Next week performance test : turn off cross healing again

Sarousse

Hi devs,

Could you please next week turn off cross healing so we can see how goes the performance without proc sets ?

Obviously turning on cross healing + removing proc sets was a bad move as we can't identify which one is the worst for performance.

Thanks.

relentless_turnip

I agree I would like to see this. It does seem worse despite disabling most the sets. I think the majority of us believe switching ally healing back on equated in a net gain for server calculations and an overall worsening performance.
It should be done just to put our minds at ease.

NeillMcAttack

I think it should be pretty obvious that cross healing, or more precisely, “Smart” AOE checks are about 75% of the leading causes of performance issues. And I am even quite convinced now that it isn’t a processing speed problem on the server, but more a bandwidth/congestion and latency issue, all those pings from individuals to the server to individuals, back to the server until a “state” on the server can be determined is the main issue with all the latency. Of course this was due to putting every individual player state on the server with update 25.
But yes, I totally agree, we have no need for this test to continue with the current parameters. Every other test lasted a week. There wasn’t even a Queue for the reds on RW EU last night so populations are dwindling due to the horrendous performance.

So please ZOS, we need a more conclusive outcome.

It is worth noting though OP, that some campaigns still have large issues even with these AOE checks disabled. For example. Greyhost NA is a very competitive campaign, and in it there would be a lot more ball groups than I would be used to on RW EU, as such, if that is the campaign that ZOS monitors the most, it is likely they won’t notice it being a significant enough change to base any finding on. As such, I would suggest to the devs, that they remove cross healing and in addition, reduce group sizes to 4, this way we can get much more accurate results on the effects of large “smart” AOE checks!

rbfrgsp

Yes, it was a strange decision to alter two variables at the same time during a performance test. But performance these last ten days has been worse than it was when proc sets and cross-heals we're both available, some months ago.

ethenya

As much as I enjoy healing or being healed I would be interested to see how Cyro performance is without procs and cross healing. Agree with the OP 100%.

NeillMcAttack

rbfrgsp wrote: »

Yes, it was a strange decision to alter two variables at the same time during a performance test. But performance these last ten days has been worse than it was when proc sets and cross-heals we're both available, some months ago.

I honestly don’t blame ZOS for the decision to turn on the cross healing as being able to heal your allies should be the standard for Cyro. So having the no-procs with healing on gives the best indicator as to how procs effect the ‘standard’ Cyro.
I guess you could say that the lack of cross heals was itself a test of sorts.

Sanctum74

Sarousse wrote: »

Hi devs,

Could you please next week turn off cross healing so we can see how goes the performance without proc sets ?

Obviously turning on cross healing + removing proc sets was a bad move as we can't identify which one is the worst for performance.

Thanks.

Actually it’s the opposite, had they left cross healing off then they would have no way of knowing which affects performance. Having constant variables is the key basics to testing anything. Let them get their data it will be over soon.

React

I hope they decide to disable it again before the test ends. I'm no longer confident that anything will improve, but it is definitely worth trying and there are tons of people asking for it now.

Faded

They should disable grouping too, so nobody can heal anybody else, full stop. Best way to gauge the full impact.

Tommy_The_Gun

I think that they re-enabled cross healing simply because - it is Cyrodill "default" setting...

Remember that Cross healing turned off was also a test whenever it improves performance or not. And from data ZOS gathered they concluded it did not worked.

So, they reverted Cyro to "default" setting, which is with cross healing turned on - just to have "apples - to - apples" comparison.

QuebraRegra

LOL, this again?

Wanna fix performance? Undo the client server changes that were made, requiring far more server checks for even basic things (LOS for skills, etc.).

