Campaign Performance

waitwhat

ZOS_BrianWheeler wrote: »

More players means more server calculations; that is 100% correct. However the key variable in all these scenarios is how much information is being calculated on a character by character basis depending on abilities being used, passives, armor sets, etc. This is why population is not as big a factor compared to what's being calculated on a character by character basis within that population. Let's take an example of a typical armor setup now a days.

A player wearing Viper, Velidreth, and Red Mountain doing a single heavy attack costs the server 3 times as much as a player doing Heavy attack without those sets because of calculating whether to proc those 3 sets or not. Even when a proc is on cooldown, the server needs to check per attack if the cooldown is done yet, which means every attack it checks whether it can fire or not based on either percentage, cooldown, or other situations. Factor in Champion Point passives, class passives, weapon passives and whatever temporary passive bonuses from potions, and you add to those calculations per attack/being attacked. In campaigns like Blackwater Blade and Azura, there are simply less things to calculate even when they have higher population than Trueflame.

So proc stacking is the primary cause of poor system performance in Cyrodiil.

@ZOS_BrianWheeler This is significant and intractable problem with ramifications across zones.

For performance's sake, how can the system instance zones based on population size if resource demand does not significantly relate to population size? The player in @ZOS_BrianWheeler 's example above costs the server three times as many resources only once they activate the condition checks for their proc sets, in that case by attacking. Until that player attacks and initiates the condition checks for their proc sets, however, the system doesn't know that the player will cost three times the resource amount as another player performing the same action. The sames goes for dueling in PvE zones: the system doesn't know prior to zone admittance if a given player is going to cost it multiplicatively more resources because it doesn't know if the player will perform the actions that do so.

Cooldowns on proc sets also won't help with performance, since each set's activating condition still requires a check against the cooldown. Monstrously, cooldowns actually seem to make performance worse, along with percentage chances, since cooldown checks and RNG procs are the two addtional calcuations to which @ZOS_BrianWheeler refers.

Of course, instancing won't work with the campaign mechanics of Cyrodiil, and since population caps will not resolve the issue, Cyrodiil hardware across all platforms needs a boost, and it needs to rapidly scale up and down in response to unpredictable traffic demands.

Here's some steps towards a solution:

1. NO NEW PROC SETS. The core design of proc sets--effects that have an RNG chance to activate after a cooldown on a given condition--appears to have created the present performance degradation. @ZOS_RichLambert , given the intractable difficulties proc sets impose upon resource management with their additional calculations, set design should pursue a different direction, perhaps one aimed at more constant benefits and trade-offs that more predictably drain resources.
2. NO DUELING IN CITIES. Put a 1km radius around the wayshrines in any city, within which dueling is not allowed. Honestly, this is safer for all the merchants in the Mournhold, Elden Root, and Wayrest markets anyway, and nobody needs to lag out of a guild trader mid-purchase.
3. MORE TOWNS TO CAPTURE IN CYRODIIL. @ZOS_RichLambert , @Wrobel we want to spread people out, but Vicious Death/Destro Ult spreading people off stack during a keep siege just doesn't help that much in terms of resource load and FPS. If you really want to spread people out, add more strategic objectives to Cyrodiil. Make Cheydinhal and Chorrol capturable and usable as respawn points so that groups are split up to manage them and capture them.


@ZOS_RichLambert , the goal should be to phase out proc sets gradually by introducing more desirable alternatives with more constant resource requirements. The other option is a rollback on proc sets in all but name, as the calculations at the core of their design are the cause of our present predicament.
@Wrobel , Much attention has been paid to the mechanics of individual abilities and items. It appears the system will benefit from mechanic adjustments on a larger scale that prevent large numbers of players from congregating in one place and performing actions that disproportionately drain server resources.

No one should expect those changes to happen in update 13, but they should be on PTS about two months after update 13 is released.

akredon_ESO

So... did no one have any for thought before they put these sets in to think that this would/could cause server performance issues? or was this a case of ah screw it we are just gonna put it in and worry about it later. I get you want to create engaging and diverse builds but at what cost ? While i appreciate the information that @ZOS_BrianWheeler provided. I think there should have been some for thought before something like this was decided.

gard

akredon_ESO wrote: »

So... did no one have any for thought before they put these sets in to think that this would/could cause server performance issues? or was this a case of ah screw it we are just gonna put it in and worry about it later. I get you want to create engaging and diverse builds but at what cost ? While i appreciate the information that @ZOS_BrianWheeler provided. I think there should have been some for thought before something like this was decided.

