Do you also get increased lag whenever there is a ballgroup nearby?

InvictusApollo

I'm not talking about a normal group of random people. I'm talking about those "skilled", "elite" groups that stack closely and allways spam buffs and aoe heals being almost invincible, wait for their ults to charge and then "coordinate" an ult drop.
Every time I fight one, I get much higher lag. Usually it doubles just because of that groups presence. It doesn't happen with normal groups. For a normal group to affect my lag that much, it has to be two or three times larger than a ball group.
Is anyone else experiencing this?

Gilvoth

yes,
we all do.

Savos_Saren

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

zyk

Without a doubt. But there are different reasons.

They contribute disproportionately to server processing lag because they layer so many calculations on each other. Skilled players in large groups playing in a tight formation have the potential to create the most calculations on a per-player basis.

The very good groups you reference who claim to fight 50+ randoms on a regular basically create the same conditions as a multi-raid PUG like AOTP can.

Also, a lot of these groups love backdooring already big, laggy fights. This usually makes things much worse for everyone. This is extremely aggravating. There are times when this must happen -- like when a faction is turtling on scrolls/emp, but most of the time, it's just for laughs I think.

In the past, when there were more ball groups, they used to keep each other in check a little better than today. Often only one side has such a group running and they can be like kids in a candy store and drawing dozens of randoms wherever they go.

Ruckly

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

Even worse is when you get smoked by a hard-drive targeted DOS and can't even close ESO. This seems like the hip new thing amongst some of Cyrodills finer patrons.

LeifErickson

Yes

JaZ2091

Ruckly wrote: »

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

Even worse is when you get smoked by a hard-drive targeted DOS and can't even close ESO. This seems like the hip new thing amongst some of Cyrodills finer patrons.

Thats a little extreme for mediocre PVP lol.

idk

It does not have to do with a 'ball group" but just the sheer number of players in the area. Attacks on keeps tend to draw crowds and in a busy campaign like Vivek many will not be in groups or will be in small groups.

zyk

idk wrote: »

It does not have to do with a 'ball group" but just the sheer number of players in the area. Attacks on keeps tend to draw crowds and in a busy campaign like Vivek many will not be in groups or will be in small groups.

This isn't accurate. The number of players does not directly correlate to server processing lag. What matters is the number of concurrent calculations the server must process. The number of players present is one factor in that, but what they do is just as important. 50 players just light attacking probably aren't going to stress the server. 50 players on top of each other spamming AEs will.

Because ball groups play in a tight formation, they contribute far more to server lag than the same players would spread apart because their AE abilities and passives won't affect each other as much, reducing the number of calculations the server must process.

idk

zyk wrote: »

idk wrote: »

It does not have to do with a 'ball group" but just the sheer number of players in the area. Attacks on keeps tend to draw crowds and in a busy campaign like Vivek many will not be in groups or will be in small groups.

This isn't accurate. The number of players does not directly correlate to server processing lag. What matters is the number of concurrent calculations the server must process. The number of players present is one factor in that, but what they do is just as important. 50 players just light attacking probably aren't going to stress the server. 50 players on top of each other spamming AEs will.

Because ball groups play in a tight formation, they contribute far more to server lag than the same players would spread apart because their AE abilities and passives won't affect each other as much, reducing the number of calculations the server must process.

Umm, exactly. It is irrelevant how they are grouped. So ball groups are not the cause in itself.

Of course a bunch of players just standing around with their thumbs up their rumps are not going to stress the server much. That seemed extremely obvious, but thx for making sure just in case some did not innately understand that.

zyk

idk wrote: »

Of course a bunch of players just standing around with their thumbs up their rumps are not going to stress the server much. That seemed extremely obvious, but thx for making sure just in case some did not innately understand that.

The point is that being balled is more calculation intensive. So yes, players who are balled up cause more server processing lag than those who aren't.

Basically, strong ball groups are the perfect storm for server processing lag:
- tight formation
- large group
- skilled players executing abilities on cooldown
- optimal builds allowing for high sustain and utilizing proc sets
- use AEs (damage/heals/utility)
- can draw 2-5 times their numbers to one location

idk

zyk wrote: »

idk wrote: »

Of course a bunch of players just standing around with their thumbs up their rumps are not going to stress the server much. That seemed extremely obvious, but thx for making sure just in case some did not innately understand that.

The point is that being balled is more calculation intensive. So yes, players who are balled up cause more server processing lag than those who aren't.

Basically, strong ball groups are the perfect storm for server processing lag:
- tight formation
- large group
- skilled players executing abilities on cooldown
- optimal builds allowing for high sustain and utilizing proc sets
- use AEs (damage/heals/utility)
- can draw 2-5 times their numbers to one location

You make it look as though eliminating ball groups solves all the problems most of the time which is horribly misleading because it is factually incorrect.

zyk

idk wrote: »

You make it look as though eliminating ball groups solves all the problems most of the time which is horribly misleading because it is factually incorrect.

If it looks that way to you, don't blame me because I didn't write that.

This is what ZOS has been asking us to do the entire time: spread out. Ball groups are the polar opposite of that.

Xsorus

Yes.. I feel like I need to post all my old videos to show this has been a problem forever.

frostz417

No I don’t lag near ball groups. I lag near 80 man zergs

frostz417

zyk wrote: »

idk wrote: »

You make it look as though eliminating ball groups solves all the problems most of the time which is horribly misleading because it is factually incorrect.

