Bad connection question

[Gooberking]

Lets say there are two people playing, and one of then has a good connection while the other does not. We will say player A will be the good connection, player B will be the bad connection.

The person B with a "bad" connection is introducing ping swings (jitter) of something like 40-70ms. Some large amount where it's clearly messing with the game, but not so bad that it spikes pings up like 300ms+ and causes lots of auto pausing. (pausing being where inputs are so delayed the game just has to stop to wait for inputs and can't use rollbacks to cover it up; as discussed in the NRS netcode keynote

)

My question is do both people have to deal with that instability, or is only person A with the good connection the only one dealing with the timings being erratic? It seems like the person causing the problem would not even see how much rollback they are inflicting onto person A. Their game is going to cook like normal as long as it doesn't force pauses right? It will just be sending input history using pack mules, leaving gamer A to hold all the rollback.

I think I always assumed the experience was mutual, but thinking about how stuff works and seeing how people play, I've started to think that isn't the case. That when you wonder how someone can play in that connection, then answer may be that they aren't.

 

[AREZ God of War]

You can probably use mirror matches to test it by seeing if you are both able to do the same stuff.

 

[Rodney Quillz]

Lets say there are two people playing, and one of then has a good connection while the other does not. We will say player A will be the good connection, player B will be the bad connection.

The person B with a "bad" connection is introducing ping swings (jitter) of something like 40-70ms. Some large amount where it's clearly messing with the game, but not so bad that it spikes pings up like 300ms+ and causes lots of auto pausing. (pausing being where inputs are so delayed the game just has to stop to wait for inputs and can't use rollbacks to cover it up; as discussed in the NRS netcode keynote

)

My question is do both people have to deal with that instability, or is only person A with the good connection the only one dealing with the timings being erratic? It seems like the person causing the problem would not even see how much rollback they are inflicting onto person A. Their game is going to cook like normal as long as it doesn't force pauses right? It will just be sending input history using pack mules, leaving gamer A to hold all the rollback.

I think I always assumed the experience was mutual, but thinking about how stuff works and seeing how people play, I've started to think that isn't the case. That when you wonder how someone can play in that connection, then answer may be that they aren't.

They have more than likely adapted to their connection.

 

[fr stack]

Yeah they are one with the matrix of mc donalds wifi
The mc matrix if u will

 

[Gooberking]

You can probably use mirror matches to test it by seeing if you are both able to do the same stuff.

I'd be curious to see what it's like. I guess the good connection's inputs would still be coming in compromised, so it probably does suck all around.

FYI, if anyone thinks it's normal to have 30+ ping swings in matches, it's not.

 

[THTB]

The short answer is yes, both sides will deal with the instability, provided the coding is designed to keep the games as in sync as possible.

The long answer, which hopefully I'm explaining correctly...the way rollback (properly implemented) works is that it works frame-by-frame to keep both connections in sync on a clock level, and will only allow one player to be out of sync by a specific maximum value of frames. With a spiky connection, the game can and will start to desync at various points. After that point, the game starts trying to resync by pausing the game in bursts to get the clocks back on track. If the jitter manages to spike the connection and put things out of sync without reaching this threshold, you should be seeing an increase in momentary instances of rollback correction, such as correcting whether an attack hits or not or a lot of startup frames being cut, until the connection sorts itself out.

This is all assuming the developers put in the coding to work to keep sync between games. In a game that is infamous for this not being the case (pre-CE fix SFV), the game would fall out of sync, and one player would be ahead by a potential 15 frames without any correction until the next round, causing one or both sides to have tons of skipping that would persist until round conclusion. And even if this doesn't get that extreme, the rollback will oftentimes be problematic on one end, but not the other. In games with proper rollback, this one-sided rollback isn't uncommon, but the game is constantly working to never let this ever really show in meaningful manners for long.

Again, I could be missing some important details, so anyone who fully understands rollback would probably give a much better explanation, but the point is, yes...it is very possible for games to feel one-sided in terms of connection quality, but given the way NRS codes their netplay (which is regarded as extremely robust), it's likely not often you'll even notice this, and would require footage from both ends.