Change game code from TCP to UDP.

enjoi!

edges_endgame

Cross healing is not part of the test. It's back and here to stay.

slan77

Yes. Leave no proc and turn off cross healing. Performace is fine dealing with small groups, but when 30+ zerg comes it is horrible

Ranger209

NeillMcAttack wrote: »

I think it should be pretty obvious that cross healing, or more precisely, “Smart” AOE checks are about 75% of the leading causes of performance issues. And I am even quite convinced now that it isn’t a processing speed problem on the server, but more a bandwidth/congestion and latency issue, all those pings from individuals to the server to individuals, back to the server until a “state” on the server can be determined is the main issue with all the latency. Of course this was due to putting every individual player state on the server with update 25.
But yes, I totally agree, we have no need for this test to continue with the current parameters. Every other test lasted a week. There wasn’t even a Queue for the reds on RW EU last night so populations are dwindling due to the horrendous performance.

So please ZOS, we need a more conclusive outcome.

It is worth noting though OP, that some campaigns still have large issues even with these AOE checks disabled. For example. Greyhost NA is a very competitive campaign, and in it there would be a lot more ball groups than I would be used to on RW EU, as such, if that is the campaign that ZOS monitors the most, it is likely they won’t notice it being a significant enough change to base any finding on. As such, I would suggest to the devs, that they remove cross healing and in addition, reduce group sizes to 4, this way we can get much more accurate results on the effects of large “smart” AOE checks!

If they were to do this I would like to see them remove it completely whether in group or not. Make it so everyone can only heal themselves so that no one has a healing advantage based on being in a group. It isn't right that some people can benefit from the heals of others while other people cannot.

Ranger209

Faded wrote: »

They should disable grouping too, so nobody can heal anybody else, full stop. Best way to gauge the full impact.

Agree this would be a ground 0 test on what impact grouping has, not only from a smart healing standpoint but also from how the grouping UI may be involved with potential issues, as well as not having to run through any group "in combat" checks for XP/CP purposes of kills to be shared by the group.

Soul_Demon

Sanctum74 wrote: »

Sarousse wrote: »

Hi devs,

Could you please next week turn off cross healing so we can see how goes the performance without proc sets ?

Obviously turning on cross healing + removing proc sets was a bad move as we can't identify which one is the worst for performance.

Thanks.

Actually it’s the opposite, had they left cross healing off then they would have no way of knowing which affects performance. Having constant variables is the key basics to testing anything. Let them get their data it will be over soon.

They turned off cross healing while proc sets were on still and we had worse performance---they came out and said there was no appreciable impact cross heals made and simply turned them back on now. But since the proc sets are off and we still have major issues and no improvement in performance I feel like there is a set of players out there who will continue to ignore the facts right in front of them and claim they know the solutions---even after they have been tested. Why? To what end?

Faded

Soul_Demon wrote: »

Sanctum74 wrote: »

Sarousse wrote: »

Hi devs,

Could you please next week turn off cross healing so we can see how goes the performance without proc sets ?

Obviously turning on cross healing + removing proc sets was a bad move as we can't identify which one is the worst for performance.

Thanks.

Actually it’s the opposite, had they left cross healing off then they would have no way of knowing which affects performance. Having constant variables is the key basics to testing anything. Let them get their data it will be over soon.

They turned off cross healing while proc sets were on still and we had worse performance---they came out and said there was no appreciable impact cross heals made and simply turned them back on now. But since the proc sets are off and we still have major issues and no improvement in performance I feel like there is a set of players out there who will continue to ignore the facts right in front of them and claim they know the solutions---even after they have been tested. Why? To what end?

An excellent question.

sunandstars77

I'm no tech expert. But i have played other MMO's. Where factions in key battles on the PvP map were using some software to locally cause lag to change the outcome of a fight in their favor. I don't know if this was true or even possible. Maybe someone can say if something like that is even possible.

Aerenthir

sunandstars77 wrote: »

I'm no tech expert. But i have played other MMO's. Where factions in key battles on the PvP map were using some software to locally cause lag to change the outcome of a fight in their favor. I don't know if this was true or even possible. Maybe someone can say if something like that is even possible.

That has happened in WoW but it was in order to prevent another guild from their raid progress. Won't work if you also want to raid. Aka in this case if they lag the battle, they will lag themselves.

Not happening.