If it looks that way to you, don't blame me because I didn't write that.

This is what ZOS has been asking us to do the entire time: spread out. Ball groups are the polar opposite of that.

You act as if people who are in 60-80
Man Zergs aren’t the problem but 12 man ball groups are... if anything ball groups help split people up since they farm 60-80 man Zergs with AOE abilities.. ball groups aren’t the issue, faction stacking 60-80 man pug Zergs are the issue.

zyk

frostz417 wrote: »

You act as if people who are in 60-80
Man Zergs aren’t the problem but 12 man ball groups are... if anything ball groups help split people up since they farm 60-80 man Zergs with AOE abilities.. ball groups aren’t the issue, faction stacking 60-80 man pug Zergs are the issue.

No, I don't. I very specifically explained why balled players contribute more lag per-player than the same players when not balled.

I act like the problem is too many concurrent calculations for the server to process smoothly, which it is.

visionality

Indeed ball-groups intensify lag due to the way they play. But they dont do it out of spite, they would prefer to play lag-free as much as anybody else in the game (possibly more). The server just cannot handle the kind of gameplay a group of experienced players can do.

ZOS' call to spread out to reduce lag is ridiculous, though. Players will always go where they feel gaming is best, and usually thats the one big X on the map - more kills, more AP, more healing from other players around. If ZOS wants ppl to NOT clog up in one big blob, they need to introduce a mechanic that makes 30+ zergs an increasing disadvantage for players. E.g. by introducing some kind of plague starting to weaken and kill players as soon as a critical number of them is in the vicinity (no matter whether grouped or not).

TriangularChicken

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

This! The day ZOS comes up with an idea to make people spread over the map is the day that a) lag will be gone and b) ball groups will disappear. Exactly in that order, noobs stacking in one place to get 6k tick for not doing anything are the problem in Cyrodiil. Ball groups contribute to lag, but not as much as faction stackers do. That's coming from somebody that does not play in a ball group.

InvictusApollo

RealDolanTrump wrote: »

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

This! The day ZOS comes up with an idea to make people spread over the map is the day that a) lag will be gone and b) ball groups will disappear. Exactly in that order, noobs stacking in one place to get 6k tick for not doing anything are the problem in Cyrodiil. Ball groups contribute to lag, but not as much as faction stackers do. That's coming from somebody that does not play in a ball group.

We had a day like that. Even several. It was during the "siege bug". Ping was great. Game was fluid. Ball groups didn't exist because they were too afraid to stack.
We had epic battles. And even though there were dozens of players on the battlefield we still had great ping.
I think it was because they were spread out so their aoe abilities weren't affecting other players.

Experiences of so many players with ball groups inducing lag prove that spamming multiple aoe abilities on multiple players produces significant strain on servers which in effect causes lag.

I haven't seen ESO's source code but given your reports I can make an educated guess that the algorythm looks like follows:

1. Player casts some ability. If it is single target, go to step 3. If it is aoe go to step 2.
2. A query to the database is executed. Query returns a list of all players that are within aoe range.
3. System makes calculations on all affected players. If there was one target then we get only one calculation. If there were multiple targets then we get multiple calculations.

This means that when one player casts an aoe ability (heal for example) on n players, then we get n calculations. In ball groups however all players cast on average one aoe ability per second. This means that a ball group generates n^2 calculations per second. And this is assuming that players cast only aoe abilities that last for one second or put an effect that lasts more seconds but doesn't stay on ground. If players cast ground based aoe effects that remain on the ground we get even more calculations.
This number of calculations per second is vastly increased whenever a ball group rolls over another group. In that case the n^2 is multiplied by the number of players in opposing group.

Lets analyze two scenarios:
1. A ballgroup the size of Nb fights a normal group of Nn players. Where Nn = Nb*k.
Total player number: Tp = Nb + Nb*k = Nb*(1+k) = Nn + Nn/k = Nn*(1+1/k)
We have two phases during a fight:
Phase 1: Balllgroup charges their ultimates. During that time every player from the normal group either casts aoe ability on ball group, single target ability on anyone or aoe ability on just several players of normal group. Lets assume the most optimisitic choice (for the number of calculations) which is that he casts a single target ability on someone.
In this phase we have total number of calculations per second (Tcs):
Tcs = Nb*Nb + Nn

Phase 2: Ballgroup rushes in spamming aoes and activating their aoe ultimates. Every ballgroup player affects multiple players from normal group and all players from ball group. Lets say that j percent of all players from the normal group are affected.
Tcs = Nb*(Nb + Nn*j) + Nn


2. 2. A normal group of Nn players fights a normal group of Nn players. AOEs may be used but if everyone is spread out then an aoe affects only one player and induces only additional database query to check which player was affected in comparison to activating a single target ability that already returns an id of affected player.
Tcs = Nn+Nn

Those calculations do not take into consideration proc sets that increase calculations per second. This means that a ball group effect on lag is even greater since they often use proc sets (Earthgore for example).

As you can see a ball group produces significantly more calculations per second.
Lets talk about some real numbers to visualize you this.
Lets say that there are 100 players in a normal group and 20 players in ballgroup. And during ballgroup charge they affect one tenth of all players from normal group per second.
In that case we get total number of calculations per second:
Tcs = 20*(20 + 100/10) + 100 = 700
In comparison if there are 100 players on two normal groups, the total number of calculations is:
Tcs = 100+100 = 200
Which is a lot less calculations with almost twice the number of all players.

