Jump to content
  • Sign Up

WvW Performance and Map Cap Testing


Cal Cohen.2358

Recommended Posts

@Zok.4956 said:

@"Cal Cohen.2358" said:Hi Everyone,

We know skill lag is an issue in WvW. We’re actively investigating performance improvements and optimizations, but this is not a quick process and is going to take some time. While these investigations are ongoing, we want to see if there are other opportunities for some short-term improvements.

We’re planning to run some tests to see what performance gains we can get by reducing the total number of players on the map. The current target for the initial test is with EU reset on this Friday, July 10, where we’ll be temporarily reducing the map caps in WvW for both NA and EU by 21 (7 per team) and monitoring the impact on performance over the next week. The key feedback that we’ll need from all of you is to get a sense of how the maps feel with fewer players. We still want WvW to feel like WvW, and we realize that any reduction to player counts will have a negative impact on the overall gameplay experience. We will be weighing that negative impact against the positives of any performance improvements to determine if it makes sense to move forward with lower map caps beyond this initial test. If we do move forward with lower map caps, we will also be keeping a close eye on queue times and will increase the number of tiers as needed.

This is not a long-term solution. Even if we find that reducing map caps leads to big performance improvements and the impact on gameplay is minimal, we will be continuing our investigations into performance with the plan of incrementally restoring map caps as improvements are made.

Hi CalIt was unplayable for me today off peak hours. We played during oceanic night time which started at 11am GMT and finished at 1pm GMT. My ping seemed to be quite ordinary (300-400 ave) but I couldn't activate skills and the screen was freezing quite a lot. It was really impossible to play. We are only ~ 20 men squad.I have not experienced this issue for many years. As oceanic player I have always been playing with higher ping but it is manageable as long as the ping is consistent without spike. But what happen today it wasn't spike or high ping, its rubber banding and delayed skills activation. When I try to turn the view around it seems like it takes longer for one frame at a time delay.. its horrible. Instead of sending a support ticket I think a report here may help to provide you a feedback for the server fix.FYI - this is FoW server linked with GunnarI am calling from Australia.I have a short video of the frame lag here I can send it over if it helps.

Thanks.RegardsTalindra

Rubberbanding with delayed skill activation is network latency and/or packet loss, not the server struggling to calculate everyone's skill usage as this topic is about. Network issues can occur at any point between you and the server. It is not necessarily something fixable by Anet. The only thing they can do is ensure that their network and upstream connection are copasetic.

It could be a network problem or it could be server a problem (server not able to push its message queues fast enough or looses messages internally). We (as players) can not ne sure about the root problem. But I do know that even rubberbanding can happen to several players at exactely the same time (some time ago, not recently, it happend to all players in my squad and shortly after this the map crashed and we were all disconnected).

simplified rule of thumb: If it happens only to one player -> player problem, if it happens to several players -> Anet problem.

What I am wondering: Skill lag is not a new problem in WvW and it did happen before Anet moved to cheaper Amazon servers. What is now the actual motivation to reduce skill lag in WvW? Is it something that happens with the work for the next expansion and WvW is just the test-case? Or did someone higher in the food-chain gave the order to fix this? Or is it more of a pet-project of 1-2 devs? Or is this only a kind of proof that they are doing something for WvW (we have reduced the map cap as a temporary fix until we have a real solution, but only after we finished the alliance-system)?

I do not know. But I do know that Anet could have the server performance data/metrics since months and they could replay the load patterns with automatic-test clients if this problem were important for Anet.

The server not being able to push its message queues fast enough is exactly what we're talking about in this thread. In a large three way battle where the network is completely fine, the server can get overwhelmed by the skill calculations. Condition damage really affects this because the server calculates damage for every "tick" and the variables change every "tick". The player experience of that is pushing buttons and they don't activate. That's been known from hundreds of past threads on this topic by both players and devs. These calculations are what is being referenced when Cal talks about "optimizations". We went through this before after HoT was released because HoT lifted the 25 stack cap on many conditions and skill lag really suffered until Anet did some "optimizations" to the biggest offending skills.

Rubberbanding doesn't really have anything to do with the server because the server doesn't enforce player position - it only reports player's last known position to other clients (Edit: I'll add that rubberbanding tends to be what other players see, not what the person who is rubberbanding sees). If it did, there wouldn't be teleport hacks. It's the primary reason the orb mechanic was removed from WvW.

