Update 28 Game Performance Preview

MishMash

I wish they would SHARE the results of each test with us. We deserve to know as we were the ones that performed the tests for them!

IF they've already posted the results elsewhere for the players to review, I would like to see them.

LarsS

There is a simple reason that the mayhem events is less laggy, as was the first test week. In both cases the organized guilds and experienced players prefer to go elswhere. So the load on the servers decreases.

AMeanOne

Week 1 test should be invalid. Re run the test if you want, but you can't take these results seriously.

Seyer530

Even with the double AP, being stuck at home due to COVID-19, and finding some cool people to play with, I find myself RAGE QUITTING more than ever before. I have played this game since launch, on and off for breaks, but I may not come back to this game if GCDs are a thing in PVP.

I promise you, every other MMO on the market has significantly much less issues than this game, with GCDs, and will probably be more fun to play. No one likes 20 FPS combat.

TheMightyRevan

ZOS_GinaBruno wrote: »

Kingslayer513 wrote: »

ZOS_GinaBruno wrote: »

FYI: Just corrected a typo in the post. Originally one of the bullet points stated the global cooldown test brought server FPS in high-intensity situations down almost 45% - this should've said 25%, and has been corrected in the original post.

Pretty sure that should say "ping" or "latency" instead of FPS. FPS makes no sense in that context and you don't measure FPS in milliseconds.

In this case, we're using FPS to refer to server frames.

Milliseconds is not the unit for FPS, FPS doesnt have a unit, its a dimensionless quantity, in this case. So writing: our ideal server fps is less than 30 ms, is just wrong. if you want to use a unit for FPS use either nothing or hertz

Ostonoha

You are all talking about how is test 1 had the best reduction in server lag. But how is test 1 even considered a valid test when most of the pvpers did not even touch cyro and went right into IC for the event? [snip] I dont know how any of you dont understand how if you run a pvp test in cyro while a pvp event is going on in IC makes any data you collected that week useless. [snip]

[Edited for Bashing]

Can edit my comments and delete them but it does not make any of it less true.

Seriously cannot wait now till these AOE cooldowns happen. We all know the test(s) were a failure and wont help the lag. But with the cooldowns we will see months and months of class rebalancing, changing of stats, rework on animations, ext to make half the skills or classes (templar lol) even useable. Instead of that time being taken to add new content that could be sold. Im sure the higher ups must love hearing that the devs will be spending their time reworking and rebalancing things instead of making new sellable content. Unless your all going to just leave the classes how they are and let half of them be totally useless in pvp. All in all it will all be one giant waste of time since in the end the lag will still be there and hours of workhours will be spent wasted.

Loralai_907

I have questions. Assuming you do a gcd, how do you imagine that to work? With the thought being that there has never been a separation of how things work in PvP vs how they work in PvE, whats the plan here? Will you be having skills do one thing in PvE and another in PvP, OR, will you now have to look at PvE content as well and adjust things as needed, since there is going to be a shift in how much damage classes can be doing, in my head that means that there might be certain things that just cannot be done now. Obviously, I dont expect an answer here, but these are things that I need to know so I can figure out where I go from here. And I need to get an idea for the scale of how much I am trusting you to change/fix/get right.

Mostly, I'm just not at all interested in the idea of a gcd being a permanent thing, I've been away since a bit before this all started because just knowing where you all were heading caused me to lose my heart in all of this. I could maybe readjust my thinking and do more in the dungeons and trials area, and just not PvP anymore, but that will depend on how everything plays out. I know you just want Cyro to work right, we all do. But it just makes me super sad that this is how you want to do that. I know I am not alone in saying if this is the direction you go, I can't see myself spending much time if any in Cyro going forward.

RMerlin

Calypso589 wrote: »

Spend some o dat Microsoft money and upgrade the servers.

Upgrade out of DX9 too.

Sale not expected to close until the end of 2021.

I will have quit this game before that.

RMerlin

TheMightyRevan wrote: »

ZOS_GinaBruno wrote: »

Kingslayer513 wrote: »

ZOS_GinaBruno wrote: »

FYI: Just corrected a typo in the post. Originally one of the bullet points stated the global cooldown test brought server FPS in high-intensity situations down almost 45% - this should've said 25%, and has been corrected in the original post.