Sanctum74

Soul_Demon wrote: »

Sanctum74 wrote: »

Sarousse wrote: »

Hi devs,

Could you please next week turn off cross healing so we can see how goes the performance without proc sets ?

Obviously turning on cross healing + removing proc sets was a bad move as we can't identify which one is the worst for performance.

Thanks.

Actually it’s the opposite, had they left cross healing off then they would have no way of knowing which affects performance. Having constant variables is the key basics to testing anything. Let them get their data it will be over soon.

They turned off cross healing while proc sets were on still and we had worse performance---they came out and said there was no appreciable impact cross heals made and simply turned them back on now. But since the proc sets are off and we still have major issues and no improvement in performance I feel like there is a set of players out there who will continue to ignore the facts right in front of them and claim they know the solutions---even after they have been tested. Why? To what end?

I’m not ignoring any facts or claiming I have solutions, did you even read my comment?

I’m just trying to explain how a basic test is conducted, you don’t change multiple variables at the same time. Thats why every test is done separately so they know exactly how much cross healing, aoes, proc sets, etc affect performance individually. Then they can make a better determination on what to work on based on actual data instead of just changing something because someone don’t like it.

Personally I’m conflicted with the cross healing, it was great to be able to play some of my magic characters again and actually be able to heal myself reliably as a solo dd, but on the other hand I thought it sucked for the people that have done nothing but heal for 5 years. There was already a crown shortage and they clipped group sizes at the same time making it impossible for them to play their role.

QuebraRegra

sunandstars77 wrote: »

I'm no tech expert. But i have played other MMO's. Where factions in key battles on the PvP map were using some software to locally cause lag to change the outcome of a fight in their favor. I don't know if this was true or even possible. Maybe someone can say if something like that is even possible.

name another MMO that uses TCP exclusively, or even TCP for communicating combat, etc.... I'll be here waiting.

there's a reason most if not all use UDP for these functions... ESO code and engine are the issue. Spend some of that $93 million taken in last year on rewriting the code to go from TCP to UDP, using RTP sequencing and application echo/ack.

Jackey

QuebraRegra wrote: »

name another MMO that uses TCP exclusively, or even TCP for communicating combat, etc.... I'll be here waiting.

World of Warcraft?

erio

The no healing test should of been the control. We see performance before and right after. Now youve switched another variable

Sanctum74

erio wrote: »

The no healing test should of been the control. We see performance before and right after. Now youve switched another variable

All of the other tests were done with healing except for the no cross healing test so no they did not switch another variable, they returned it to default so they can test procs separately. Pretty straight forward way of testing.

Unless you want to go through all of the aoe and skill cool down tests all over again with no healing?

erio

Sanctum74 wrote: »

erio wrote: »

The no healing test should of been the control. We see performance before and right after. Now youve switched another variable

All of the other tests were done with healing except for the no cross healing test so no they did not switch another variable, they returned it to default so they can test procs separately. Pretty straight forward way of testing.

Unless you want to go through all of the aoe and skill cool down tests all over again with no healing?

They should of kept it off. now im trying to remember weeks back to when there was healing and I cant. How am I supposed to really tell the difference?

Ackwalan

It can't be just cross healing that causes this much lag. ZOS must have done something else in the background besides procs and cross healing.

Joy_Division

Sanctum74 wrote: »

Soul_Demon wrote: »

Sanctum74 wrote: »

Sarousse wrote: »

Hi devs,

Could you please next week turn off cross healing so we can see how goes the performance without proc sets ?

Obviously turning on cross healing + removing proc sets was a bad move as we can't identify which one is the worst for performance.

Thanks.

Actually it’s the opposite, had they left cross healing off then they would have no way of knowing which affects performance. Having constant variables is the key basics to testing anything. Let them get their data it will be over soon.

They turned off cross healing while proc sets were on still and we had worse performance---they came out and said there was no appreciable impact cross heals made and simply turned them back on now. But since the proc sets are off and we still have major issues and no improvement in performance I feel like there is a set of players out there who will continue to ignore the facts right in front of them and claim they know the solutions---even after they have been tested. Why? To what end?

