Update on Cyrodiil Performance & Upcoming AOE Tests

Elsonso

relentless_turnip wrote: »

Elsonso wrote: »

relentless_turnip wrote: »

-Reducing group sizes to 12(reduces radius check)
-Healing to only effect group members(reduces radius check)
-Only one type of each heal on a group member i.e. recasting rapids would refresh the effect rather than stacking it.

I dunno. Honestly, this sounds like the same band-aid solutions that got us to this point in the first place. Ultimately, this sort of tweaking is the same thing that got us here. I think that this will have us back here talking about performance before too long.

Do you feel it is anymore of a band aid fix than putting a cooldown on 60% of skills in the game?

I'm not sure, but I know which makes the gameplay feel worse.

ZOS is being more technical and aggressive in their proposed tests. Both are band-aids, but the ZOS method also includes some disincentive, and that is really where the focus should be. ZOS is approaching this with a little more punishment than I think they need to do, but this is just a test, not a solution, and these tests should help prove the direction that the need to go.

ZOS obviously feels that the solution lies somewhere along the lines of getting players to reduce how often they reuse AoE. The main question seems to be determining the level where AoE use is not a problem, whether that can be single player or needs to be across the entire group, and whether it should be a solution that involves timing limits or cost limits.

At this time, I don't expect that any of these tests are potential solutions. I think they are just gathering facts.

Sugaroverdose

jcmlongb16_ESO wrote: »

I think a lot of the lag etc is due to packet loss and high latency.

For example, 2am this morning BST, on XBox EU there were a total of approx 14 DC and 30 AD, zero bars showing on all alliances, the mini load screens were still triggering, even withing southern high rock. In a 2v2 it took 3-5 button presses for leap to go off.

Any skill that requires a 'lock on' say heavy attack with a lightning staff or toppling charge wont go off without a light attack first, the only ability that consistently land are AOE, if this problem was fixed there would be less AOE's used.

One thing I find very strange is when something is changed to see if it fixes an issue when it doesn't the change is left in place, this is a very odd debugging process as you will never know if the subsequent change would have been effective if it wasn't for a previous change being left in.

High packet loss and high latency is a consequences of high server stress, not a reason.

SneaK

What a joke

Sugaroverdose

Czekoludek wrote: »

@ZOS_RichLambert Please, someone from ZOS should watch @FENGRUSH response. Great vid with so many valid points and propositions that will definietely hurt game less then what we see in OP.

https://www.youtube.com/watch?v=nO-NLBjSO1U

I thought about these proposed tests and think that Zeni take the wrong approach here, looking at wrong direction.

"Nerf templars ability to remove negative effects" absolutely unbiased decision to fix lag in cyrodiil, well part of his point have reason it's even less radical than my counter-zerg ideas, but still, ZoS will hear only about templar nerf

npuk

Sugaroverdose wrote: »

High packet loss and high latency is a consequences of high server stress, not a reason.

You can have an under-worked server and but a poor network infrastructure and that would also cause it, I use BT as my isp and I don't get anywhere near the disconnections that my friend on Virgin get. There was also noticeable increase here in the UK when we went into lock down and people started working from home.

I'm not disputing that the servers may need an upgrade, but that wont help if the data centre is letting the servers down.

relentless_turnip

Elsonso wrote: »

relentless_turnip wrote: »

Elsonso wrote: »

relentless_turnip wrote: »

-Reducing group sizes to 12(reduces radius check)
-Healing to only effect group members(reduces radius check)
-Only one type of each heal on a group member i.e. recasting rapids would refresh the effect rather than stacking it.

I dunno. Honestly, this sounds like the same band-aid solutions that got us to this point in the first place. Ultimately, this sort of tweaking is the same thing that got us here. I think that this will have us back here talking about performance before too long.

Do you feel it is anymore of a band aid fix than putting a cooldown on 60% of skills in the game?

I'm not sure, but I know which makes the gameplay feel worse.

ZOS is being more technical and aggressive in their proposed tests. Both are band-aids, but the ZOS method also includes some disincentive, and that is really where the focus should be. ZOS is approaching this with a little more punishment than I think they need to do, but this is just a test, not a solution, and these tests should help prove the direction that the need to go.