I'd bet a clown crate that somewhere at ZOS there was at least one developer saying, "Ermagerd this is gonna be bad!"

However, keep in mind that there was pretty bad lag before proc sets. And it hasn't gotten 3x worse. Has it?
I play mostly in Azura's, it doesn't feel 3x worse to me.

I think that there are probably several different issues which all fall under the category of "lag" (at least when discussed/reported by p;layers.)
For example, even in Azura's I find that my framerate {GTX980) will steadily decrease from 60+ to 15 over a long play session in Cyrodiil.
The framerate stays low even if I'm in a keep where there are few enemy players. If I quit the client and restart, it'll go back to normal and the cycle begins again.

The point is it feels laggy even though it really isn't a latency problem.

Katahdin

How about making another CP keeps campaign or switching Haderas to include keeps.

I have Haderas as my guest campaignbut I hardly go there. Many, including myself don't want to play there because it doesn't include keeps.

This might help spread people out a bit more. (?)

Armitas

ZOS_BrianWheeler wrote: »

More players means more server calculations; that is 100% correct. However the key variable in all these scenarios is how much information is being calculated on a character by character basis depending on abilities being used, passives, armor sets, etc. This is why population is not as big a factor compared to what's being calculated on a character by character basis within that population. Let's take an example of a typical armor setup now a days.

A player wearing Viper, Velidreth, and Red Mountain doing a single heavy attack costs the server 3 times as much as a player doing Heavy attack without those sets because of calculating whether to proc those 3 sets or not. Even when a proc is on cooldown, the server needs to check per attack if the cooldown is done yet, which means every attack it checks whether it can fire or not based on either percentage, cooldown, or other situations. Factor in Champion Point passives, class passives, weapon passives and whatever temporary passive bonuses from potions, and you add to those calculations per attack/being attacked. In campaigns like Blackwater Blade and Azura, there are simply less things to calculate even when they have higher population than Trueflame.

@ZOS_BrianWheeler

I have a question about this type of lag.

If a group of 20 EP and 20 DC are at a keep and all 20EP have as many proc sets sets as possible and the 20 DC have none will both groups experience the same lag or will one group experience less lag than the other group?

akredon_ESO

@Armitas I would think it would be for both. Because at the end of the day its the Server that is having to process all those calculations.

ZOS_BrianWheeler

@Armitas Both teams at that conflict, as well as other places in Cyrodiil, would feel the effects of that battle. The server processes all information from Cyrodiil together, so server performance is effected across the entire zone. However that performance doesn't bridge across zones so while there may be spikes in Cyrodiil, they would not effect Imperial City, Imperial Sewers, or any of the Cyrodiil caves.

ZOS_BrianWheeler

@Recremen In the next update we are doing some of exactly what you have suggested in putting frequently referenced abilities/requirements into a different server process.

Bashev

ZOS_BrianWheeler wrote: »

@Recremen In the next update we are doing some of exactly what you have suggested in putting frequently referenced abilities/requirements into a different server process.

In the next update you mean update 13, right?

ZOS_BrianWheeler

Correct Bashev.

Bashev

ZOS_BrianWheeler wrote: »

Correct Bashev.

Thanks @ZOS_BrianWheeler I have another question: when I am low on resources and I cannot cast a skill because I have no resources, the skill icon is greyed. But I noticed that in these situation it flashes every 50-100ms like it is checking weather I have the resources or not. Are these effects based on such calculations? If yes are these calculations are done by the server or by the client?

Armitas

ZOS_BrianWheeler wrote: »

@Armitas Both teams at that conflict, as well as other places in Cyrodiil, would feel the effects of that battle. The server processes all information from Cyrodiil together, so server performance is effected across the entire zone. However that performance doesn't bridge across zones so while there may be spikes in Cyrodiil, they would not effect Imperial City, Imperial Sewers, or any of the Cyrodiil caves.

Thanks Brian.