Pretty sure that should say "ping" or "latency" instead of FPS. FPS makes no sense in that context and you don't measure FPS in milliseconds.

In this case, we're using FPS to refer to server frames.

Milliseconds is not the unit for FPS, FPS doesnt have a unit, its a dimensionless quantity, in this case. So writing: our ideal server fps is less than 30 ms, is just wrong. if you want to use a unit for FPS use either nothing or hertz

I thought that maybe she was rather referring to frame latency, tho still I have a hard time seeing how this would be relevant in a server environment, since frames are rendered on the client, not on the server.

She was probably referring to processing latency, i.e. the delay in ms between the moment the server receives a request, and the moment it sends back the response to your client.

Giving her a break tho, this is pretty technical stuff

forzajuve212

forthwinds wrote: »

On PC/NA there were never 2 factions poplocked at a time during the first week of testing. I highly doubt, even if there was one campaign that reached pop lock across, that there were huge faction fights like in the other weeks of testing. This is because Test 1 was during the Imperial City Event. The devs also have to realize that putting cooldowns in this game will ruin it and completely change the core combat.

A lot of us play this game because the combat is amazing compared to other MMOs. If you go through with putting cooldowns in this game, there's no reason why people would stay here vs. another MMO that has better performance. It's pretty sad that this change to core combat is even being discussed over 5 years after release.

This ^^^

Parasaurolophus

I just don't understand why so many people speak so badly about tests? Is it really that hard to just wait a few weeks and change your build a little? Is it better than continuing to play with terrible lags? THIS IS ONLY ONLY TESTS FOR COLLECTING SOME DATA AND ALL THAT

virtus753

BrownChicken wrote: »

I just don't understand why so many people speak so badly about tests? Is it really that hard to just wait a few weeks and change your build a little? Is it better than continuing to play with terrible lags? THIS IS ONLY ONLY TESTS FOR COLLECTING SOME DATA AND ALL THAT

They’re not just a data-gathering exercise, though. The devs were very explicit that whatever they found worked well in testing would be implemented in some way, even if that meant changing the combat experience significantly.

DigitalHype

TheMightyRevan wrote: »

ZOS_GinaBruno wrote: »

Kingslayer513 wrote: »

ZOS_GinaBruno wrote: »

FYI: Just corrected a typo in the post. Originally one of the bullet points stated the global cooldown test brought server FPS in high-intensity situations down almost 45% - this should've said 25%, and has been corrected in the original post.

Pretty sure that should say "ping" or "latency" instead of FPS. FPS makes no sense in that context and you don't measure FPS in milliseconds.

In this case, we're using FPS to refer to server frames.

Milliseconds is not the unit for FPS, FPS doesnt have a unit, its a dimensionless quantity, in this case. So writing: our ideal server fps is less than 30 ms, is just wrong. if you want to use a unit for FPS use either nothing or hertz

The servers simulation rate can be measured in different units. Typically this is refered to as the "tick rate" of the server. IN that case it is measured in Hertz (Hz). Using Hz as the unit of measurement would mean higher numbers are better. But, it can also be represented in units of time per server tick (or what they are calling frame). In that case it is in millisecond (ms) of time. And this is what ZOS is calling "Server FPS". Lower is better with this unit of measurement.

A 30hz tickrate means the server takes approximate 33ms of time per tick. A 60hz tickrate would mean 16ms. This 30ms means they are trying to hold a tickrate of 30hz. In FPS games this would be a low-end server tickrate and commonly what is seen on older gen console games. Keep in mind those have client side predicitions and other things to help reduce the delay for the player.

What this "30ms server FPS" translates to for the player is 30ms of delay + latency to the server before your skills fire. So if you have 70ms latency, then you can expect 100ms of delay if the server can hold that rate. This is a simplification, as there can be other factors.

The data reported they are seeing "spikes above 50ms in large battles". The key in this statement is "above". We don't know how high those spikes go. But I'm guessing they are probably 100's of ms or more in some cases. Meaning your combat experience is going to be abysmal.