ZOS obviously feels that the solution lies somewhere along the lines of getting players to reduce how often they reuse AoE. The main question seems to be determining the level where AoE use is not a problem, whether that can be single player or needs to be across the entire group, and whether it should be a solution that involves timing limits or cost limits.

At this time, I don't expect that any of these tests are potential solutions. I think they are just gathering facts.

I agree with you, I believe they are just tests to gather data. That data will help them come up with an effective solution hopefully.

I would much prefer they reduced the amount of area checks, rather than reducing the amount of area effects that can be used. With suggestions like healing not stacking, not being used outside a group and reducing group size. I'm sure they could also put an aoe cap on all abilities and siege, this too would reduce the calculations caused by aoe abilities.

This will change the way we play in cyrodill, but won't ruin combat in cyrodill or anywhere else.

peacenote

ZOS_RichLambert wrote: »

Over the years, player power has grown considerably. With the addition of the Champion System, various armor/weapons sets, and changes to abilities, we have reached the stage where players - with the right build - can cast near-infinite numbers of abilities.

At launch, Cyrodiil’s processes were able to keep up with the number of AOEs cast, because most players couldn’t cast that many of them: they ran out of Magicka or Stamina, so they had to use AOEs judiciously. Over time, as player knowledge grew and regen builds grew in power, more players could cast more and more AOEs before running out of resources.

Premise: Power/resource creep is causing the performance issue.

Wouldn't the logical conclusion be to temper the power/resource creep in Cyro, via Battle Spirit, as a first test? If the problem is unlimited resources, fix the unlimited resources?

Adjusting resource regeneration across the board or for the specific abilities/sets that are causing it to over-perform in PvP seems like a better solution that completely gutting a well-beloved combat system.

technohic

jcmlongb16_ESO wrote: »

Sugaroverdose wrote: »

High packet loss and high latency is a consequences of high server stress, not a reason.

You can have an under-worked server and but a poor network infrastructure and that would also cause it, I use BT as my isp and I don't get anywhere near the disconnections that my friend on Virgin get. There was also noticeable increase here in the UK when we went into lock down and people started working from home.

I'm not disputing that the servers may need an upgrade, but that wont help if the data centre is letting the servers down.

Seeing as they are targeting AOE impact on servers. I'm assuming they are seeing their actual resources being consumed on their server. I dont think that means you just slap more hardware and it just takes care of itself. There has to be a capability to pool them together as needed where needed

React

It's tough to believe that this was even proposed as a viable solution to the lag.

When you see the fact that defensive and offensive abilities share the same CD, it is very clear that the developers who proposed this DO NOT care about how it affects combat. Not even a developer with minimal time on ESO should be able to look at that and say "Yep, this is a good solution to the problem". From the PTS just now, a player posted a picture on a templar where his bar was Jabs, rapid regen, channeled focus, ritual, breath of life, and crescent sweep. He cast his armor buff, and his entire bar of abilities became unusable.

This is an MMO lite. You do not have 20-50 abilities available to use. Cooldowns WILL NOT WORK for this game, and implementing them will fundamentally DESTROY combat, not just change it.

There have been so many good suggestions from players that you could implement where the performance WOULD improve without altering base combat mechanics. You could make healing go to group members only. Cap healing to 6 players maximum (abilities like purge too). REMOVE CP FROM CYRODIIL ENTIRELY. I don't know why CP wasn't identified as the primary issue here - CP is what enables the spamming of abilities more than anything, and the calculations related to CP are exponentially higher than the ones that occur in the same scenario within a NO-CP environment.

No matter what, this change will show an increase in performance during the test. If zenimax sees this and decides to follow through with it, it will be the final "F*** you" to any combat oriented PVP player left in the game. You are quite literally spitting in our faces with this.

Muizer

IMHO the game should have been much more single target based from the beginning, especially in PvP so in that respect I'd welcome whatever changes would move us in that direction.

nml

Czekoludek wrote: »

@ZOS_RichLambert Please, someone from ZOS should watch @FENGRUSH response. Great vid with so many valid points and propositions that will definietely hurt game less then what we see in OP.

https://www.youtube.com/watch?v=nO-NLBjSO1U

I thought about these proposed tests and think that Zeni take the wrong approach here, looking at wrong direction.