Recremen

ZOS_BrianWheeler wrote: »

@Recremen In the next update we are doing some of exactly what you have suggested in putting frequently referenced abilities/requirements into a different server process.

So cool! I love science.

SwaminoNowlino

ZOS_BrianWheeler wrote: »

@Recremen In the next update we are doing some of exactly what you have suggested in putting frequently referenced abilities/requirements into a different server process.

@ZOS_BrianWheeler,

Thanks for the update Brian. But as you mentioned earlier, regardless of all of the calculation stuff, there's still an issue of reaching a critical amount of players in one area. While deferring or segregating some calculations may hopefully help, isn't it more of a band-aid type fix? It also sounds like a pretty technical band-aid that could pretty easily break some things inadvertently.

For instance it looks like on the PTS build you've relegated CC's to this type of differential server process, and they all appear to be pretty significantly broken. It sounds like the fix you're talking about is akin to adding something like an express lane on a freeway. These help in certain situations, but they also can cause more accidents and calamity. It also doesn't address the bottleneck of translating information from the server to the client, which leads us back to the critical mass of players type issue.

I love that you're trying to think of technical ways to address this, but I'm worried this may miss the mark. You can keep putting someone down for open heart surgery to perform bypasses, but if the person is still cramming triple cheeseburgers down their gullet multiple times a day and clogging their arteries, does it really help? Would it not be more helpful to work towards changing the individual's behavior?

Create incentives for people to spread out.

As @BigES suggested in another thread, a creative disincentive to running in huge blobs is make individual keeps more important to the overall success of factions in the alliance war. Make factions need to be concerned about the status of all of their keeps, so they can't just zerg from one to another.

Another suggestion is to implement diminishing returns on healing. Basically, make healing less effective if you have a a ton of people around you, like greater than 12 for instance. This allows the change to be implemented easily, without impacting trials. You heal for a normal value until you have more than 12 people within your active radius, then decrease the effectiveness of healing done for each person over that threshold. This insures that people are more conscious of their surroundings and have less incentive to rely on herd protection. You can still run in large groups effectively if you want, you just have to build for it thus imposing a penalty on those larger groups.

Just a couple ideas. Hopefully battlegrounds are close and implemented properly, so the majority of the player base concerned with performance and enjoyable, competitive PVP will never have to enter Cyrodiil again. But I think you all could get really close to solving a lot of these overarching performance issues through some creative tweaks to the base game without having to devote a lot of time and resources.

Hurika

My main concern about breaking correlated tasks into separate processes is shared data. There's a big risk that the processes become out of sync causing odd things to happen. Especially if one process or thread gets bogged down there has to be a lot of deliberate effort to ensure other related process know and behave accordingly.

You don't want situations where you run non proc sets and kill someone but that process is so bogged down the opponent has procs firing on you because the proc process is running faster and doesn't know they are already dead. May not be exactly the case but it illustrates that concurrent processes and threads require much more thought and planning to avoid unintended consequences.

Recremen

SwaminoNowlino wrote: »

ZOS_BrianWheeler wrote: »

@Recremen In the next update we are doing some of exactly what you have suggested in putting frequently referenced abilities/requirements into a different server process.

@ZOS_BrianWheeler,

Thanks for the update Brian. But as you mentioned earlier, regardless of all of the calculation stuff, there's still an issue of reaching a critical amount of players in one area. While deferring or segregating some calculations may hopefully help, isn't it more of a band-aid type fix? It also sounds like a pretty technical band-aid that could pretty easily break some things inadvertently.

For instance it looks like on the PTS build you've relegated CC's to this type of differential server process, and they all appear to be pretty significantly broken. It sounds like the fix you're talking about is akin to adding something like an express lane on a freeway. These help in certain situations, but they also can cause more accidents and calamity. It also doesn't address the bottleneck of translating information from the server to the client, which leads us back to the critical mass of players type issue.

I love that you're trying to think of technical ways to address this, but I'm worried this may miss the mark. You can keep putting someone down for open heart surgery to perform bypasses, but if the person is still cramming triple cheeseburgers down their gullet multiple times a day and clogging their arteries, does it really help? Would it not be more helpful to work towards changing the individual's behavior?

Create incentives for people to spread out.