Link to comment
Share on other sites

https://massivelyop.com/2020/07/11/guild-wars-2-looks-for-short-term-wvw-performance-improvements/For you guys to see what I'm talking about. They will reduce population in meantime.

So this is recording I finally got to capture after remembering to install the dx9 to dx12 wrapper mod. I went from 20/25fps to 70/90fps after installing the mod.

Doesn't it all go back to the fact, whether they want to admit it, Guild Wars 2's game engine needs some updating. Especially to match todays hardware.I like that WvW finally got attention but, same problem is found on PvE.

What's your thoughts? If Anet did a Fundme/backers fund to help start the new game engine build. Would you help them?

What's your thoughts on their current mitigation for the problem?

Link to comment
Share on other sites

You manage to link an article that talks about an issue, but then you talk about a completely separate issue that has nothing to do with exactly that.

There is also this thread for further reading for you: https://en-forum.guildwars2.com/discussion/109755/wvw-performance-and-map-cap-testing/p1

Skill lag literally comes from the game server not being able to handle all the calculations and packages, the time it takes for your skills to activate and be registered as such. Lowering player count would be an obvious fix to that and they are just testing whether that has the intended effect or not, while also exploring other possibilities.

Link to comment
Share on other sites

So like aside from like two people who said there was a lot less lag on EU, do we have any demonstrable metrics about how performance was on reset, or are people still going to sit here and speculate/whine?

I get we needed a fix to this a really long time ago, but seriously, progress being made is a huge freaking deal, and might give us some more insights on how to fix it.

My question is whether or not EU reset had 3-way map-sized blob fighting during reset or not, because if it didn't, the metrics aren't intrinsically as valuable as we might think, considering the lag only really seems to occur (at least on NA) when all three servers are involved and bringing huge numbers. Even 50v50 from two servers rarely seems to induce the skill lag issues.

How EU's WvW metas stack up to others is also a pretty important facet to determining data validity, too, as there is a trend and a previous dev post from some years ago attributing lag to condi and boon application/ticks/clearing in mass quantities, so a power-centric meta may have fewer issues than a condi-centric one.

If you're all gonna complain, at the very least be somewhat knowledgeable and ask important questions, first.

Link to comment
Share on other sites

@"Eventine.8024" said:And for everyone that is posting: please write only if you understand the difference between skill lag and fps-based lag, and if you tested the game with a basic ping <40

If "skill lag" is pressing a skill attack and it takes 3-5seconds to actually initiate, while the other two skills you had already activated are also lagging behind.Than this occured in PVE as well. So it technically wouldn't be WvW specific. WvW would just make it easier to overload the network. I can play with 150-200ping and not experience this at times.

@GalvinWolf.3284 said:MAKE EoTM VIABLE AGAIN, Treat EoTM as if it were a 5th BL that people can fight over,OR get rid of EoTM so you can impalement 'OverFlow' maps to EBG and all the BLs

it's alos weird that you would sstress test lowering the cap by7, why not go lower 10 or 20 players, then add to the cap to see where it throttles

Good point.

Link to comment
Share on other sites

yeap, like @Blocki.4931 said, it's a server side issue that is causing skill lag and character location de-sync, your client's fps have nothing to do with how much network lags you are getting

the fundamental problem comes from the smoothness the combat and the access to quickness and alacrity = more skill spamming per sec from players that the server needs to crunch through

Link to comment
Share on other sites

@Cal Cohen.2358 said:Hi Everyone,

We know skill lag is an issue in WvW. We’re actively investigating performance improvements and optimizations, but this is not a quick process and is going to take some time. While these investigations are ongoing, we want to see if there are other opportunities for some short-term improvements.

We’re planning to run some tests to see what performance gains we can get by reducing the total number of players on the map. The current target for the initial test is with EU reset on this Friday, July 10, where we’ll be temporarily reducing the map caps in WvW for both NA and EU by 21 (7 per team) and monitoring the impact on performance over the next week. The key feedback that we’ll need from all of you is to get a sense of how the maps feel with fewer players. We still want WvW to feel like WvW, and we realize that any reduction to player counts will have a negative impact on the overall gameplay experience. We will be weighing that negative impact against the positives of any performance improvements to determine if it makes sense to move forward with lower map caps beyond this initial test. If we do move forward with lower map caps, we will also be keeping a close eye on queue times and will increase the number of tiers as needed.

This is not a long-term solution. Even if we find that reducing map caps leads to big performance improvements and the impact on gameplay is minimal, we will be continuing our investigations into performance with the plan of incrementally restoring map caps as improvements are made.