Conclusions:
1. Ballgroups significantly increase lag compared to normal groups.
2. The bigger the ballgroup, the more lag we get. It has been numerously observed that a ballgroup and normal group large enough can even cause a crash.


How to deal with this?

• Increase server efficiency by a factor of five. That is impossible. Code optimization could decrease lag only a little. Server upgrade would also decrease lag only a little.
• Implement game mechanics that encourage players to spread out. Not only spread out on the map but also spread out on the battlefield. Implementing daedric artifacts won't be enough. This will only create one more big battle on the map, but players will still stack.
• Put a limit of affected players on every aoe ability. For example healing springs would affect only two players at once. This is unacceptable. And would require massive balance changes.
• Move calculations from server to client computers. This would reduce the lag by the time it takes a server to make all calculations but would also increase it, because each computer would have to report more to the server. What is unacceptable with this solution is that cheating would be extremely easy and almost undetectable.

idk

zyk wrote: »

idk wrote: »

You make it look as though eliminating ball groups solves all the problems most of the time which is horribly misleading because it is factually incorrect.

If it looks that way to you, don't blame me because I didn't write that.

This is what ZOS has been asking us to do the entire time: spread out. Ball groups are the polar opposite of that.

You are acting as though the only time you can ever have a large number of players in close proximity attacking each other is with ball groups. I am telling you that you are wrong. Players tend to gravitate to where the action is.

Soul_Demon

InvictusApollo wrote: »

RealDolanTrump wrote: »

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

This! The day ZOS comes up with an idea to make people spread over the map is the day that a) lag will be gone and b) ball groups will disappear. Exactly in that order, noobs stacking in one place to get 6k tick for not doing anything are the problem in Cyrodiil. Ball groups contribute to lag, but not as much as faction stackers do. That's coming from somebody that does not play in a ball group.

We had a day like that. Even several. It was during the "siege bug". Ping was great. Game was fluid. Ball groups didn't exist because they were too afraid to stack.
We had epic battles. And even though there were dozens of players on the battlefield we still had great ping.
I think it was because they were spread out so their aoe abilities weren't affecting other players.

Experiences of so many players with ball groups inducing lag prove that spamming multiple aoe abilities on multiple players produces significant strain on servers which in effect causes lag.

I haven't seen ESO's source code but given your reports I can make an educated guess that the algorythm looks like follows:

1. Player casts some ability. If it is single target, go to step 3. If it is aoe go to step 2.
2. A query to the database is executed. Query returns a list of all players that are within aoe range.
3. System makes calculations on all affected players. If there was one target then we get only one calculation. If there were multiple targets then we get multiple calculations.

This means that when one player casts an aoe ability (heal for example) on n players, then we get n calculations. In ball groups however all players cast on average one aoe ability per second. This means that a ball group generates n^2 calculations per second. And this is assuming that players cast only aoe abilities that last for one second or put an effect that lasts more seconds but doesn't stay on ground. If players cast ground based aoe effects that remain on the ground we get even more calculations.
This number of calculations per second is vastly increased whenever a ball group rolls over another group. In that case the n^2 is multiplied by the number of players in opposing group.

Lets analyze two scenarios:
1. A ballgroup the size of Nb fights a normal group of Nn players. Where Nn = Nb*k.
Total player number: Tp = Nb + Nb*k = Nb*(1+k) = Nn + Nn/k = Nn*(1+1/k)
We have two phases during a fight:
Phase 1: Balllgroup charges their ultimates. During that time every player from the normal group either casts aoe ability on ball group, single target ability on anyone or aoe ability on just several players of normal group. Lets assume the most optimisitic choice (for the number of calculations) which is that he casts a single target ability on someone.
In this phase we have total number of calculations per second (Tcs):
Tcs = Nb*Nb + Nn

Phase 2: Ballgroup rushes in spamming aoes and activating their aoe ultimates. Every ballgroup player affects multiple players from normal group and all players from ball group. Lets say that j percent of all players from the normal group are affected.
Tcs = Nb*(Nb + Nn*j) + Nn


2. 2. A normal group of Nn players fights a normal group of Nn players. AOEs may be used but if everyone is spread out then an aoe affects only one player and induces only additional database query to check which player was affected in comparison to activating a single target ability that already returns an id of affected player.
Tcs = Nn+Nn

Those calculations do not take into consideration proc sets that increase calculations per second. This means that a ball group effect on lag is even greater since they often use proc sets (Earthgore for example).

As you can see a ball group produces significantly more calculations per second.
Lets talk about some real numbers to visualize you this.
Lets say that there are 100 players in a normal group and 20 players in ballgroup. And during ballgroup charge they affect one tenth of all players from normal group per second.
In that case we get total number of calculations per second:
Tcs = 20*(20 + 100/10) + 100 = 700
In comparison if there are 100 players on two normal groups, the total number of calculations is:
Tcs = 100+100 = 200
Which is a lot less calculations with almost twice the number of all players.

Conclusions:
1. Ballgroups significantly increase lag compared to normal groups.
2. The bigger the ballgroup, the more lag we get. It has been numerously observed that a ballgroup and normal group large enough can even cause a crash.


How to deal with this?