As @BigES suggested in another thread, a creative disincentive to running in huge blobs is make individual keeps more important to the overall success of factions in the alliance war. Make factions need to be concerned about the status of all of their keeps, so they can't just zerg from one to another.

Another suggestion is to implement diminishing returns on healing. Basically, make healing less effective if you have a a ton of people around you, like greater than 12 for instance. This allows the change to be implemented easily, without impacting trials. You heal for a normal value until you have more than 12 people within your active radius, then decrease the effectiveness of healing done for each person over that threshold. This insures that people are more conscious of their surroundings and have less incentive to rely on herd protection. You can still run in large groups effectively if you want, you just have to build for it thus imposing a penalty on those larger groups.

Just a couple ideas. Hopefully battlegrounds are close and implemented properly, so the majority of the player base concerned with performance and enjoyable, competitive PVP will never have to enter Cyrodiil again. But I think you all could get really close to solving a lot of these overarching performance issues through some creative tweaks to the base game without having to devote a lot of time and resources.

They've been creating incentives for people to spread out. Town capture, new resource AP rewards, and the slow and steady realization by Trueflame players that resource sneaking is becoming an increasingly important part of the campaign victory condition.

As far as "express lanes" and "freeways" go, I'd trust the professional engineers to know what they're doing on that. The change is doubtless coming from a steady process of code refinement and based on verifiable performance metrics that we players can only guess at, but which are super obvious and informative to the actual engineers. Also, sometimes a change in a system is objectively good by every measure with no real downside. If you're familiar with programming at all, look at the difference in performance for single-object referencing between a Map and an Array data structure. The Map requires more memory to work properly, but memory is, as a rule, way less expensive than processing time, and so Maps perform better, period, in all cases for that specific task.

The only thing I'm worried about is what appears to be the steadily declining ability to make a lot of AP in a group setting. I hope that as these performance changes and incentives continue to get refined, they don't forget about the arduous AP grind to Grand Overlord!

BigES

ZOS_BrianWheeler wrote: »

More players means more server calculations; that is 100% correct. However the key variable in all these scenarios is how much information is being calculated on a character by character basis depending on abilities being used, passives, armor sets, etc. This is why population is not as big a factor compared to what's being calculated on a character by character basis within that population. Let's take an example of a typical armor setup now a days.

A player wearing Viper, Velidreth, and Red Mountain doing a single heavy attack costs the server 3 times as much as a player doing Heavy attack without those sets because of calculating whether to proc those 3 sets or not. Even when a proc is on cooldown, the server needs to check per attack if the cooldown is done yet, which means every attack it checks whether it can fire or not based on either percentage, cooldown, or other situations. Factor in Champion Point passives, class passives, weapon passives and whatever temporary passive bonuses from potions, and you add to those calculations per attack/being attacked. In campaigns like Blackwater Blade and Azura, there are simply less things to calculate even when they have higher population than Trueflame.

Brian -

Isn't a passive skill on a skill tree worth the same amount of lag as a single proc set (in terms of server performance)? So if I have 4 proc sets on, does it really matter if I have 30 skill passives the server needs to do a calculation for?

I'm just worried people are going to misinterpret your post as "ZOMG PROC SETS CAUSE ALL THE LAG!" when that doesn't appear to be what's going on here.

frozywozy

waitwhat wrote: »

ZOS_BrianWheeler wrote: »

More players means more server calculations; that is 100% correct. However the key variable in all these scenarios is how much information is being calculated on a character by character basis depending on abilities being used, passives, armor sets, etc. This is why population is not as big a factor compared to what's being calculated on a character by character basis within that population. Let's take an example of a typical armor setup now a days.

A player wearing Viper, Velidreth, and Red Mountain doing a single heavy attack costs the server 3 times as much as a player doing Heavy attack without those sets because of calculating whether to proc those 3 sets or not. Even when a proc is on cooldown, the server needs to check per attack if the cooldown is done yet, which means every attack it checks whether it can fire or not based on either percentage, cooldown, or other situations. Factor in Champion Point passives, class passives, weapon passives and whatever temporary passive bonuses from potions, and you add to those calculations per attack/being attacked. In campaigns like Blackwater Blade and Azura, there are simply less things to calculate even when they have higher population than Trueflame.

