Update on Cyrodiil Performance & Upcoming AOE Tests

NotTaylorSwift

https://drive.google.com/file/d/1HZSJAfe2B0KBHzZ2D2NQ8EnJlkJdcoMG/view?usp=sharing

Worst performance yesterday since even before any of the aoe tests on no cp pc eu, as well as nice perma slow bug (psijic ult bug?) in Roebeck which lasted for almost 5 whole minutes considering the clip is 2 mins long, we ressed at camp, returned and all players were still slowed. Also happened 3 or 4 times in the 40 ish minutes we were there.

FlaviusPK

Few moments later - https://coub.com/view/2jtxj2

relentless_turnip

Terrible performance this evening, I'm really not sure aoe's are the issue. Can you please just revert whatever change you made for harrowstorm. Performance got dramatically worse and it hasn't recovered...

All of these tests just feel futile as there has been no improvement, just the added frustration of have 3 second delay on a skill you have to spam.

I have taken part excessively in every test so far and I can assure you, it hasn't shown any improvements. In fact I would say it is often worse than it was before the tests.

casparian

relentless_turnip wrote: »

Can you please just revert whatever change you made for harrowstorm. Performance got dramatically worse and it hasn't recovered...

They cannot. They’re committed to Stadia now, and moving more processes server side (without investing money in improving server performance) was part of what it took to make that happen.

We’ll probably never see pre-Harrowstorm levels of performance again, let alone 2015-18 performance.

olsborg

Demonhunter wrote: »

It's amazing to me when people say that Cyrodiil is un playable they assume it's on all the servers, servers like blackreach for me never lags probably because it's never population capped on all faction?

I had severe lagg on blackreach yesterday and the factions was high and medium popped. Same with gray sometimes, its only high popped , but the lagg is severe. The only time I dont lag at all is mornings.

techyeshic

Last night; was the worst I ever saw lag and it was really pre-primetime. During maybe the most harsh AOE test yet.

relentless_turnip

techyeshic wrote: »

Last night; was the worst I ever saw lag and it was really pre-primetime. During maybe the most harsh AOE test yet.

I totally agree! Pc EU ravenwatch was the worst I'd ever seen it...

techyeshic

Seriously. Tonight on PCNA crashed twice in an hour. Abilities worse ever delay, all single target take longer than AOEs with CDs. Clearly; the AOE on CD and ramping cost does not work. PTS being live on the other hand; proves once again to impact performance, bigly.

psk177ub17_ESO

So AOE changes during this performance test, in my experience in Greyhost, Blackreach, and Ravenwatch aren't really showing much of a performance increase or decrease. The performance is bad just as it was without the AOE changes. I hope the data you guys have is consistent with this otherwise, I might just need to find another game (no threat) but it would indicate that my location/pc or something will never really let me play this game at it's full potential. I hope to hear good news soon. Take care.

relentless_turnip

psk177ub17_ESO wrote: »

So AOE changes during this performance test, in my experience in Greyhost, Blackreach, and Ravenwatch aren't really showing much of a performance increase or decrease. The performance is bad just as it was without the AOE changes. I hope the data you guys have is consistent with this otherwise, I might just need to find another game (no threat) but it would indicate that my location/pc or something will never really let me play this game at it's full potential. I hope to hear good news soon. Take care.

No mate, I concur... The performance is just as bad as before the aoe tests!

alainjbrennanb16_ESO

NotTaylorSwift wrote: »

https://drive.google.com/file/d/1HZSJAfe2B0KBHzZ2D2NQ8EnJlkJdcoMG/view?usp=sharing

Worst performance yesterday since even before any of the aoe tests on no cp pc eu, as well as nice perma slow bug (psijic ult bug?) in Roebeck which lasted for almost 5 whole minutes considering the clip is 2 mins long, we ressed at camp, returned and all players were still slowed. Also happened 3 or 4 times in the 40 ish minutes we were there.

but you do run as a ball group, try and expand your groups area, not sure why iam saying that, if you stay as a ball then its easier to kill you ;p, but it does show the issues at the mo and ta for this it will help mos break you up and the rest of ep on ravenwatch

NotTaylorSwift

Undaunted orbs synergy not working. The synergy appears and can be activated but it doesn't actually do anything and can be activated over and over with no cooldown.

Svenja

NotTaylorSwift wrote: »

Undaunted orbs synergy not working. The synergy appears and can be activated but it doesn't actually do anything and can be activated over and over with no cooldown.