• Increase server efficiency by a factor of five. That is impossible. Code optimization could decrease lag only a little. Server upgrade would also decrease lag only a little.
• Implement game mechanics that encourage players to spread out. Not only spread out on the map but also spread out on the battlefield. Implementing daedric artifacts won't be enough. This will only create one more big battle on the map, but players will still stack.
• Put a limit of affected players on every aoe ability. For example healing springs would affect only two players at once. This is unacceptable. And would require massive balance changes.
• Move calculations from server to client computers. This would reduce the lag by the time it takes a server to make all calculations but would also increase it, because each computer would have to report more to the server. What is unacceptable with this solution is that cheating would be extremely easy and almost undetectable.

Would like to throw out one more solution- before the gold farmer 'Fix' implementation the lag was almost non existent in this game that was designed for 100's of players at one place at one time. There are videos out there before the ZOS rush to stop farmers and gold sellers early on that is the genesis of all lag we currently see today. So, if code exists that was nearly lag free and they introduced 'fix' to protect their profits in the store (that is what they did) now that we have a fraction of that population- an additional solution would be to assess now if the lag we tolerate is the only way to protect those profits or potential profits they introduced it to protect...namely farmers and gold sellers- do they present the problem with tiny fraction of population now that they did in the beginning of the game? Question ZOS will have to answer, they have the original 'unfixed' version without lag.

InvictusApollo

Soul_Demon wrote: »

InvictusApollo wrote: »

RealDolanTrump wrote: »

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

This! The day ZOS comes up with an idea to make people spread over the map is the day that a) lag will be gone and b) ball groups will disappear. Exactly in that order, noobs stacking in one place to get 6k tick for not doing anything are the problem in Cyrodiil. Ball groups contribute to lag, but not as much as faction stackers do. That's coming from somebody that does not play in a ball group.

We had a day like that. Even several. It was during the "siege bug". Ping was great. Game was fluid. Ball groups didn't exist because they were too afraid to stack.
We had epic battles. And even though there were dozens of players on the battlefield we still had great ping.
I think it was because they were spread out so their aoe abilities weren't affecting other players.

Experiences of so many players with ball groups inducing lag prove that spamming multiple aoe abilities on multiple players produces significant strain on servers which in effect causes lag.

I haven't seen ESO's source code but given your reports I can make an educated guess that the algorythm looks like follows:

1. Player casts some ability. If it is single target, go to step 3. If it is aoe go to step 2.
2. A query to the database is executed. Query returns a list of all players that are within aoe range.
3. System makes calculations on all affected players. If there was one target then we get only one calculation. If there were multiple targets then we get multiple calculations.

This means that when one player casts an aoe ability (heal for example) on n players, then we get n calculations. In ball groups however all players cast on average one aoe ability per second. This means that a ball group generates n^2 calculations per second. And this is assuming that players cast only aoe abilities that last for one second or put an effect that lasts more seconds but doesn't stay on ground. If players cast ground based aoe effects that remain on the ground we get even more calculations.
This number of calculations per second is vastly increased whenever a ball group rolls over another group. In that case the n^2 is multiplied by the number of players in opposing group.

Lets analyze two scenarios:
1. A ballgroup the size of Nb fights a normal group of Nn players. Where Nn = Nb*k.
Total player number: Tp = Nb + Nb*k = Nb*(1+k) = Nn + Nn/k = Nn*(1+1/k)
We have two phases during a fight:
Phase 1: Balllgroup charges their ultimates. During that time every player from the normal group either casts aoe ability on ball group, single target ability on anyone or aoe ability on just several players of normal group. Lets assume the most optimisitic choice (for the number of calculations) which is that he casts a single target ability on someone.
In this phase we have total number of calculations per second (Tcs):
Tcs = Nb*Nb + Nn

Phase 2: Ballgroup rushes in spamming aoes and activating their aoe ultimates. Every ballgroup player affects multiple players from normal group and all players from ball group. Lets say that j percent of all players from the normal group are affected.
Tcs = Nb*(Nb + Nn*j) + Nn


2. 2. A normal group of Nn players fights a normal group of Nn players. AOEs may be used but if everyone is spread out then an aoe affects only one player and induces only additional database query to check which player was affected in comparison to activating a single target ability that already returns an id of affected player.
Tcs = Nn+Nn

Those calculations do not take into consideration proc sets that increase calculations per second. This means that a ball group effect on lag is even greater since they often use proc sets (Earthgore for example).

As you can see a ball group produces significantly more calculations per second.
Lets talk about some real numbers to visualize you this.
Lets say that there are 100 players in a normal group and 20 players in ballgroup. And during ballgroup charge they affect one tenth of all players from normal group per second.
In that case we get total number of calculations per second:
Tcs = 20*(20 + 100/10) + 100 = 700
In comparison if there are 100 players on two normal groups, the total number of calculations is:
Tcs = 100+100 = 200
Which is a lot less calculations with almost twice the number of all players.

Conclusions:
1. Ballgroups significantly increase lag compared to normal groups.
2. The bigger the ballgroup, the more lag we get. It has been numerously observed that a ballgroup and normal group large enough can even cause a crash.


How to deal with this?

• Increase server efficiency by a factor of five. That is impossible. Code optimization could decrease lag only a little. Server upgrade would also decrease lag only a little.
• Implement game mechanics that encourage players to spread out. Not only spread out on the map but also spread out on the battlefield. Implementing daedric artifacts won't be enough. This will only create one more big battle on the map, but players will still stack.
• Put a limit of affected players on every aoe ability. For example healing springs would affect only two players at once. This is unacceptable. And would require massive balance changes.
• Move calculations from server to client computers. This would reduce the lag by the time it takes a server to make all calculations but would also increase it, because each computer would have to report more to the server. What is unacceptable with this solution is that cheating would be extremely easy and almost undetectable.