I’m not ignoring any facts or claiming I have solutions, did you even read my comment?

I’m just trying to explain how a basic test is conducted, you don’t change multiple variables at the same time. Thats why every test is done separately so they know exactly how much cross healing, aoes, proc sets, etc affect performance individually. Then they can make a better determination on what to work on based on actual data instead of just changing something because someone don’t like it.

Personally I’m conflicted with the cross healing, it was great to be able to play some of my magic characters again and actually be able to heal myself reliably as a solo dd, but on the other hand I thought it sucked for the people that have done nothing but heal for 5 years. There was already a crown shortage and they clipped group sizes at the same time making it impossible for them to play their role.

This test isn't too compare it with no cross healing Cryodiil. They tried that option and decided to get rid of it because it didn;t help much and there were a lot of complaints about it. In short, it wasn't worth keeping. Why would they test a variable that they decided to drop? That would be just wasting time. There are years of data to compare what cyrodiil is like with proc set (launch - Sept 2020) and without (these three weeks). That's a valid test. And I suppose we learned something: procs aren;t the lag causing problem. Then again, I'm not sure why people go their hopes up to begin with, but that's a different conversation.

I have no desire to wait the months and months of the sort of let's try one variable at a time approach some people are advocating. None. Zero. And I highly doubt ZoS does either. It is ridiculous as it is that I have to pay to be a Beta tester for a game that has been out 7 years already. This is like the only possible test they could run that I don;t mind because I can't stand proc dominated gameplay. But that's just me. There are people out there who like procs and don't like this test.

badmojo

Just remove healing entirely, no self heals, no healing from potions, nothing.

I am not in favor of it, but if we are going to gut the game in the name of performance we might as well go all the way. No heals means quicker deaths which means less server calculations.

VaranisArano

erio wrote: »

Sanctum74 wrote: »

erio wrote: »

The no healing test should of been the control. We see performance before and right after. Now youve switched another variable

All of the other tests were done with healing except for the no cross healing test so no they did not switch another variable, they returned it to default so they can test procs separately. Pretty straight forward way of testing.

Unless you want to go through all of the aoe and skill cool down tests all over again with no healing?

They should of kept it off. now im trying to remember weeks back to when there was healing and I cant. How am I supposed to really tell the difference?

You aren't going to remember. Do you really remember the details of your own performance back to Greymoor before the testing started? How about before Harrowstorm when stuff got moved server-side for Stadia? Did you keep notes about your performance during all past tests to compare? I don't have a perfect memory of it myself! Just a general perception that "Things sure haven't gotten better."

Look, I get it. It's really tempting to think that we are tester and evaluator both. That our experience is what matters most during these tests. So if we can't test a specific variable right now, it doesn't feel right.

Thing is, we as individuals are not tester and evaluator, both. For one thing, the varying experiences people have had with performance from this test and indeed every other test should make that clear. Also, you know, I'm not getting paid to keep detailed notes on my performance issues over the testing, just getting double AP to play. Instead, ZOS is looking at their data, aggregated from the whole playerbase that participates in the tests - and they don't need to "remember". They still have their old data they can directly compare.

In that sense, whether or not you and I "remember" really doesn't matter. Even if we took detailed notes, individual experiences don't mean that much in the grand scheme of things. ZOS should be making their decisions based on their current and saved data from the whole of the playerbase during testing.

And if we want to be really sticklers about "variables", then ZOS should have exempted Cyrodiil from all base game updates since the testing started. That didn't happen because ZOS isn't running a clinical experiment. They are trying to test stuff on a Live server while still keeping the game moving and balancing their need to test certain variables against the knowledge that their playerbase doesn't necessarily have unlimited patience for testing. It's messier than, say, lab testing, but not necessarily inaccurate for ZOS' purposes.


Side note: I don't remember all this hand-wringing about variables when ZOS announced they were changing two things, reducing group size and eliminating cross-heals for "behavioral reasons" rather than performance, but now that they've brought back cross heals, suddenly it's all "You can't change two things, ZOS, how will we know what made the difference in performance?!"