No, I watched the video - Fengrush is clearly speaking about what he wants from the game that suits his playstyle not what is best for the game overall.

Wanting heals to only work within your group is absurd, this is Alliance vs. Alliance combat so those on the same alliance should be able to throw out a heal to somebody near them even if they don't know each other.

Cronopoly

ESO Community: But no one wanted to believe. Believe they even existed. And when the truth finally dawns, it dawns in fire.
But, there's one they fear. In their tongue, he's Dovahkiin: Dragonborn!

ESO Devs: "Foos Ro Dah!!!!!" (Sensible combat gets blasted away...)

😁😅😙 I couldn't help it lol

Tammany

nml wrote: »

No, I watched the video - Fengrush is clearly speaking about what he wants from the game that suits his playstyle not what is best for the game overall.

Fenrush bias is maximum level, no point to listen his hours long preaches.

MincVinyl

There are plenty of other mechanical layers of the game that zos could look into....(imo these should be looked into before gutting aoes but hey)

• Cp
• Procs
• Cross healing/buffing/set proccing (many ways to implement, best: 1-4players can do outside of group 5-24 players can only interact with group)
• Over time effects stacking (previously how the game worked, less calculations, more skill/class/morph diversity,
• The hammer and destroyable gates (Leads to faction stacks)
• Keep ticks being the go to ap farm (resource incentives could spread out playes during seiges)
• Mountspeed being too fast (people bypassing field fights just to get to the ball seige at keeps)
• Abilities with too many mechanics (think of how complicated dizzy swing has become since it was a long cast hard hit knockup)
• Catapult seige and persistent ground effects
• The anticheat system (cant see zos removing this in the long run, but if they could disable it for a week it would be interesting to see what happens)

I would be perfectly fine doing more live tests every few months or whenever pts cycles happen

Cronopoly

The temptation to NGE the game 's combat has returned... SONY(SOE) did it. What's the worst that could happen? 🙄

Pot, Frog, Boil, Jump....sigh

W0lf_z13

bigdavid11b16_ESO wrote: »

If anybody remembers. the day that nobody could group. is the day that cyrodiil had zero lag. Maybe you all should have a little ponder as to why. The reasoning is that there were no people spamming abilities that were linked to group utility
example : PURGE

Why can we not just start at once place. Group size. Lower it to a max of 12 and go from there.

completely agree with this... from what I have seen, the main lag comes when there is a ball group around, mainly due to the group based heals being spammed and the server having to calculate all the healing to designated targets on said group... i get SOME lag while fighting a zerg... but not nearly as much as when the ball groups are around

Neoakropolis

Thank god the Alpha comes out for Ashes of Creation soon.. Turn based ESO? No thanks.

Sugaroverdose

nml wrote: »

Czekoludek wrote: »

@ZOS_RichLambert Please, someone from ZOS should watch @FENGRUSH response. Great vid with so many valid points and propositions that will definietely hurt game less then what we see in OP.

https://www.youtube.com/watch?v=nO-NLBjSO1U

I thought about these proposed tests and think that Zeni take the wrong approach here, looking at wrong direction.

No, I watched the video - Fengrush is clearly speaking about what he wants from the game that suits his playstyle not what is best for the game overall.

Wanting heals to only work within your group is absurd, this is Alliance vs. Alliance combat so those on the same alliance should be able to throw out a heal to somebody near them even if they don't know each other.

Solo heal spammers does nothing good to alliance vs alliance battle btw

TineaCruris

Dusk_Coven wrote: »

Put PvP on its own server running it's own ruleset so it won't keep messing up what's fun in PvE.
If this goes through on PvE you will probably have to rebalance all the endgame PvE achievements or have a cutoff where the "original" achievement can no longer be attained because of a drastically different ruleset.

I'm still trying to figure out why they can't just do what ever they did to make the MYM event work pretty smoothly, but do it all the time.

What did they do that made it better? And why can't they be doing that all the time?

Elsonso

.

TineaCruris wrote: »

Dusk_Coven wrote: »

Put PvP on its own server running it's own ruleset so it won't keep messing up what's fun in PvE.
If this goes through on PvE you will probably have to rebalance all the endgame PvE achievements or have a cutoff where the "original" achievement can no longer be attained because of a drastically different ruleset.