Would like to throw out one more solution- before the gold farmer 'Fix' implementation the lag was almost non existent in this game that was designed for 100's of players at one place at one time. There are videos out there before the ZOS rush to stop farmers and gold sellers early on that is the genesis of all lag we currently see today. So, if code exists that was nearly lag free and they introduced 'fix' to protect their profits in the store (that is what they did) now that we have a fraction of that population- an additional solution would be to assess now if the lag we tolerate is the only way to protect those profits or potential profits they introduced it to protect...namely farmers and gold sellers- do they present the problem with tiny fraction of population now that they did in the beginning of the game? Question ZOS will have to answer, they have the original 'unfixed' version without lag.

What "Fix" are you talking about? (Also please use commas to make your sentences more understandable.)

Edirt_seliv

Soul_Demon wrote: »

InvictusApollo wrote: »

RealDolanTrump wrote: »

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

This! The day ZOS comes up with an idea to make people spread over the map is the day that a) lag will be gone and b) ball groups will disappear. Exactly in that order, noobs stacking in one place to get 6k tick for not doing anything are the problem in Cyrodiil. Ball groups contribute to lag, but not as much as faction stackers do. That's coming from somebody that does not play in a ball group.

We had a day like that. Even several. It was during the "siege bug". Ping was great. Game was fluid. Ball groups didn't exist because they were too afraid to stack.
We had epic battles. And even though there were dozens of players on the battlefield we still had great ping.
I think it was because they were spread out so their aoe abilities weren't affecting other players.

Experiences of so many players with ball groups inducing lag prove that spamming multiple aoe abilities on multiple players produces significant strain on servers which in effect causes lag.

I haven't seen ESO's source code but given your reports I can make an educated guess that the algorythm looks like follows:

1. Player casts some ability. If it is single target, go to step 3. If it is aoe go to step 2.
2. A query to the database is executed. Query returns a list of all players that are within aoe range.
3. System makes calculations on all affected players. If there was one target then we get only one calculation. If there were multiple targets then we get multiple calculations.

This means that when one player casts an aoe ability (heal for example) on n players, then we get n calculations. In ball groups however all players cast on average one aoe ability per second. This means that a ball group generates n^2 calculations per second. And this is assuming that players cast only aoe abilities that last for one second or put an effect that lasts more seconds but doesn't stay on ground. If players cast ground based aoe effects that remain on the ground we get even more calculations.
This number of calculations per second is vastly increased whenever a ball group rolls over another group. In that case the n^2 is multiplied by the number of players in opposing group.

Lets analyze two scenarios:
1. A ballgroup the size of Nb fights a normal group of Nn players. Where Nn = Nb*k.
Total player number: Tp = Nb + Nb*k = Nb*(1+k) = Nn + Nn/k = Nn*(1+1/k)
We have two phases during a fight:
Phase 1: Balllgroup charges their ultimates. During that time every player from the normal group either casts aoe ability on ball group, single target ability on anyone or aoe ability on just several players of normal group. Lets assume the most optimisitic choice (for the number of calculations) which is that he casts a single target ability on someone.
In this phase we have total number of calculations per second (Tcs):
Tcs = Nb*Nb + Nn

Phase 2: Ballgroup rushes in spamming aoes and activating their aoe ultimates. Every ballgroup player affects multiple players from normal group and all players from ball group. Lets say that j percent of all players from the normal group are affected.
Tcs = Nb*(Nb + Nn*j) + Nn


2. 2. A normal group of Nn players fights a normal group of Nn players. AOEs may be used but if everyone is spread out then an aoe affects only one player and induces only additional database query to check which player was affected in comparison to activating a single target ability that already returns an id of affected player.
Tcs = Nn+Nn

Those calculations do not take into consideration proc sets that increase calculations per second. This means that a ball group effect on lag is even greater since they often use proc sets (Earthgore for example).

As you can see a ball group produces significantly more calculations per second.
Lets talk about some real numbers to visualize you this.
Lets say that there are 100 players in a normal group and 20 players in ballgroup. And during ballgroup charge they affect one tenth of all players from normal group per second.
In that case we get total number of calculations per second:
Tcs = 20*(20 + 100/10) + 100 = 700
In comparison if there are 100 players on two normal groups, the total number of calculations is:
Tcs = 100+100 = 200
Which is a lot less calculations with almost twice the number of all players.

Conclusions:
1. Ballgroups significantly increase lag compared to normal groups.
2. The bigger the ballgroup, the more lag we get. It has been numerously observed that a ballgroup and normal group large enough can even cause a crash.


How to deal with this?

• Increase server efficiency by a factor of five. That is impossible. Code optimization could decrease lag only a little. Server upgrade would also decrease lag only a little.
• Implement game mechanics that encourage players to spread out. Not only spread out on the map but also spread out on the battlefield. Implementing daedric artifacts won't be enough. This will only create one more big battle on the map, but players will still stack.
• Put a limit of affected players on every aoe ability. For example healing springs would affect only two players at once. This is unacceptable. And would require massive balance changes.
• Move calculations from server to client computers. This would reduce the lag by the time it takes a server to make all calculations but would also increase it, because each computer would have to report more to the server. What is unacceptable with this solution is that cheating would be extremely easy and almost undetectable.

Would like to throw out one more solution- before the gold farmer 'Fix' implementation the lag was almost non existent in this game that was designed for 100's of players at one place at one time. There are videos out there before the ZOS rush to stop farmers and gold sellers early on that is the genesis of all lag we currently see today. So, if code exists that was nearly lag free and they introduced 'fix' to protect their profits in the store (that is what they did) now that we have a fraction of that population- an additional solution would be to assess now if the lag we tolerate is the only way to protect those profits or potential profits they introduced it to protect...namely farmers and gold sellers- do they present the problem with tiny fraction of population now that they did in the beginning of the game? Question ZOS will have to answer, they have the original 'unfixed' version without lag.

You also have to remember that back during the times you are talking about, ball groups (and pretty much everyone in cyrodil) were almost unanimously wearing warlock and seducer. There are hundreds more sets now with added calculations, there are more passives, more abilities etc.

Now don't get me wrong, I far from disagree with you regarding ball groups. By zyks 'the sky is falling' estimation of the subject one would be inclined to think that ball group v ball group fights would basically set fire to your pc. Which is the complete opposite to what you will experience out there. Gvgs are the best examples. I have been a part of many gvg fights in the middle of nowhere with other ball groups of 16-24+.

These fights were always the most fun, not only because it was more engaging than fighting mindless pugs, but because it was often the most crisp gameplay, lagfree compared to taking a populated keep etc.

So to answer the OPs question, when in a vacuum, no I do not get lag when ball groups are nearby. Now, understandably "in a vacuum" isn't exactly a pursuasive caveat, but the point is, if you just want to close your eyes and point your finger at ball groups and say "there's your lag fellas" then you'll wind up just looking as sallty as our good friend zyklosis here.

Alucardo

Whoever answers no is obviously lying, or didn't understand the question.

Mr_Walker

Damn some of you get upset at people playing as a team. I can taste the salt in the OP.

Soul_Demon

InvictusApollo wrote: »

Soul_Demon wrote: »

InvictusApollo wrote: »

RealDolanTrump wrote: »

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

This! The day ZOS comes up with an idea to make people spread over the map is the day that a) lag will be gone and b) ball groups will disappear. Exactly in that order, noobs stacking in one place to get 6k tick for not doing anything are the problem in Cyrodiil. Ball groups contribute to lag, but not as much as faction stackers do. That's coming from somebody that does not play in a ball group.

We had a day like that. Even several. It was during the "siege bug". Ping was great. Game was fluid. Ball groups didn't exist because they were too afraid to stack.
We had epic battles. And even though there were dozens of players on the battlefield we still had great ping.
I think it was because they were spread out so their aoe abilities weren't affecting other players.

Experiences of so many players with ball groups inducing lag prove that spamming multiple aoe abilities on multiple players produces significant strain on servers which in effect causes lag.

I haven't seen ESO's source code but given your reports I can make an educated guess that the algorythm looks like follows:

1. Player casts some ability. If it is single target, go to step 3. If it is aoe go to step 2.
2. A query to the database is executed. Query returns a list of all players that are within aoe range.
3. System makes calculations on all affected players. If there was one target then we get only one calculation. If there were multiple targets then we get multiple calculations.

This means that when one player casts an aoe ability (heal for example) on n players, then we get n calculations. In ball groups however all players cast on average one aoe ability per second. This means that a ball group generates n^2 calculations per second. And this is assuming that players cast only aoe abilities that last for one second or put an effect that lasts more seconds but doesn't stay on ground. If players cast ground based aoe effects that remain on the ground we get even more calculations.
This number of calculations per second is vastly increased whenever a ball group rolls over another group. In that case the n^2 is multiplied by the number of players in opposing group.

Lets analyze two scenarios:
1. A ballgroup the size of Nb fights a normal group of Nn players. Where Nn = Nb*k.
Total player number: Tp = Nb + Nb*k = Nb*(1+k) = Nn + Nn/k = Nn*(1+1/k)
We have two phases during a fight:
Phase 1: Balllgroup charges their ultimates. During that time every player from the normal group either casts aoe ability on ball group, single target ability on anyone or aoe ability on just several players of normal group. Lets assume the most optimisitic choice (for the number of calculations) which is that he casts a single target ability on someone.
In this phase we have total number of calculations per second (Tcs):
Tcs = Nb*Nb + Nn

Phase 2: Ballgroup rushes in spamming aoes and activating their aoe ultimates. Every ballgroup player affects multiple players from normal group and all players from ball group. Lets say that j percent of all players from the normal group are affected.
Tcs = Nb*(Nb + Nn*j) + Nn


2. 2. A normal group of Nn players fights a normal group of Nn players. AOEs may be used but if everyone is spread out then an aoe affects only one player and induces only additional database query to check which player was affected in comparison to activating a single target ability that already returns an id of affected player.
Tcs = Nn+Nn

Those calculations do not take into consideration proc sets that increase calculations per second. This means that a ball group effect on lag is even greater since they often use proc sets (Earthgore for example).

As you can see a ball group produces significantly more calculations per second.
Lets talk about some real numbers to visualize you this.
Lets say that there are 100 players in a normal group and 20 players in ballgroup. And during ballgroup charge they affect one tenth of all players from normal group per second.
In that case we get total number of calculations per second:
Tcs = 20*(20 + 100/10) + 100 = 700
In comparison if there are 100 players on two normal groups, the total number of calculations is:
Tcs = 100+100 = 200
Which is a lot less calculations with almost twice the number of all players.