Even 50ms is considered bad for any other online game with realtime combat. Other MMO's typically do not have to worry about this because they use cooldowns on skills and the combat is at a much slower pace. But ESO's combat is more real time and action oriented. In other games of this nature the commonly accepted "Server FPS" would be 16ms or less. They are begging for 30ms and not even getting that.

As you can see, this is why they are trying to slowly move the community to a more cool-down based combat system. With everything being server side now (anti-cheat + stadia), they simply don't have the architecture (either server or software, or both) to handle the combat as it was originally intended.

Crash427

DigitalHype wrote: »

TheMightyRevan wrote: »

ZOS_GinaBruno wrote: »

Kingslayer513 wrote: »

ZOS_GinaBruno wrote: »

FYI: Just corrected a typo in the post. Originally one of the bullet points stated the global cooldown test brought server FPS in high-intensity situations down almost 45% - this should've said 25%, and has been corrected in the original post.

Pretty sure that should say "ping" or "latency" instead of FPS. FPS makes no sense in that context and you don't measure FPS in milliseconds.

In this case, we're using FPS to refer to server frames.

...

Even 50ms is considered bad for any other online game with realtime combat. Other MMO's typically do not have to worry about this because they use cooldowns on skills and the combat is at a much slower pace. But ESO's combat is more real time and action oriented. In other games of this nature the commonly accepted "Server FPS" would be 16ms or less. They are begging for 30ms and not even getting that.

As you can see, this is why they are trying to slowly move the community to a more cool-down based combat system. With everything being server side now (anti-cheat + stadia), they simply don't have the architecture (either server or software, or both) to handle the combat as it was originally intended.

So in other words Cyro the way we want may simply not be feasible at this point for a number of reasons? I'd have more respect for Zos if they just came out and said that and started looking for an alternative. Or even if they just said, "hey given the limitations of x,y, and z this is as good as it gets. Here. Have a 12v12 arena as a consolation prize."

DigitalHype

Crash427 wrote: »

So in other words Cyro the way we want may simply not be feasible at this point for a number of reasons? I'd have more respect for Zos if they just came out and said that and started looking for an alternative. Or even if they just said, "hey given the limitations of x,y, and z this is as good as it gets. Here. Have a 12v12 arena as a consolation prize."

Rich has basically said this in his original post about the testing. Some form of it is going to happen. Minus the consolation prize.

Daviiid_ESO

On test #5, October 5-12, the performance seemed a lot better for a lot of people. Better than any of the tests performance wise and is the least intruding in terms of gameplay. Why not provide data or mention those before introducing new tests? @ZOS_GinaBruno

Rather introduce this test to live that actually improved performance, and then balance pvp, after the fact, than adding a GCD and putting a lot of uneccessary restrictions on players as in the new tests.

Going into PVP as a new player with all these restrictions will be a very jarring experience I'm sure, compared to their experience in PVE.

Edit:
Also it would be cool if you could include some sort of reference to server load to previous patches as these numbers mean nothing to us as they are by themselves. How was the server load on your side pre-summerset for example - in terms of "server FPS"? - The game worked a lot better in PVP back then (especially in terms of positional desyncs, which started in summerset and became worse patch after patch.)

pma_pacifier

virtus753 wrote: »

BrownChicken wrote: »

I just don't understand why so many people speak so badly about tests? Is it really that hard to just wait a few weeks and change your build a little? Is it better than continuing to play with terrible lags? THIS IS ONLY ONLY TESTS FOR COLLECTING SOME DATA AND ALL THAT

They’re not just a data-gathering exercise, though. The devs were very explicit that whatever they found worked well in testing would be implemented in some way, even if that meant changing the combat experience significantly.

Imagine if the devs found that testing worked well, that it was "less server load on their end, e.g. 25%"
While players felt no change or worse lag.

Would ANY form of adjustments, considering wholesale adjustments be even enough to improve the player experience?
Please note, player experience should be a combination of 1. Gameplay experience 2. Combat experience
Can we even expect a NET increase in player experience?? I think it will be a NET DECREASE in any case.

For the sake of "less stress on their servers " and "more stress on their players " ????

ESO_Nightingale

Are they going to push the global aoe cooldown through now? Wtf? Well i guess you can't have lag when no-one plays your game anymore.