We noticed that too. It didn't matter if the caster was in our group or not.
Short video of the bug:
https://youtu.be/ReVEFoSJLKQ

Tigor

The fifth test: This does not make us happy. I would like to quickly go to the last test, which I expect will be more fun again.

After that I also hope to see results of the tests soon, and implementation of needed improvements. Thank you.

Adding a note: During this test the latency and fps are working fine, so nothing wrong with the connection to the server. Only something is still causing a disadvantage concerning particular skills that I use. Some examples applicable from my StaNec.

- Blastbones are not running off after 2.5 seconds, falling apart.
- Invasion is only stunning after a few seconds and weapon swap is not directly accessable.
- Carve is hitting nothing but air.
- Sudden deaths.
- etc.

SturgeHammer

Ward Ally is still targeting outside of group. Don't know if that is intentional where its not a heal. I was looking forward to it only hitting players in my group, as that is a buff to the ability in most situations.

PoppaVape

How about just having a dragon swoop down and blast players anytime there is more than 20 of one faction in a certain radius? Perhaps that would keep factions from stacking so deep.

silentxthreat

After playing these tests and seeing server performance during each tests it really leans towards the ball groups causing lag. Every time they start farming at any place the server suffers. I don't think its anything aoe related though that me part of it but I think that its add ons that communicate with the server that stresses the server. ZOS should have the tools to see what the pings are that overload the server at any time. If its as simple as a add on that tracks group damage ult resources ect just make that against tos and let the groups farm as normal? just my 2 cent rant and wanting a playable server

Idinuse

...

silentxthreat wrote: »

After playing these tests and seeing server performance during each tests it really leans towards the ball groups causing lag. Every time they start farming at any place the server suffers. I don't think its anything aoe related though that me part of it but I think that its add ons that communicate with the server that stresses the server. ZOS should have the tools to see what the pings are that overload the server at any time. If its as simple as a add on that tracks group damage ult resources ect just make that against tos and let the groups farm as normal? just my 2 cent rant and wanting a playable server

IMO ZOS should make an audit as to what add ons the ball groups are using. Could be interesting finds.

asneakybanana

silentxthreat wrote: »

After playing these tests and seeing server performance during each tests it really leans towards the ball groups causing lag. Every time they start farming at any place the server suffers. I don't think its anything aoe related though that me part of it but I think that its add ons that communicate with the server that stresses the server. ZOS should have the tools to see what the pings are that overload the server at any time. If its as simple as a add on that tracks group damage ult resources ect just make that against tos and let the groups farm as normal? just my 2 cent rant and wanting a playable server

The main addon that ball groups use is RDK group tools which pings the map once ever second or two but ZoS already has a limit on how often the map can be pinged and if you ping too often you get kicked for spamming. These pings allow the addons to share information based on user's ultimate status, damage done, and synergy cooldowns (Theoretically it could be used for anything but those are the main uses) but all those different things use the same ping cooldown.

The main reason it lags when a ball group starts a farm somewhere is because of how it attracts the enemy factions. When they go to an enemy tri keep and farm the 3rd floor attracting 40-60 enemies, that means those 40-60 enemies are all in range of each other, causing every heal to do a radius check to all 60 of those players and then do further calculations on a large amount of players from there. Ball group farms is not the only time the game lags either, faction stack fights at things like last emp keep or scroll fights or any long fight that attracts 60 players from both factions is going to cause lag.

I think the above statement could best be seen last week during the group buffs only going to your own group test. While this obviously isn't a great solution long term as it basically ruins cyrodiil for anyone looking to play a healer outside of a group, it can be used as a basis for future tests or changes. Maybe if your group is over 4 you can only heal your group, this way a small group of players can still do something and feel they are helping their faction while also greatly limiting the amount of cross healing that will be done between groups (Such as what currently occurs in these large scale fights).

SgtNuttzmeg

I have to say, I wasn't sure how I was feeling about the 12 person group caps at the start but I am really enjoying it.

SgtNuttzmeg

After playing through all the testing phase in both a group and solo setting I would half to say the biggest decrease of lag definitely occurred during the fifth testing environment. Forcing heals to remain within a group had the most significant impact on performance. That being said, lag was still present and I noticed a significant difference in-between playing solo and in a group from a performance stand point. Performance is usually significantly worse when you play by yourself. The other important element I noted was the effects these changes had on sustain. Cooldowns without cost increases created an environment where most roles over sustained without any investment into sustain.