I'm still trying to figure out why they can't just do what ever they did to make the MYM event work pretty smoothly, but do it all the time.

What did they do that made it better? And why can't they be doing that all the time?

I have some thoughts on that, given that @ZOS_RichLambert has not added this to his discussion about Cyrodiil performance. ¯\_(ツ)_/¯ I was in one of the new Campaigns, but I stayed away from everyone while I was in there.

1. They radically dropped the population cap and fewer people requires fewer computations. The performance improvement was smoke and mirrors. I don't really think this is the case, but I dunno.
2. They changed an internal setting that they don't want to leave set that way. If we knew what that was, we might not want it, either. For example, maybe they simply stopped processing all of the AoE attacks and simply ignored them. Maybe they made it so that AoE only applied to 2 people, instead of 6. That sort of thing.

Beyond that, I am at a loss.

FrankonPC

I am all for doing tests to figure out what is causing server strain and lag in pvp. I personally believe gear sets and aoe checks on spammed healing abilities are extremely problematic because of how they function.

Earthgore on live is running checks on everyone in the area to see if they're at low health before proccing, same with Troll King. You also have other conditional proc sets like bogdan and symphony of blades. These sets looking for conditions to drop off of cool down in a big area is probably creating a lot of stress, if the spamming of these aoe abilities also looks to be the culprit.

I do not think a 3 second global cool down is a fix though, and I think it is problematic to the tests being proposed. Every magicka class the way they are currently constructed will struggle vs any stamina class 1 v 1 with a 3 second global cool down on aoe check abilities. If you catch a burst that takes half your health bar, your casting of hots and burst heals EVERY 3 seconds is a much longer time than it takes the average class with a spammable like surprise attack or dswing. you can get off 2 dswings in the time it takes to do one heal. You will just die once you attempt to recover, and it will be frustrating to pvp, even with the double AP incentive.

I think some adjustments need to be made first to class kits before you attempt a test like this because as it is, even on live people do not want to engage in fights where their characters do not function. I don't think a 3 second aoe cool down is a good option regardless so unless you're just doing it as a baseline test to see what the server looks like, I'd scrap it entirely. If you are using it as a baseline to see how the server functions, please let the community know so that we can fully embrace the poor pvp experience to give the information you need to make the game better.

I think if we were all told a 3 second GCD isn't an option for a fix, but just is being used as a benchmark to compare data we'd all be a lot more relieved about the direction of this game.

silver1surfer69

jcmlongb16_ESO wrote: »

I think a lot of the lag etc is due to packet loss and high latency.

For example, 2am this morning BST, on XBox EU there were a total of approx 14 DC and 30 AD, zero bars showing on all alliances, the mini load screens were still triggering, even withing southern high rock. In a 2v2 it took 3-5 button presses for leap to go off.

Any skill that requires a 'lock on' say heavy attack with a lightning staff or toppling charge wont go off without a light attack first, the only ability that consistently land are AOE, if this problem was fixed there would be less AOE's used.

One thing I find very strange is when something is changed to see if it fixes an issue when it doesn't the change is left in place, this is a very odd debugging process as you will never know if the subsequent change would have been effective if it wasn't for a previous change being left in.

Yeah i had in the night from sunday to monday lag and delayed input problem at 5am cest dueling in alrikr (PC EU). This was more than in IC during the day.

Neloth

Paske wrote: »

CTO: It is offical, AOE-s are killing the performance. We need stronger hardware.

CFO: Hold on, let me crunch some numbers .... Yep just as I thought. It will cost money. No can do ...

Third guy: But what if we simply limp the combat system. Like with a 3 seconds cooldown. I mean, they will hardly notice.

The whole room: GENIUS !

Blizzard hearing of this: GENIUS !

Players: Why do you hate your player base. Is the money we are giving you somehow offensive and you decide not to invest this offensive money into hardware infrastructure.

You clearly don't understand basic principles of HPC scalability.