Thechuckage

*popcorn*

Leslik

pma_pacifier wrote: »

WHY WAS THE LAG BETTER DURING MIDYEAR MAYHEM. IT WAS BETTER THAN DURING THESE TESTS. YOU GUYS WERE DOING SOMETHING RIGHT! I WONDER...

jad11mumbler wrote: »

pma_pacifier wrote: »

WHY WAS THE LAG BETTER DURING MIDYEAR MAYHEM. IT WAS BETTER THAN DURING THESE TESTS. YOU GUYS WERE DOING SOMETHING RIGHT! I WONDER...

During the events they usually increase server capacity, or such, which costs more to run from what I've heard.

^^ This

It may be a while before the purchase by ms results in much change but you could still put your money where your mouth is rather than unpopular changes. I hear ms have a couple of datacenters. Maybe you could use some of that power.

Miswar

BrownChicken wrote: »

I just don't understand why so many people speak so badly about tests? Is it really that hard to just wait a few weeks and change your build a little? Is it better than continuing to play with terrible lags? THIS IS ONLY ONLY TESTS FOR COLLECTING SOME DATA AND ALL THAT

Because most of us know and as the "test" have shown these have not increased the performance whatsoever.

There is your answer and there will no better until they start investing in proper server farms etc. [snip]

The lag is bad as ever, skills dont fire, desyncs are horrid etc... there is your test results in nutshell.

[Edited to remove Bashing]

Sarannah

Question to ZOS: Could Cyrodiil's dynamic map cause most of the lag?... As it constantly keeps updating every objective to every player in there. Which does not just use servercapacity, but also takes up bandwidth to/from the server.

virtus753

Leslik wrote: »

pma_pacifier wrote: »

WHY WAS THE LAG BETTER DURING MIDYEAR MAYHEM. IT WAS BETTER THAN DURING THESE TESTS. YOU GUYS WERE DOING SOMETHING RIGHT! I WONDER...

jad11mumbler wrote: »

pma_pacifier wrote: »

WHY WAS THE LAG BETTER DURING MIDYEAR MAYHEM. IT WAS BETTER THAN DURING THESE TESTS. YOU GUYS WERE DOING SOMETHING RIGHT! I WONDER...

During the events they usually increase server capacity, or such, which costs more to run from what I've heard.

^^ This

It may be a while before the purchase by ms results in much change but you could still put your money where your mouth is rather than unpopular changes. I hear ms have a couple of datacenters. Maybe you could use some of that power.

It will be quite a while for sure. The deal isn’t anticipated to close until second half FY2021, so between April and September. Unless there were terms written into the contract about game improvements, which seems very unlikely, that’s the earliest we’d see any changes here related to Microsoft. Even then, ZOS has said, and Microsoft has agreed, that things will be run much the same afterwards. I’m not optimistic about what that means, especially for server upgrades. I can only hope I’m wrong.

Hexquisite

@ZOS_GinaBruno
I feel that the ZOS was disingenous by doing the Gobal AOE cooldowns during the IC event.

My groups play pretty hardcore on PC/NA Grey Host DC about 4 to 5 hours a night, BUT during the IC event we stayed in the sewers the whole time. This is true for every other group I know that plays in the same time slot. In Fact, the campaign was so empty of both DC and AD, that EP won by an astronomical amount of points, and many EP players campaign swapped because of it. I always check campaign populations, and never saw DC or AD above 1 bar during that first week, and the 2nd week sometimes they were at 2 bars.

The IC CP campaigns were poplocked a lot during the first week, and had ques. Yet, the lag we experienced was nothing like Prime Time Lag in Cyro.

Our group did try out these tests for about 15 mins, it was clunky, unenjoyable, and if you played Magic classes, the test really sucked for you. ..esp if you run in a smaller group, and soloing was pretty much out of the question unless it was just 1v1 against another magic user.

The only people I know who played in Cyro during those first 2 weeks did so on a Stam Blade. IN FACT PEOPLE MADE STAM BLADES just for those 2 weeks of tests.

Most of our group members haven't bothered to post here, because they are just super disappointed and feel ZOS won't listen, I myself included, this why this post is late--and then we saw the 2 new tests added, and it seems that the writing is on the wall, without even looking at the group limit size test results.