However my recommendations from testing would be:

Imposing a 3 second cooldown on every unique aoe buff, DOT and HOT skills. People shouldn't be spaming them anyway.

Buffing the number of targets hit by a single cast Mutogen can hit but ensuring that only one instance per caster could ever be cast.

Limit all friendly AOEs to group members in groups of 4 or more.

Limiting group size to 12 players.

Escalating costs of all AOE Spamables. Treat them all like streak.

Reduce the damage from Siege or remove the ability to stack siege damage. Reduce the damage of DOT proc sets, shift damage more towards procs that burst. Then change purge. This skill stills need to be tweaked as it definitely is a big source of lag. Reduce the need for it and change it so it is no longer a spamable. (The only exception to this is the Pestilent Host.)

Look at the Undaunted tree or create a new tree with some new single target spamable utility abilities. A lot of the initial complaints I heard came from people lacking stuff to do in combat. We need more single target abilities for Healers and Supports that are relevant to PVP group play. More skills like Silver Leash and Bone Shield. We also need a single target heal ability for healers that is not Mend Wounds or Mend Wounds needs a buff.

Stamina needs a generic delayed burst skill for pvp combat. Most damage done by Stam DDs comes from Cleeve or Spin2win. Other DDs simply outclass them do the fact that proxy det exists. Either look at tweaking the Azurablight set or make a new skill. Ideally whatever this is should synergize with Pestilent Host. (Pestilent Host should only trigger on player death too.)

Obviously Jabs will need to be reworked.

It is a lot easier to outplay ball groups with some of these changes as ballgroups need to focus their attacks during very specific windows. This gives small scale players openings and opportunities because they can wait for a Ball Group to attack and then go in and try to disrupt it. This combined with the nerfs to buffs/debuffs will definitely lower the skill celling of group vs group combat and will definitely help to gut ball groups. Reducing the ability to stack HOTs as mentioned above should also help to further curtail ball groups. I think if you played around with these suggestions you could find a really good balance that allows for competitive group play, doesn't destroy the servers and curtails the power of ball groups.

keto3000

I know.... how about bring back 2014's:

#PVP-GROUND-OILS-FTW!!!


https://youtu.be/Imv1vQQ1GVk

Reference: Found this old thread..and couldn't resist posting https://forums.elderscrollsonline.com/en/discussion/129056/flaming-oil

ZOS_GinaBruno

Hi all,

Just posted an update in the main stickied thread, but wanted to cross-post here as well for extra visibility.

As we approach the end of our scheduled tests, we’ve determined that we’d like to run at least two more tests to gather additional data, which are outlined below. Our goal here is to see how a combination of changes from previous tests affect the performance, as we’ve found they have not had a significant impact individually.

• October 19 - 26:
  • Shared global AoE cooldown with a 3 second timer
  • Group size limited to 12 in Cyrodiil
  • Ally-targeted abilities only applied to group members
• October 26 – November 2:
  • Shared global AoE cooldown with a 3 second timer
  • Group size limited to 12 in Cyrodiil
  • Ally-targeted abilities only applied to group members
  • Global ramping cost per successive AoE cast by 50%/100%/150%
  • Regen decrease per successive AoE cast by 33%/66%/99%

Our current plan is to disable all tests and double AP once Update 28 launches for PC on November 2. At that time, we’ll need to take some time to analyze the data collected from all the tests and determine next steps. Once we have a better idea of what that entails, we’ll let everyone know.

NotTaylorSwift

How can you even say that the 3 second global timer had good results? Of course it had good results, no one could use skills. And despite most of the community saying that they would rather play with bad performance than have these global cooldowns, you’re pushing this as a fix and even expanding on it?

Are there any plans to rework every single class and skill line if these changes are the solution you are going for? Because every class except nightblade is, for the most part, unplayable with 3 second global cooldown.

Maybe it wasn’t a good idea to add more and more proc sets every single patch also. I guess now cyrodiil will simply be: slot as many proc sets as possible and just light attack away with no skills. You might as well just disable all skills at this point, maybe that will help with lag

Derra

ZOS_GinaBruno wrote: »

Hi all,

Just posted an update in the main stickied thread, but wanted to cross-post here as well for extra visibility.