On PC NA we frequently have 5-6 seconds lag spikes. This basically means that server stops responding your client for 5+ seconds, since it is busy performing some other calculations (for example processing large-scale battle happening over a distant keep). Most probably you prefer to have ~500ms latency at worst, which means that servers has to perform its calculations ~10 time faster to process the input of that large-scale battle just in time to respond your client.

Now imagine they have linear scalability of their algorithms (however its a very rare case). To provide 10 times faster calculations without changing the code, they need 10 times faster hardware.

Surprise, no company will expand their hardware infrastructure in 10 times just to have better performance in some zone. And that's clearly not because they are greedy/silly, it's just a terrible solution. What they really need to do is either:

1) put most calculations back to the client side (which they will never do, since forum warriors here cry "cheaters" all the time, and they decided not to invest into anti-cheating software).
2) significantly reduce computational complexity of server calculations, and removing AoEs from PvP is a perfect way to do it (not from the game/class design point of view though).

Kamchuk

This sounds like a good test. I have to admit that every time you guys [ZOS] even mention “Templars” in your update notes I cringe – simply because I know another Nerf is coming. But I’m “on board” with this testing if it will improve performance. I would suggest that during that week, you suspend all but one Cryo campaign, increase the population caps on that one campaign so that you can get some meaningful and good data. Let us know if you need anything specific from us to help – such as a massive fight at one particular Keep. We all want you to succeed because this is an Awesome game.

coletas

with that analysis finally it will be turn based XDDDD

relentless_turnip

nml wrote: »

Czekoludek wrote: »

@ZOS_RichLambert Please, someone from ZOS should watch @FENGRUSH response. Great vid with so many valid points and propositions that will definietely hurt game less then what we see in OP.

https://www.youtube.com/watch?v=nO-NLBjSO1U

I thought about these proposed tests and think that Zeni take the wrong approach here, looking at wrong direction.

No, I watched the video - Fengrush is clearly speaking about what he wants from the game that suits his playstyle not what is best for the game overall.

Wanting heals to only work within your group is absurd, this is Alliance vs. Alliance combat so those on the same alliance should be able to throw out a heal to somebody near them even if they don't know each other.

Except you aren't even choosing who you heal... Wouldn't you prefer your heal went to your team mate or god forbid yourself? I'm sure the guy 30 meters away with no association to you appreciates it though...

I don't really see how his suggestion fits an agenda. What he is suggesting is a damn sight better than what the Devs are suggesting.

Suite_Ice

I already left a lengthy post as to my thoughts on this band aid test. I wanted to further state the issues that are being tested a way down on my list of issue needing fixed. The lisse of lag and latency is far worse in PvE places like world bosses during events even it on 2-4 players are there, dolmens in places where people sometime grind xp are many times terrible. I have a serious gaming rig and a Gig connection and Im constantly at time getting booted from server entirely back to log in screen with error to check my internet, or back to character selection screen. I die because of the lag and the game going on without me. I have in game issue where I cant get mail because my mail box is full but the only reason its full it the game wont delete empty mail when I tell it to delete mail. These are the kinds of issues that are far more important than what happens in Cyrodiil. This is only game or action my rig wont cut through like a hot knife through butter. These are the issues to work on and not nerfing PvP's AOE

Raevin

At this point I don't give a damn anymore. Test whatever you need to test, fix whatever needs fixing. ANYTHING that can make PvP work and not have your character spaz, freeze, unable to cast I'll take it. And if this discourages the lowest life form in Cyrodiil - ballgroups - oh boy! EVEN BETTER! Those brainless 1 skill spammers need to disappear.

SpacemanSpiff1

Raevin wrote: »

At this point I don't give a damn anymore. Test whatever you need to test, fix whatever needs fixing. ANYTHING that can make PvP work and not have your character spaz, freeze, unable to cast I'll take it. And if this discourages the lowest life form in Cyrodiil - ballgroups - oh boy! EVEN BETTER! Those brainless 1 skill spammers need to disappear.

it wont discourage ballgroups. they'll adapt better than most.

HeroOfNone

I would like to see more calculations/work handled client side, but threads like this show why it wouldn't work. We're talking about a test and people think its a solution. Imagine trying to tell users they were installing denuvo anti-cheat to help protect clients as they were changing the client/server process? Posters would complain that ZOS was using it to spy when they were going to the bathroom.