Cooldowns on proc sets also won't help with performance, since each set's activating condition still requires a check against the cooldown. Monstrously, cooldowns actually seem to make performance worse, along with percentage chances, since cooldown checks and RNG procs are the two addtional calcuations to which @ZOS_BrianWheeler refers.

You forgot the part where having a global cooldown on damaging proc sets will strongly discourage people from stacking them, to not say force them to use only one which would result in a significant performance change.

@ZOS_BrianWheeler - could you please elaborate on why a global cooldown couldn't be possible? The question has been asked several times but I haven't been able to catch any explanation yet on why it is not on the table already.

A global cooldown need to be put in place or damaging proc sets need to disappear.

Proc sets shouldn't be able to proc each others.

Exploits that allow people to ignore poison cooldowns need to be fixed or poisons need to be removed.

frozywozy

gard wrote: »

akredon_ESO wrote: »

So... did no one have any for thought before they put these sets in to think that this would/could cause server performance issues? or was this a case of ah screw it we are just gonna put it in and worry about it later. I get you want to create engaging and diverse builds but at what cost ? While i appreciate the information that @ZOS_BrianWheeler provided. I think there should have been some for thought before something like this was decided.

For example, even in Azura's I find that my framerate {GTX980) will steadily decrease from 60+ to 15 over a long play session in Cyrodiil.
The framerate stays low even if I'm in a keep where there are few enemy players. If I quit the client and restart, it'll go back to normal and the cycle begins again.

The point is it feels laggy even though it really isn't a latency problem.

Addons poorly managed/updated do that. When I used to play World of Warcraft, we had an addon that would keep track of data usage per addon in kbps/mpbs and that was super useful. I wish someone would make one like that for Teso.

SwaminoNowlino

Recremen wrote: »

SwaminoNowlino wrote: »

ZOS_BrianWheeler wrote: »

@Recremen In the next update we are doing some of exactly what you have suggested in putting frequently referenced abilities/requirements into a different server process.

@ZOS_BrianWheeler,

Thanks for the update Brian. But as you mentioned earlier, regardless of all of the calculation stuff, there's still an issue of reaching a critical amount of players in one area. While deferring or segregating some calculations may hopefully help, isn't it more of a band-aid type fix? It also sounds like a pretty technical band-aid that could pretty easily break some things inadvertently.

For instance it looks like on the PTS build you've relegated CC's to this type of differential server process, and they all appear to be pretty significantly broken. It sounds like the fix you're talking about is akin to adding something like an express lane on a freeway. These help in certain situations, but they also can cause more accidents and calamity. It also doesn't address the bottleneck of translating information from the server to the client, which leads us back to the critical mass of players type issue.

I love that you're trying to think of technical ways to address this, but I'm worried this may miss the mark. You can keep putting someone down for open heart surgery to perform bypasses, but if the person is still cramming triple cheeseburgers down their gullet multiple times a day and clogging their arteries, does it really help? Would it not be more helpful to work towards changing the individual's behavior?

Create incentives for people to spread out.

As @BigES suggested in another thread, a creative disincentive to running in huge blobs is make individual keeps more important to the overall success of factions in the alliance war. Make factions need to be concerned about the status of all of their keeps, so they can't just zerg from one to another.

Another suggestion is to implement diminishing returns on healing. Basically, make healing less effective if you have a a ton of people around you, like greater than 12 for instance. This allows the change to be implemented easily, without impacting trials. You heal for a normal value until you have more than 12 people within your active radius, then decrease the effectiveness of healing done for each person over that threshold. This insures that people are more conscious of their surroundings and have less incentive to rely on herd protection. You can still run in large groups effectively if you want, you just have to build for it thus imposing a penalty on those larger groups.

Just a couple ideas. Hopefully battlegrounds are close and implemented properly, so the majority of the player base concerned with performance and enjoyable, competitive PVP will never have to enter Cyrodiil again. But I think you all could get really close to solving a lot of these overarching performance issues through some creative tweaks to the base game without having to devote a lot of time and resources.

They've been creating incentives for people to spread out. Town capture, new resource AP rewards, and the slow and steady realization by Trueflame players that resource sneaking is becoming an increasingly important part of the campaign victory condition.

