Messages
- 5,436
Online games buffer frames locally and on the server. In the case of valorant, which is a 128tick server with modern net code, it's 2 frames on the server and 3 frames on the client. So say 7.8ms x2 and 7.8ms x3 buffered on each end. If you are using DLSS + VRR, you are also seeing imagined frames separate from what the server decides is happening that you are acting on, and you are also buffering a frame there as well.
Games like modern Battlefield games do use a different method of determining "who shot first" for the occurrences where both people shoot on the same tick (the same tick as the server sees it) more or less, but considering all of the above, you don't see where the other player is temporally, accurately down to small ms levels when you are playing online games. So you don't see in sync what, where to aim at and thus when you decide to shoot to begin with before that "who shot first"calculation comes into play (most obnoxiously visible and made aware of this happening as peeker scenario, but it's a temporal gap/rubberband all of the time, the lack of sync is just more obvious and in your face in certain scenarios).
. . . .
That's not true. You are out of sync with the game server so you can shoot someone on your local screen who "isn't there anymore" as far as the server is concerned. Peeker's advantage shows this clearly, but the whole game world is out of sync, peeking just highlights just how much that is and with obnoxious results.
emphasis/bold lettering added by me in the quote below the link.
https://technology.riotgames.com/news/peeking-valorants-netcode
If they can peek in that far, then the reverse is true - they can get behind the corner for cover and from your perspective will still look like they were still standing there with you shooting them. The server sees them as having dodged behind the corner. This kind of thing, being out of sync with the game world is affecting the entire game world and gameplay, it's just most obvious in certain things.
It's not just input lag. In online games, you always see yourself ahead of where you actually are on the server, and you always see your opponent behind where they actually are on the server. The server goes back in time using the buffered frames system in an attempt to grant successful shot timing and other actions like player movement compared to (what your machine simulates to the server based on) what you saw locally. However different game's server code use their own biased design choices to resolve which player/action is successful, usually in regard to who's ping is higher or lower than the other - it's an interpolated/simulated result. As i said, the client also uses predicted frames in the online gaming system while it's waiting on the server's ticks.
A difference of a few ms of local input lag - operating on predicted frames since you are ahead of the server's tick rate, plus buffering on both ends, and then the server determining where everyone actually was at a lower rate and taking input lag compensation into account - is probably not an impactful difference. What makes a big difference is the length of the rubberband between you and the server, aka the amount of "peeker's advantage". Differences there of 10ms to 20ms would make a huge difference (e.g. a 60fpsHz player on a 128tick server would get 100ms of distortion, a 128fpsHz player would get 68ms of distortion on that same server) - - so your best case scenario is exceeding the tick rate as your fps lows and minimums (i.e. 128fps minimum, not average on a 128tick server) to reduce the peeker's advantage to the lowest the tick rate is capable of. Even then, "the move may arrive mid-frame and need to wait up to a full tick to be queued or processed" (+7.8ms), and also " may take an additional frame to render on the client" (+7.8ms), so it's sloppy there too.
That kind of meaningful gap reduction in online gameplay hinges on your fps vs the tick rate. However, a lot of game's tick rate is much lower than 128 tick , some are in the 20's ! .. so they would be much higher ms peeker's advantage gap ~ temporal distortion (and players thus being shown more predicted frames by the client etc.). Even if players are all experiencing the peeker's advantage gap more or less evenly on lower tick servers due to more achievable frame rates vs tick rates, that kind of huge gap in lower tick servers makes things much more wacky compared to the local predicted frames of gameplay you have being displayed your end while your client waits for the server's next interpolated tick. Things like teleporting around, rubberband "rewinds", and high fire rate weapons landing fewer shots as "super bullets" instead of peppering multiple lower damage shots throughout (among other things)
. .
The difference in ms between players who exceed the tick rate in fps and those who are beneath the tick rate in fps is - how far off what they are seeing is from what is happening as far as the server is concerned, rather than how fast the player can act input lag wise. Reaction wise, players react with 150ms to 180ms human reaction time plus their input lag chain, but they are reacting on predicted frames they are being shown by the client that don't correspond 1:1, and across peekers advantage distortion to how, when, and where things are on the server to begin with before they decide to act. Even with both players exceeding 128fpshz as their frame rate minimum on a 128 tick server, there will still be 72ms of distortion/"peekers advantage", plus a frame delay on the server and/or the client at times. "Frames of movement data are buffered at tick-granularity. Moves may arrive mid-frame and need to wait up to a full tick to be queued or processed." . . "Processed moves may take an additional frame to render on the client." (another frame, I believe he meant "frame" as 128tick peak, is 7.8ms)
The point I'm making is that - in online games - +2ms of input lag isn't going to make a difference overall, and on the 900D you'd still get something like 6ms input lag at over 200fpsHz, and 5ms at 240fpsHz. Such low input lag numbers won't have appreciable difference in online games for the above reasons, especially the fact that you aren't being shown were someone actually is temporily to react to compared to the server for several reasons. Local games, and LAN competitions, sure some appreciable gap in already low input lag numbers between player could make a difference in top skill level players if averaged out because all of the lower tick rates, buffering, prediction, biased interpolated results/compensation, etc won't be involved.