As we approach the end of our scheduled tests, we’ve determined that we’d like to run at least two more tests to gather additional data, which are outlined below. Our goal here is to see how a combination of changes from previous tests affect the performance, as we’ve found they have not had a significant impact individually.

• October 19 - 26:
  • Shared global AoE cooldown with a 3 second timer
  • Group size limited to 12 in Cyrodiil
  • Ally-targeted abilities only applied to group members
• October 26 – November 2:
  • Shared global AoE cooldown with a 3 second timer
  • Group size limited to 12 in Cyrodiil
  • Ally-targeted abilities only applied to group members
  • Global ramping cost per successive AoE cast by 50%/100%/150%
  • Regen decrease per successive AoE cast by 33%/66%/99%

Our current plan is to disable all tests and double AP once Update 28 launches for PC on November 2. At that time, we’ll need to take some time to analyze the data collected from all the tests and determine next steps. Once we have a better idea of what that entails, we’ll let everyone know.

Couldn´t you have like - kept those timers and debuffs to grps larger than 4 players or something?

The universial cooldowns render virtually any magica class unplayable solo or duo. Especially the last one means magica will be outright disabled - because almost all builds have core functions solely available as aoe skills.

MachineGod

@ZOS_GinaBruno

Don't get me wrong these tests are welcomed but you need to really evaluate this last test.

How can anyone play during October 26 – November 2? That's punishment beyond belief. You may as well disable AoE abilities at that point. You won't gain data for that because nothing is going to be cast in any meaningful way. If you want players to play during that test you are going to need more than just "double AP" if you go through with it.

Also Why not address siege weapons effects on cyro? Why is this not mentioned?

Satiar

If ZoS wanted to just end Cyrodiil they didn't have to do this whole song and dance. Just send me a PM, I'll create the mega-zerg and melt the megaserver down for good.

Crash427

Satiar wrote: »

If ZoS wanted to just end Cyrodiil they didn't have to do this whole song and dance. Just send me a PM, I'll create the mega-zerg and melt the megaserver down for good.

It does seem like its about time to give Cyro the old yeller treatment.

MurderMostFoul

ZOS_GinaBruno wrote: »

October 26 – November 2:

• Shared global AoE cooldown with a 3 second timer
• Group size limited to 12 in Cyrodiil
• Ally-targeted abilities only applied to group members
• Global ramping cost per successive AoE cast by 50%/100%/150%
• Regen decrease per successive AoE cast by 33%/66%/99%

I'm confused, doesn't a shared global AOE cooldown negate the effect of global ramping cost per successive AOE cast?

I thought that the shared global cooldown / ramping costs tests were supposed to run independently of one another. Are you saying with this test that we can cast jabs, wait three seconds, cast jabs again and still incur a 50% cost increase?

Why not just remove AoEs from the game?

HankTwo

This last test is just insanity beyond believe. The only sound reason I can think of is to show the financial department how underbudgeted the server hardware really is if they still have lag after that.

MurderMostFoul wrote: »

ZOS_GinaBruno wrote: »

October 26 – November 2:

• Shared global AoE cooldown with a 3 second timer
• Group size limited to 12 in Cyrodiil
• Ally-targeted abilities only applied to group members
• Global ramping cost per successive AoE cast by 50%/100%/150%
• Regen decrease per successive AoE cast by 33%/66%/99%

I'm confused, doesn't a shared global AOE cooldown negate the effect of global ramping cost per successive AOE cast?

I thought that the shared global cooldown / ramping costs tests were supposed to run independently of one another. Are you saying with this test that we can cast jabs, wait three seconds, cast jabs again and still incur a 50% cost increase?

Yeah, I was also wondering about that. The only way this can even work is if the ramping cost increase and regen debuff run on a duration longer than 3 seconds. For example, once you cast an AOE, you will get a 3 second AOE cooldown + say a 5 second long ramping cost stack.

MurderMostFoul wrote: »

ZOS_GinaBruno wrote: »

October 26 – November 2:

• Shared global AoE cooldown with a 3 second timer
• Group size limited to 12 in Cyrodiil
• Ally-targeted abilities only applied to group members
• Global ramping cost per successive AoE cast by 50%/100%/150%
• Regen decrease per successive AoE cast by 33%/66%/99%

Why not just remove AoEs from the game?

I agree, that would be far simpler but wouldn't make much of a difference compared to the last test at this point...