As far as "express lanes" and "freeways" go, I'd trust the professional engineers to know what they're doing on that. The change is doubtless coming from a steady process of code refinement and based on verifiable performance metrics that we players can only guess at, but which are super obvious and informative to the actual engineers. Also, sometimes a change in a system is objectively good by every measure with no real downside. If you're familiar with programming at all, look at the difference in performance for single-object referencing between a Map and an Array data structure. The Map requires more memory to work properly, but memory is, as a rule, way less expensive than processing time, and so Maps perform better, period, in all cases for that specific task.

The only thing I'm worried about is what appears to be the steadily declining ability to make a lot of AP in a group setting. I hope that as these performance changes and incentives continue to get refined, they don't forget about the arduous AP grind to Grand Overlord!

You're asserting that the increased AP gains will spread people out? Let's look at the facts as they are now. It is not uncommon to see a zerg of literally 100 yellows in Scourge on Xbox One. I watch streamers on PC, and quite often see zergs of similar size. Zergs are currently, and have been for quite some time, incredibly prevalent. The nature of their massive group means that they get virtually no AP from anything they do. They may split a defensive tick, but they'll be splitting it 100 ways. Then you look at the Campaign forums where you see these folks say that they do not care about AP, which is demonstrated by their in game actions. They care about winning, by any means necessary. So if they have to call 100 to dislodge a group of 8 from an outpost, they will. No concern is given to quality PVP or AP or fair fights.

The change being proposed is to provide higher base values of AP. Is this going to change that behavior? Heck no. If anything I am willing to make the bold prediction that it will make things far worse. It'll be even more worthwhile for them to take those 100 people to get their resource back.

When presented with the facts of how people play now, and what their priorities are, the proposed changes are not going to do anything positive. So the suggestion being made is make winning, not AP, more predicated on having to spread out. The dominant faction with emp in a campaign currently doesn't give two bleeps about one of those exterior keeps like Drakelowe. Sure they'll run over there if nothing is going on, but they're more than content to just hold the emp keeps all night. That should be punished. Reducing the effectiveness of healing by having giant groups and increasing the importance of every keep and resource strategically would mean these groups that care about "winning" will have to break apart.

As for the statement about trusting the engineers, I find that funny because we currently have a PTS where its quite obvious they have broken stuff in their effort to put some processes in a separate lane. Between all the crashing and performance issues on PTS, and the fact that several abilities don't work and most CC's are currently broken, its quite obvious what trusting the engineers gets you. I work in an engineering organization. Our engineers are great at coming up with band-aids or fixing things, but horrendously bad at the critical thinking/big picture part. Its not even their fault, its just the mindset and thought process that makes them effective at their job. Engineers tend to think differently than planners. You present them with a problem, they fix the immediate broken thing. They don't look at the overall process to see if 1,000 little things went into stressing that piece causing it to break.

ManDraKE

Using the proc sets as an example to explain server calculations was a bad idea. This is why:

waitwhat wrote: »

So proc stacking is the primary cause of poor system performance in Cyrodiil.

/facepalm

Karivaa

@ZOS_BrianWheeler , I have been playing on Xbox na and been in blackwater blade because I wanted to play lag and dashboard free. I got dashboarded at least 4 times yesterday! The destro staff I use causes a lag spike every time I get a kill.

Dr_Resilient

So why a empty server(Haderus) early in the morning, with low population, no fights going on on the map, except the one i was, 2v3, and yet we experience an absurd lag? So, basically the server can't handle some simple small fights?

Kirsch9

Been wondering:

If two large teams, identical in numbers and gear setups fight and the only difference is all the players on one team have a ping rate of 100 (normal) and the other team's is at 250 - 500+ (high)... would that cause more lag than if both team's had ping rates of 100? (Sorry this is so long but thank you for reading and answering the question part)

From what what I've read about ping rates and lag in pvp I understand that in small scale fights one player's ping isn't going to affect another's. It doesn't make sense to me though how it wouldn't in these large scale fights where so many players are affecting each other with abilities, passives, manual/rng proc sets, etc. I mean, in a 1v1 situation I can see where the server nor the other player is going to have an issue lagging because one person has a high ping rate. However, in a highly congested area (50 to 100 players) with all of these things happening at once is already taxing on the server and causes it to lag a bit even in ideal settings due to all the calculations it has to deal with and then transferring this information amongst all the players. Now, when half of these people have ping rates from 250 to 500+ (and I'm talking people that have these pings even when standing alone somewhere, staring at the floor), it would seem to me like that would cause more stress on the server thus causing it to lag even more. Since it's now trying to do all these mitigation, buff/debuff, possible proc, etc calculations plus waiting between a quarter to a half a second or more for the information to transfer, then be calculated and transferred yet again to the affected player(s) which if whatever it was gets blocked, reflected, causes a proc or is synergised with, that's even more calculations and wait to finish transferring time and all of this just causes a snowball effect until we all have 999 ping rates.