Kwoung

VaranisArano wrote: »

You aren't going to remember. Do you really remember the details of your own performance back to Greymoor before the testing started? How about before Harrowstorm when stuff got moved server-side for Stadia? Did you keep notes about your performance during all past tests to compare? I don't have a perfect memory of it myself! Just a general perception that "Things sure haven't gotten better."

Quite honestly, I do not remember exact details back to about a year ago on Greyhost when I played there exclusively. I do however, remember that the lag was so bad I decided to go play on Blackreach instead, just to see how PVP would be when skills and things actually worked, with no rubberbanding, no dieing to 30 attacks landing on you at once, no instant disconnect if the hammer rounded the corner, etc... and all the other stuff that was so common that I do still remember quite clearly.

It was a world of difference on BR and despite the low pop, my friends and I felt it was a much better option than the continued frustration of well... pretty much all the same things that exist today before, during and after every test they have done to date. IMHO, nothing has changed for better or worst in the past year.

Soul_Demon

Sanctum74 wrote: »

Soul_Demon wrote: »

Sanctum74 wrote: »

Sarousse wrote: »

Hi devs,

Could you please next week turn off cross healing so we can see how goes the performance without proc sets ?

Obviously turning on cross healing + removing proc sets was a bad move as we can't identify which one is the worst for performance.

Thanks.

Actually it’s the opposite, had they left cross healing off then they would have no way of knowing which affects performance. Having constant variables is the key basics to testing anything. Let them get their data it will be over soon.

They turned off cross healing while proc sets were on still and we had worse performance---they came out and said there was no appreciable impact cross heals made and simply turned them back on now. But since the proc sets are off and we still have major issues and no improvement in performance I feel like there is a set of players out there who will continue to ignore the facts right in front of them and claim they know the solutions---even after they have been tested. Why? To what end?

I’m not ignoring any facts or claiming I have solutions, did you even read my comment?

I’m just trying to explain how a basic test is conducted, you don’t change multiple variables at the same time. Thats why every test is done separately so they know exactly how much cross healing, aoes, proc sets, etc affect performance individually. Then they can make a better determination on what to work on based on actual data instead of just changing something because someone don’t like it.

Personally I’m conflicted with the cross healing, it was great to be able to play some of my magic characters again and actually be able to heal myself reliably as a solo dd, but on the other hand I thought it sucked for the people that have done nothing but heal for 5 years. There was already a crown shortage and they clipped group sizes at the same time making it impossible for them to play their role.

I understand the frustration with the lack of scientific approach, but we as a community have tested quite a bit at this point. We have put delays on AOE's, limited skills use to either an AOE or HOT one time in 3 seconds but not both along with other artificial game introduced delays on skills- We have done no champion points, no proc sets no heals, group size reductions, no addons. In virtually every case the server lags and their is only variability on either how bad it lags or how fast you can reconnect after you crash out. But in all the above scenarios we have seen absolutely not one single fix introduced by taking away a core function of the game.....even the group size reduction looks like a Dev plays for a ball group and this is being done for the sole purpose of reduction of the counters to them.

So, if we pay attention to those tests and played for them (I have) I think its easy to see either one of two things is the ONLY reason we lag. Code is FUBAR

---

or Hardware (servers) are not adequate. One of the two, maybe a combination of both. We have NOT seen over the years a concerted effort to fix bugs. At this point I would wager one could put together about a page and half of bugs listed that the Devs have outright ignored for years now. What does that do to the game? Who knows, maybe the game is now sentient and with so many bugs is plotting world domination.....but regardless some logic to the approach has to be the consistent players in game coming to forums saying "try this, try that" when they have told us they 'have a plan' and are doing it despite our frustrations while forgetting or not paying attention to the previous tests we have run. If we have already tried it and it did not dispel lag and crashes----lets just leave it off the table. It may not be perfect scientific method but at least we wont be running in circles for the sake of a few posts where most of the time the player just wasn't there for the test and or does not remember when we tested that stuff.