Many users seem to agree on the fact that we start to lag heavy when a group that spams repid regen and purge. Purge is an interesting skill because of the hierachy of the way it purges, and this is calculated everytime it is cast ( thus why we alll put 1 point in Siphon)....Its seems like these things could be changed, or even eliminted before ZOS ruins the awesome combat that is ESO.

After reading what the 2 new tests are, I am just resigned to the fact that those are the changes that are going to happen, and am deeply saddened that I will have to find a new game to play after 6 or so years.

NeillMcAttack

It literally mentions in the OP that Ravenwatch on PC/EU was pop-locked every evening throughout all testing, as it always is. Can people stop with the spam talking about the IC event. NA is not the only server.

Hexquisite

NeillMcAttack wrote: »

It literally mentions in the OP that Ravenwatch on PC/EU was pop-locked every evening throughout all testing, as it always is. Can people stop with the spam talking about the IC event. NA is not the only server.

But other people fromt hose servers have disagreed with that statement. Also that statement has loop holes, it isn't concise.

Maybe PC EU uses a dif type of server, or has a dif population cap than PC NA.

Jaraal

Orpheaus wrote: »

Additionally this question is directed at someone who has knowledge of this industry that I don't, is there a specific reason that ZOS can't upgrade their servers to create better performance? Or is this a situation where the hardware literally won't help and that's a pointless request?

The data has to go through a DDoS scrubber service (Akamai) that monitors and filters it before it gets to the servers. You can have the best servers in the world, but if you're bottlenecked at a remote firewall, it makes no difference. Reducing the amount of data that has to be calculated might be your only option.

Cyrodill ran much more smoothly with more players casting the same skills years ago, before Akamai.

virtus753

Jaraal wrote: »

Orpheaus wrote: »

Additionally this question is directed at someone who has knowledge of this industry that I don't, is there a specific reason that ZOS can't upgrade their servers to create better performance? Or is this a situation where the hardware literally won't help and that's a pointless request?

The data has to go through a DDoS scrubber service (Akamai) that monitors and filters it before it gets to the servers. You can have the best servers in the world, but if you're bottlenecked at a remote firewall, it makes no difference. Reducing the amount of data that has to be calculated might be your only option.

Cyrodill ran much more smoothly with more players casting the same skills years ago, before Akamai.

Do they just cancel Akamai for the week or two to get PvP working better for Midyear Mayhem, then?

Genuinely curious. Something improves substantially during PvP events, and I can’t imagine it’s less data, given how packed both the usual and ad hoc campaigns are. If the current performance is bottlenecked at Akamai, what happens during the event that improves it?

LarsS

It is true that there was less lag during the first week as well as it is during the mayhem events. The reason is however hardly because the server capacity is larger during mayhem nor as zos believes due to the global cool down. Both cases have one thing in common experienced pvp players with optimized builds (solo, small scale, or raids) were fewer on the standard servers. During mayhem, the queues are so long that the experienced players spread over all servers. During week 1 of the test the experienced players populated IC due to the IC event. Thus, in both cases the standard servers were populated largely by pve players with suboptimal builds for pvp, who wanted the double ap.

I am deeply worried that zos will use this flawed test as a base for radical changes of the eso battle system.

techyeshic

virtus753 wrote: »

Jaraal wrote: »

Orpheaus wrote: »

Additionally this question is directed at someone who has knowledge of this industry that I don't, is there a specific reason that ZOS can't upgrade their servers to create better performance? Or is this a situation where the hardware literally won't help and that's a pointless request?

The data has to go through a DDoS scrubber service (Akamai) that monitors and filters it before it gets to the servers. You can have the best servers in the world, but if you're bottlenecked at a remote firewall, it makes no difference. Reducing the amount of data that has to be calculated might be your only option.

Cyrodill ran much more smoothly with more players casting the same skills years ago, before Akamai.

Do they just cancel Akamai for the week or two to get PvP working better for Midyear Mayhem, then?

Genuinely curious. Something improves substantially during PvP events, and I can’t imagine it’s less data, given how packed both the usual and ad hoc campaigns are. If the current performance is bottlenecked at Akamai, what happens during the event that improves it?

If they got Akamai; they must have felt the need for security. I doubt we'd ever know if they turned it off officially. Just could guess if someone traces during the event