Online gaming displays to the player and resolves "fuzzy" to put it mildly. In order to get the lowest rubberband/peeker's advantage in online games you'd have to exceed the tick rate as your real frame rate minimum (not frame generated), and even then you are still experiencing some distortion of at least 72ms on a 128 tick server, with a lot of games way worse than that with much lower tick rates.
Games like modern Battlefield games do use a different method of determining "who shot first" for the occurrences where both people shoot on the same tick (the same tick as the server sees it) more or less, but considering all of the above, you don't see where the other player is temporally, accurately down to small ms levels when you are playing online games. So you don't see in sync what, where to aim at and thus when you decide to shoot to begin with before that "who shot first"calculation comes into play (most obnoxiously visible and made aware of this happening as peeker scenario, but it's a temporal gap/rubberband all of the time, the lack of sync is just more obvious and in your face in certain scenarios).
. . . .
That's not true. You are out of sync with the game server so you can shoot someone on your local screen who "isn't there anymore" as far as the server is concerned. Peeker's advantage shows this clearly, but the whole game world is out of sync, peeking just highlights just how much that is and with obnoxious results.
emphasis/bold lettering added by me in the quote below the link.
https://technology.riotgames.com/news/peeking-valorants-netcode
If they can peek in that far, then the reverse is true - they can get behind the corner for cover and from your perspective will still look like they were still standing there with you shooting them. The server sees them as having dodged behind the corner. This kind of thing, being out of sync with the game world is affecting the entire game world and gameplay, it's just most obvious in certain things.
It's not just input lag. In online games, you always see yourself ahead of where you actually are on the server, and you always see your opponent behind where they actually are on the server. The server goes back in time using the buffered frames system in an attempt to grant successful shot timing and other actions like player movement compared to (what your machine simulates to the server based on) what you saw locally. However different game's server code use their own biased design choices to resolve which player/action is successful, usually in regard to who's ping is higher or lower than the other - it's an interpolated/simulated result. As i said, the client also uses predicted frames in the online gaming system while it's waiting on the server's ticks.
A difference of a few ms of local input lag - operating on predicted frames since you are ahead of the server's tick rate, plus buffering on both ends, and then the server determining where everyone actually was at a lower rate and taking input lag compensation into account - is probably not an impactful difference. What makes a big difference is the length of the rubberband between you and the server, aka the amount of "peeker's advantage". Differences there of 10ms to 20ms would make a huge difference (e.g. a 60fpsHz player on a 128tick server would get 100ms of distortion, a 128fpsHz player would get 68ms of distortion on that same server) - - so your best case scenario is exceeding the tick rate as your fps lows and minimums (i.e. 128fps minimum, not average on a 128tick server) to reduce the peeker's advantage to the lowest the tick rate is capable of. Even then, "the move may arrive mid-frame and need to wait up to a full tick to be queued or processed" (+7.8ms), and also " may take an additional frame to render on the client" (+7.8ms), so it's sloppy there too.
That kind of meaningful gap reduction in online gameplay hinges on your fps vs the tick rate. However, a lot of game's tick rate is much lower than 128 tick , some are in the 20's ! .. so they would be much higher ms peeker's advantage gap ~ temporal distortion (and players thus being shown more predicted frames by the client etc.). Even if players are all experiencing the peeker's advantage gap more or less evenly on lower tick servers due to more achievable frame rates vs tick rates, that kind of huge gap in lower tick servers makes things much more wacky compared to the local predicted frames of gameplay you have being displayed your end while your client waits for the server's next interpolated tick. Things like teleporting around, rubberband "rewinds", and high fire rate weapons landing fewer shots as "super bullets" instead of peppering multiple lower damage shots throughout (among other things)
. .
The difference in ms between players who exceed the tick rate in fps and those who are beneath the tick rate in fps is - how far off what they are seeing is from what is happening as far as the server is concerned, rather than how fast the player can act input lag wise. Reaction wise, players react with 150ms to 180ms human reaction time plus their input lag chain, but they are reacting on predicted frames they are being shown by the client that don't correspond 1:1, and across peekers advantage distortion to how, when, and where things are on the server to begin with before they decide to act. Even with both players exceeding 128fpshz as their frame rate minimum on a 128 tick server, there will still be 72ms of distortion/"peekers advantage", plus a frame delay on the server and/or the client at times. "Frames of movement data are buffered at tick-granularity. Moves may arrive mid-frame and need to wait up to a full tick to be queued or processed." . . "Processed moves may take an additional frame to render on the client." (another frame, I believe he meant "frame" as 128tick peak, is 7.8ms)
The point I'm making is that - in online games - +2ms of input lag isn't going to make a difference overall, and on the 900D you'd still get something like 6ms input lag at over 200fpsHz, and 5ms at 240fpsHz. Such low input lag numbers won't have appreciable difference in online games for the above reasons, especially the fact that you aren't being shown were someone actually is temporily to react to compared to the server for several reasons. Local games, and LAN competitions, sure some appreciable gap in already low input lag numbers between player could make a difference in top skill level players if averaged out because all of the lower tick rates, buffering, prediction, biased interpolated results/compensation, etc won't be involved.
Online gaming displays to the player and resolves "fuzzy" to put it mildly. In order to get the lowest rubberband/peeker's advantage in online games you'd have to exceed the tick rate as your real frame rate minimum (not frame generated), and even then you are still experiencing some distortion of at least 72ms on a 128 tick server, with a lot of games way worse than that with much lower tick rates.
Last edited:
