Damnationie wrote: »Can we just put this "we have to do this on live" nonsense to bed.
Outside the gaming industry, in the more general IT world, if you are doing performance investigations, live is the last place you go to do performance testing. I've over two decades in programming, a lot of spent as a troubleshooter of misbehaving systems and this idea you can't find performance problems on the test system is a warning sign I've come across a number of times. What it normally translates too is
- We don't want to spend money on a proper test system that is a proper scaled down replica of our live system.
- We don't know enough about how the system works to set up a proper test system.
- Proper testing automation and load testing tools have not been purchased for use by the developers.
- Trying to get people without the experience / expertise in trouble shooting to investigate complex issues because they won't spend for specialist help.
None of what ZOS is doing makes logical sense if they actually want to solve their issues. For a performance test to be any good you need to have control over the inputs and be able to repeat the exact same activity every time you adjust the code.
If the conditions are not identical than you cannot properly gauge the impact of a change. For example, currently there is a global GCD cool down in Cyrodiil. As a result a lot of people seem to be just removing those skills off their skill bars, maybe leaving just one. So the test is not actually seeing the impact of the changes to performance, rather the impact of people using different skills altogether. As they try the different scenarios they have suggested people will again alter their characters to suit. So each test will not be comparable to each other. Any conclusions they draw will be flawed, and no, you can't compensate for that. Been there, watched as a lot of people made bad decisions with confidence resulting in a major disaster.
What they should be doing is going and getting a load of bot scripts (If they don't have test automation tooling they'd work just as fine for this) and running them on their internal test system to replicate standard player activity. With monitoring you can see the impact each bot has and the impact of each of the skills. You can enable debug level tracing of the code and properly determine where the code bottlenecks are. Debug level logging enabled on most production applications would crash them. Its why you don't do this type of testing in live.
You can start with one simulated player and just ramp up the numbers each run watching performance. As you add simulated players you'd start to see the pain points and bottlenecks in the code. They may not get as dramatic as what is happening on live but it should be detectable. And from that you can figure where to look and once you make changes you can get a reliable read on whether they had any impact or not. What they are currently doing is hoping they can find a solution.
What they are currently doing falls into the less then professional end of the IT world.
WuChiWuGen wrote: »Theirs an easy solution just like another PVP MMO. Test how many players can be on a map at a time if it lags lower the number of players allowed on a map like and add a server. Around 80 each side has worked in other MMOs.
Oreyn_Bearclaw wrote: »Not gonna lie, I was pretty impressed with everything on Tuesday Primetime. It wasnt jammed, but there was a small queue and things seemed pretty crowded. Performance was great, gameplay was Meh, but expected. I felt like it was going to be a solid test.
By Wednesday, well as stated, it was a ghost town. Cant remember the last time EP was at 2-3 bars during primetime. I cant imagine they are getting useful data at this point.
They really need to fix the double AP. With no incentive, people aren't going to show up for this.
Non-CP EU is fully pop locked and alive, the test goes well, plenty of people
So it's ur not populated server issue. don't blame zos for NA players being passiveDevs should stop babysitting NA and focus on EU servers since its Cyro test is more representative due to populationPCNA Cyrodiil is dead even in prime time. No way you can do a reasonable test right now.
Non-CP EU is fully pop locked and alive, the test goes well, plenty of people
So it's ur not populated server issue. don't blame zos for NA players being passiveDevs should stop babysitting NA and focus on EU servers since its Cyro test is more representative due to populationPCNA Cyrodiil is dead even in prime time. No way you can do a reasonable test right now.
Damnationie wrote: »Can we just put this "we have to do this on live" nonsense to bed.
Outside the gaming industry, in the more general IT world, if you are doing performance investigations, live is the last place you go to do performance testing. I've over two decades in programming, a lot of spent as a troubleshooter of misbehaving systems and this idea you can't find performance problems on the test system is a warning sign I've come across a number of times. What it normally translates too is
- We don't want to spend money on a proper test system that is a proper scaled down replica of our live system.
- We don't know enough about how the system works to set up a proper test system.
- Proper testing automation and load testing tools have not been purchased for use by the developers.
- Trying to get people without the experience / expertise in trouble shooting to investigate complex issues because they won't spend for specialist help.
None of what ZOS is doing makes logical sense if they actually want to solve their issues. For a performance test to be any good you need to have control over the inputs and be able to repeat the exact same activity every time you adjust the code.