Conclusions:
1. Ballgroups significantly increase lag compared to normal groups.
2. The bigger the ballgroup, the more lag we get. It has been numerously observed that a ballgroup and normal group large enough can even cause a crash.


How to deal with this?

• Increase server efficiency by a factor of five. That is impossible. Code optimization could decrease lag only a little. Server upgrade would also decrease lag only a little.
• Implement game mechanics that encourage players to spread out. Not only spread out on the map but also spread out on the battlefield. Implementing daedric artifacts won't be enough. This will only create one more big battle on the map, but players will still stack.
• Put a limit of affected players on every aoe ability. For example healing springs would affect only two players at once. This is unacceptable. And would require massive balance changes.
• Move calculations from server to client computers. This would reduce the lag by the time it takes a server to make all calculations but would also increase it, because each computer would have to report more to the server. What is unacceptable with this solution is that cheating would be extremely easy and almost undetectable.

Would like to throw out one more solution- before the gold farmer 'Fix' implementation the lag was almost non existent in this game that was designed for 100's of players at one place at one time. There are videos out there before the ZOS rush to stop farmers and gold sellers early on that is the genesis of all lag we currently see today. So, if code exists that was nearly lag free and they introduced 'fix' to protect their profits in the store (that is what they did) now that we have a fraction of that population- an additional solution would be to assess now if the lag we tolerate is the only way to protect those profits or potential profits they introduced it to protect...namely farmers and gold sellers- do they present the problem with tiny fraction of population now that they did in the beginning of the game? Question ZOS will have to answer, they have the original 'unfixed' version without lag.

What "Fix" are you talking about? (Also please use commas to make your sentences more understandable.)

The 'Fix' I am referring to was to stop gold farmers and bots in the game. Not quite sure how to make that any clearer, but, did, read, the, part about more, comma's, so you could, understand better, what was said. That help?

Soul_Demon

Edirt_seliv wrote: »

Soul_Demon wrote: »

InvictusApollo wrote: »

RealDolanTrump wrote: »

Savos_Saren wrote: »

Definitely. But the worst is trying to defend against 80+ people. That's when the disconnects come...

This! The day ZOS comes up with an idea to make people spread over the map is the day that a) lag will be gone and b) ball groups will disappear. Exactly in that order, noobs stacking in one place to get 6k tick for not doing anything are the problem in Cyrodiil. Ball groups contribute to lag, but not as much as faction stackers do. That's coming from somebody that does not play in a ball group.

We had a day like that. Even several. It was during the "siege bug". Ping was great. Game was fluid. Ball groups didn't exist because they were too afraid to stack.
We had epic battles. And even though there were dozens of players on the battlefield we still had great ping.
I think it was because they were spread out so their aoe abilities weren't affecting other players.

Experiences of so many players with ball groups inducing lag prove that spamming multiple aoe abilities on multiple players produces significant strain on servers which in effect causes lag.

I haven't seen ESO's source code but given your reports I can make an educated guess that the algorythm looks like follows:

1. Player casts some ability. If it is single target, go to step 3. If it is aoe go to step 2.
2. A query to the database is executed. Query returns a list of all players that are within aoe range.
3. System makes calculations on all affected players. If there was one target then we get only one calculation. If there were multiple targets then we get multiple calculations.

This means that when one player casts an aoe ability (heal for example) on n players, then we get n calculations. In ball groups however all players cast on average one aoe ability per second. This means that a ball group generates n^2 calculations per second. And this is assuming that players cast only aoe abilities that last for one second or put an effect that lasts more seconds but doesn't stay on ground. If players cast ground based aoe effects that remain on the ground we get even more calculations.
This number of calculations per second is vastly increased whenever a ball group rolls over another group. In that case the n^2 is multiplied by the number of players in opposing group.

Lets analyze two scenarios:
1. A ballgroup the size of Nb fights a normal group of Nn players. Where Nn = Nb*k.
Total player number: Tp = Nb + Nb*k = Nb*(1+k) = Nn + Nn/k = Nn*(1+1/k)
We have two phases during a fight:
Phase 1: Balllgroup charges their ultimates. During that time every player from the normal group either casts aoe ability on ball group, single target ability on anyone or aoe ability on just several players of normal group. Lets assume the most optimisitic choice (for the number of calculations) which is that he casts a single target ability on someone.
In this phase we have total number of calculations per second (Tcs):
Tcs = Nb*Nb + Nn

Phase 2: Ballgroup rushes in spamming aoes and activating their aoe ultimates. Every ballgroup player affects multiple players from normal group and all players from ball group. Lets say that j percent of all players from the normal group are affected.
Tcs = Nb*(Nb + Nn*j) + Nn


2. 2. A normal group of Nn players fights a normal group of Nn players. AOEs may be used but if everyone is spread out then an aoe affects only one player and induces only additional database query to check which player was affected in comparison to activating a single target ability that already returns an id of affected player.
Tcs = Nn+Nn

Those calculations do not take into consideration proc sets that increase calculations per second. This means that a ball group effect on lag is even greater since they often use proc sets (Earthgore for example).