I don't know, I could be wrong but I've even noticed a performance difference between helping friends. One from the U.S. (where I'm at) with both of us having a good ping rate (100-120), and the other one from Sri Lanka with him having a not so great ping rate that seemed to make mine go up to about 120-175 (pre-grind mind you). I helped them both with grinding out some CP in Craglorn and in the same place even, only on separate occasions. When the guy from the U.S. and I were doing the "you go left I go right and we'll meet back here and AoE/Ultimate bomb them" pulls, everything was fine, maybe a slight frame stutter here and there but other than that everything was fine, time and time again. Now, when the other guy from Sri Lanka and I did the same exact thing... it was the absolute opposite! Had to use shields and potions to stay alive along with having to use our ultimates one after the other because when we'd fire them off at the same time, oh man, it would create such a huge lag spike that we'd get to see about 3 separate frames and then we'd just be laying on the floor dead lol.

themdogesbite

6k base value for the keeps is actually so much ap togheter with the offensive ticks when it gets capped, i can definetly see this beeing abused.

Yuke

ZOS_BrianWheeler wrote: »

@Recremen In the next update we are doing some of exactly what you have suggested in putting frequently referenced abilities/requirements into a different server process.

Sorry to bring this up, but the servers still cant handle the final keep fight to dethrone an emperor.

At this point (after two years) its safe to say that skills like healing springs or magicka detonation need a drastic change in mechanics, so ballgroups cant spam and stack them anymore while moving around

NBrookus

This past week performance has been very poor. Lag freezes, FPS drops, input lag, long load screens, crashes.
Then of course the invisible people plus unbreakable fear has now been joined by the unbreakable encase spam.

It has been a frustrating week.

On the plus side, I get out of combat quickly and the templar horse disease is greatly diminished.

frozywozy

NBrookus wrote: »

This past week performance has been very poor. Lag freezes, FPS drops, input lag, long load screens, crashes.
Then of course the invisible people plus unbreakable fear has now been joined by the unbreakable encase spam.

It has been a frustrating week.

On the plus side, I get out of combat quickly and the templar horse disease is greatly diminished.

I support what @NBrookus just said. The screen freezes are getting worse. I can confirm that some specific players get that at the exact same time, sometimes several times in a row. Those freezes can last from 0.5 second to 2-3seconds.

There is also a memory leak happening. You have to entirely restart your client to get ride of it. Sometimes it can come back not more than 10minutes after restarting. I get a 10-15fps permanent fps drop with alot of studdering.

@ZOS_BrianWheeler

Zaldan

since Homestead I've been getting odd bugs with abilities not casting or casting but only playing split second of their animation, strangely doesn't seem to be linked to lag.

Rapid Regeneration - constantly refuses to cast
Honor the Dead - casts but animation doesn't play, charcter just twitches
Mages Wrath - constantly refuses to cast
Burning siege weapons - animation only plays for a sec then stops and siege remains intact
Crystal Fragments - animation plays in full but ability doesn't cast, one crystal fragment appears briefly next to character

Oxalias

Zaldan wrote: »

since Homestead I've been getting odd bugs with abilities not casting or casting but only playing split second of their animation, strangely doesn't seem to be linked to lag.

Rapid Regeneration - constantly refuses to cast
Honor the Dead - casts but animation doesn't play, charcter just twitches
Mages Wrath - constantly refuses to cast
Burning siege weapons - animation only plays for a sec then stops and siege remains intact
Crystal Fragments - animation plays in full but ability doesn't cast, one crystal fragment appears briefly next to character

Yes all of the above plus + healing ward casting without an animation. Also streak mixed with other factors freezing your character.