If the conditions are not identical than you cannot properly gauge the impact of a change. For example, currently there is a global GCD cool down in Cyrodiil. As a result a lot of people seem to be just removing those skills off their skill bars, maybe leaving just one. So the test is not actually seeing the impact of the changes to performance, rather the impact of people using different skills altogether. As they try the different scenarios they have suggested people will again alter their characters to suit. So each test will not be comparable to each other. Any conclusions they draw will be flawed, and no, you can't compensate for that. Been there, watched as a lot of people made bad decisions with confidence resulting in a major disaster.
What they should be doing is going and getting a load of bot scripts (If they don't have test automation tooling they'd work just as fine for this) and running them on their internal test system to replicate standard player activity. With monitoring you can see the impact each bot has and the impact of each of the skills. You can enable debug level tracing of the code and properly determine where the code bottlenecks are. Debug level logging enabled on most production applications would crash them. Its why you don't do this type of testing in live.
You can start with one simulated player and just ramp up the numbers each run watching performance. As you add simulated players you'd start to see the pain points and bottlenecks in the code. They may not get as dramatic as what is happening on live but it should be detectable. And from that you can figure where to look and once you make changes you can get a reliable read on whether they had any impact or not. What they are currently doing is hoping they can find a solution.
What they are currently doing falls into the less then professional end of the IT world.
It’s funny to see people telling others to just go in and test when they have no basic understanding of classes like templar and warden who the majority of their skills are aoe. If they are a healer then they are even more screwed because all their skills are aoe. Using 1 skill every 3 seconds is not a proper test and just results in people logging off since they are unable to play their character or roll.
If you’ve been here long enough you’d know that people have been spamming aoes for years when the population was much higher and the performance was much better.
All the performance issues were introduced from bad coding in patches that they ignored for years and now they are trying to butcher the game rather than invest the millions they've made to actually fix it, but yeah let’s blame the players.
Charge them with a Lotus Fan (wait three seconds), hit Whirling Blades (wait three seconds), proc Sheer Venom (wait thr..... never mind, they killed me already).
Everything is clearly working as intended:
Premise: AoE spam causes lag in PvP
Test setup: put cooldowns on AoEs, making most of them outright unusable or nonviable to use
Test result: a lot less lag in Cyrodiil due to low player population and almost no AoE use
Lapin_Logic wrote: »
trackdemon5512 wrote: »I mean if you protest by optioning to not participate for data they absolutely need then don’t [snip] and whine when changes you don’t like are made. ZOS is asking for a huge amount of feedback well beyond the Test Server and with the needs of large population of Live Players. They’re being open and forthright about it AND the various ways they’re trying to test/accommodate in order to accomplish the constantly maligned large-scale PVP.
The entire last year was ZOS attempting to make performance enhancements to prevent any of this happening and the developers have said that the improvements thus far haven’t been close to enough. So this is what we get.
All I know is that the Q1 update next year when these changes are likely to go into effect will be interesting and I’ll have a ton of popcorn for reading the forum complaints.
[Edited for Censor Bypass]
techyeshic wrote: »TineaCruris wrote: »....and those that are in cyrodiil have all changed their builds and bars to not use any aoe's.
I'm pretty sure thats the idea behind the tests. They are saying they think it is the AOEs doing calculations to determine who it hit in an area and by how much so they want to see if less AOEs are used, if performance gets better. I don't think they care much about how well it works on abilities as they have said they would have to review abilities once they coclude which tests work at reducing the lag.
The unfortunate thing is; this really came to a head at Update 25 with degrading performance leading up to that and for whatever reason; they can't seem to go back. That also was the patch where they took out a huge chunk of client file size
Correct. However, Zos is saying that it is that we can lay down AoE at a faster pace than intended and we used to do. Our sustain has improved over time. I guess Zos has grown tired of crimping our sustain. In the end, Zos is really just collecting information here. They are not testing to see what design will work best going forward as much as they are working to collect information and then they will devise a solution.
Templar jabs (magic and stamina) is worse sustainwise and slightly slower that it EVER was and it is still on the AOE cd (which basically breaks the class in Cyrodill).
I would suggest this is not about one skill. That can be altered easily.
Also, Zos could take a different path than what people think is obvious right now. The best we can do is provide well thought and constructive feedback from actual testing. I say that because much of the posts and threads have been complaining for the sake of complaining and pretty sure Zos can see right through that.