As you can see a ball group produces significantly more calculations per second.
Lets talk about some real numbers to visualize you this.
Lets say that there are 100 players in a normal group and 20 players in ballgroup. And during ballgroup charge they affect one tenth of all players from normal group per second.
In that case we get total number of calculations per second:
Tcs = 20*(20 + 100/10) + 100 = 700
In comparison if there are 100 players on two normal groups, the total number of calculations is:
Tcs = 100+100 = 200
Which is a lot less calculations with almost twice the number of all players.

Conclusions:
1. Ballgroups significantly increase lag compared to normal groups.
2. The bigger the ballgroup, the more lag we get. It has been numerously observed that a ballgroup and normal group large enough can even cause a crash.


How to deal with this?

• Increase server efficiency by a factor of five. That is impossible. Code optimization could decrease lag only a little. Server upgrade would also decrease lag only a little.
• Implement game mechanics that encourage players to spread out. Not only spread out on the map but also spread out on the battlefield. Implementing daedric artifacts won't be enough. This will only create one more big battle on the map, but players will still stack.
• Put a limit of affected players on every aoe ability. For example healing springs would affect only two players at once. This is unacceptable. And would require massive balance changes.
• Move calculations from server to client computers. This would reduce the lag by the time it takes a server to make all calculations but would also increase it, because each computer would have to report more to the server. What is unacceptable with this solution is that cheating would be extremely easy and almost undetectable.

Would like to throw out one more solution- before the gold farmer 'Fix' implementation the lag was almost non existent in this game that was designed for 100's of players at one place at one time. There are videos out there before the ZOS rush to stop farmers and gold sellers early on that is the genesis of all lag we currently see today. So, if code exists that was nearly lag free and they introduced 'fix' to protect their profits in the store (that is what they did) now that we have a fraction of that population- an additional solution would be to assess now if the lag we tolerate is the only way to protect those profits or potential profits they introduced it to protect...namely farmers and gold sellers- do they present the problem with tiny fraction of population now that they did in the beginning of the game? Question ZOS will have to answer, they have the original 'unfixed' version without lag.

You also have to remember that back during the times you are talking about, ball groups (and pretty much everyone in cyrodil) were almost unanimously wearing warlock and seducer. There are hundreds more sets now with added calculations, there are more passives, more abilities etc.

Now don't get me wrong, I far from disagree with you regarding ball groups. By zyks 'the sky is falling' estimation of the subject one would be inclined to think that ball group v ball group fights would basically set fire to your pc. Which is the complete opposite to what you will experience out there. Gvgs are the best examples. I have been a part of many gvg fights in the middle of nowhere with other ball groups of 16-24+.

These fights were always the most fun, not only because it was more engaging than fighting mindless pugs, but because it was often the most crisp gameplay, lagfree compared to taking a populated keep etc.

So to answer the OPs question, when in a vacuum, no I do not get lag when ball groups are nearby. Now, understandably "in a vacuum" isn't exactly a pursuasive caveat, but the point is, if you just want to close your eyes and point your finger at ball groups and say "there's your lag fellas" then you'll wind up just looking as sallty as our good friend zyklosis here.

Agreed, there were precious few sets that were proc sets then (compared to now) but we still had plenty of them in game. I would concede that the number of calculations for the newer sets as well as the volume in general of them would have to impact the lag as well. But the playability of the game prior to that time vs after was evident to all playing. It was the beginning of the lag wars time.....20 min of no one taking any damage at all, no skills going through to suddenly ALL the damage going through all at one time. It has improved since then, but seems that is the point where they made a choice and it has been with us ever since in differing degrees impacting the gaming experience.

Xsorus

https://www.youtube.com/watch?v=BJiWSWA4aLQ

Hell it use to be even worse

https://www.youtube.com/watch?v=DBxXjYjgSIc

https://www.youtube.com/watch?v=_2BqTrX8zVk

zyk

Edirt_seliv wrote: »

Now don't get me wrong, I far from disagree with you regarding ball groups. By zyks 'the sky is falling' estimation of the subject one would be inclined to think that ball group v ball group fights would basically set fire to your pc. Which is the complete opposite to what you will experience out there. Gvgs are the best examples. I have been a part of many gvg fights in the middle of nowhere with other ball groups of 16-24+.

You need to work on your comprehension skills. At no point have I ever said ball groups are responsible for all the lag in Cyrodiil. Also, in case you haven't noticed, the sky fell in Cyrodiil years ago. The chicken littles have been replaced with masochists and pollyannas.

InvictusApollo wrote: »

What "Fix" are you talking about? (Also please use commas to make your sentences more understandable.)

This is a player theory that ZOS denies.

There were widespread complaints about severe lag weeks before the patch in question. That patch happened to introduce a brutal client fps bug which took weeks to fix.

ZOS has said that the reason the game was less laggy early on was because most players were very low level with bad gear. They have pointed to the performance on the non-vet/sub-50 servers as an example of this because when they were popular and poplocked nightly, the servers weren't stressed as much as the level 50 campaigns. Plus if you watch the early videos, most players are very immobile without a clue about what they're doing. In short, lag increased as players became more capable of stressing the server.

Plus it's clear security was not significantly improved then as memory hacking and bots continue to be an issue today. The only difference is that we don't have memory hacking bots anymore because no one bothers to do that -- but both still happen individually, obviously.

Dojohoda

Absolutely. Ground bugs and spin-to-win ball group sucks the life out of the server as well as the desire to play.

I just wonder, are their buttons broken? The rest of us use up to 10 skills and the occasional ultimate. Maybe they don't know how to play?

Ya, I have no respect for 1-2 button ball groups. Deal with it.