Lapin_Logic wrote: »
Except all gap closers are bugged.Lotus Fan, Stampede, Shield Charge, Leap, chains, etc.
They want people to test different stuff, but stuff doesn't work. Ironic, or par for the course?
TineaCruris wrote: »They can fix this WITHOUT giving us a cooldown. They proved that during the Midyear Mayem event. That is the problem.
The whole premise of the test is fatally flawed to begin with, as ZOS is literally asking players to test an entirely different game.Lapin_Logic wrote: »Everything is clearly working as intended:
Premise: AoE spam causes lag in PvP
Test setup: put cooldowns on AoEs, making most of them outright unusable or nonviable to use
Test result: a lot less lag in Cyrodiil due to low player population and almost no AoE use
If you can propose a "Test setup" that Both stops people using/spamming AOE but also gives "Test Result" Keeps Pop lock on all servers so we see the effect on lag with a high Cyrodiil population without endless rage threads by experts then let ZOS know, im sure they would love this golden key.
TineaCruris wrote: »techyeshic wrote: »TineaCruris wrote: »....and those that are in cyrodiil have all changed their builds and bars to not use any aoe's.
I'm pretty sure thats the idea behind the tests. They are saying they think it is the AOEs doing calculations to determine who it hit in an area and by how much so they want to see if less AOEs are used, if performance gets better. I don't think they care much about how well it works on abilities as they have said they would have to review abilities once they coclude which tests work at reducing the lag.
The unfortunate thing is; this really came to a head at Update 25 with degrading performance leading up to that and for whatever reason; they can't seem to go back. That also was the patch where they took out a huge chunk of client file size
Correct. However, Zos is saying that it is that we can lay down AoE at a faster pace than intended and we used to do. Our sustain has improved over time. I guess Zos has grown tired of crimping our sustain. In the end, Zos is really just collecting information here. They are not testing to see what design will work best going forward as much as they are working to collect information and then they will devise a solution.
Templar jabs (magic and stamina) is worse sustainwise and slightly slower that it EVER was and it is still on the AOE cd (which basically breaks the class in Cyrodill).
I would suggest this is not about one skill. That can be altered easily.
Also, Zos could take a different path than what people think is obvious right now. The best we can do is provide well thought and constructive feedback from actual testing. I say that because much of the posts and threads have been complaining for the sake of complaining and pretty sure Zos can see right through that.
The problem is the path they appear to be taking is to pass the performance debacle that ZOS created onto the shoulders of the loyal customers that have been pumping money into ZOS for six years now. They can fix this WITHOUT giving us a cooldown. They proved that during the Midyear Mayem event. That is the problem.
TineaCruris wrote: »techyeshic wrote: »TineaCruris wrote: »....and those that are in cyrodiil have all changed their builds and bars to not use any aoe's.
I'm pretty sure thats the idea behind the tests. They are saying they think it is the AOEs doing calculations to determine who it hit in an area and by how much so they want to see if less AOEs are used, if performance gets better. I don't think they care much about how well it works on abilities as they have said they would have to review abilities once they coclude which tests work at reducing the lag.
The unfortunate thing is; this really came to a head at Update 25 with degrading performance leading up to that and for whatever reason; they can't seem to go back. That also was the patch where they took out a huge chunk of client file size
Correct. However, Zos is saying that it is that we can lay down AoE at a faster pace than intended and we used to do. Our sustain has improved over time. I guess Zos has grown tired of crimping our sustain. In the end, Zos is really just collecting information here. They are not testing to see what design will work best going forward as much as they are working to collect information and then they will devise a solution.
Templar jabs (magic and stamina) is worse sustainwise and slightly slower that it EVER was and it is still on the AOE cd (which basically breaks the class in Cyrodill).
I would suggest this is not about one skill. That can be altered easily.
Also, Zos could take a different path than what people think is obvious right now. The best we can do is provide well thought and constructive feedback from actual testing. I say that because much of the posts and threads have been complaining for the sake of complaining and pretty sure Zos can see right through that.
The problem is the path they appear to be taking is to pass the performance debacle that ZOS created onto the shoulders of the loyal customers that have been pumping money into ZOS for six years now. They can fix this WITHOUT giving us a cooldown. They proved that during the Midyear Mayem event. That is the problem.
Evasion: Casting this ability and its morphs now requires that you wear 5 pieces of Medium Armor.