Would data compression help? I know it's different but kinda like how they went from video compression from h264 to h265 (1.7gb at 4k to 453mb at 4k both at 60fps)H266 just got introduced and I think is half of h265.

So would engineering a even superior file compression software(than what is currently being used) help not just to mitigate the issue but speed up actual data transfer. Making communication quicker between the servers and clients? Increasing the time the cores or GPU are free to make more calculations.

What about a more controversial tactic. Why not separate the WvW mode from the actual game itself? Make it into it's own game engine optimized for that exact setup. This would help separate the load from the PVE/PvP servers right? Same thing could be done for PvP. Anything to reduce the load right?

Link to comment
Share on other sites

@jishi.7568 said:

@Cal Cohen.2358 said:Hi Everyone,

We know skill lag is an issue in WvW. We’re actively investigating performance improvements and optimizations, but this is not a quick process and is going to take some time. While these investigations are ongoing, we want to see if there are other opportunities for some short-term improvements.

We’re planning to run some tests to see what performance gains we can get by reducing the total number of players on the map. The current target for the initial test is with EU reset on this Friday, July 10, where we’ll be temporarily reducing the map caps in WvW for both NA and EU by 21 (7 per team) and monitoring the impact on performance over the next week. The key feedback that we’ll need from all of you is to get a sense of how the maps feel with fewer players. We still want WvW to feel like WvW, and we realize that any reduction to player counts will have a negative impact on the overall gameplay experience. We will be weighing that negative impact against the positives of any performance improvements to determine if it makes sense to move forward with lower map caps beyond this initial test. If we do move forward with lower map caps, we will also be keeping a close eye on queue times and will increase the number of tiers as needed.

This is not a long-term solution. Even if we find that reducing map caps leads to big performance improvements and the impact on gameplay is minimal, we will be continuing our investigations into performance with the plan of incrementally restoring map caps as improvements are made.

Would data compression help? I know it's different but kinda like how they went from video compression from h264 to h265 (1.7gb at 4k to 453mb at 4k both at 60fps)H266 just got introduced and I think is half of h265.

Data compression means heavier CPU load to unpack it and the main suspect for skill lag is server CPU load, not bandwidth so yeah... no... Not to mention you kind of need the information being sent or you wouldnt program the game to send it.

But I suppose if you want players to move in hexagonal turn based combat we could turn alot of those floating point values into integers!

Link to comment
Share on other sites

@Helicity.3416 said:

@Helicity.3416 said:Nice.

Even longer queues for what will
almost certainly
not give a measurable performance increase.

so you would rather them do nothing and people to continue to complain about it?

I would rather they actually look at what causes it (it's boons and condis) than attempt the lowest effort possible "fix" (really, it's boons and condis) to attempt to make things work better (it's the boons and condis, look at the boons and condis) yes.

It's not boons and condis. It's the constant back and forth from the server, where are you? I'm here. What are you doing now? I'm doing this.

You can tell when a blob is coming well before you can see them just by the lag and they will have minimal boons on them and there are no condis flying around.

Link to comment
Share on other sites

@Chaba.5410 said:

@Zok.4956 said:

@"Cal Cohen.2358" said:Hi Everyone,

We know skill lag is an issue in WvW. We’re actively investigating performance improvements and optimizations, but this is not a quick process and is going to take some time. While these investigations are ongoing, we want to see if there are other opportunities for some short-term improvements.

We’re planning to run some tests to see what performance gains we can get by reducing the total number of players on the map. The current target for the initial test is with EU reset on this Friday, July 10, where we’ll be temporarily reducing the map caps in WvW for both NA and EU by 21 (7 per team) and monitoring the impact on performance over the next week. The key feedback that we’ll need from all of you is to get a sense of how the maps feel with fewer players. We still want WvW to feel like WvW, and we realize that any reduction to player counts will have a negative impact on the overall gameplay experience. We will be weighing that negative impact against the positives of any performance improvements to determine if it makes sense to move forward with lower map caps beyond this initial test. If we do move forward with lower map caps, we will also be keeping a close eye on queue times and will increase the number of tiers as needed.

This is not a long-term solution. Even if we find that reducing map caps leads to big performance improvements and the impact on gameplay is minimal, we will be continuing our investigations into performance with the plan of incrementally restoring map caps as improvements are made.

Hi CalIt was unplayable for me today off peak hours. We played during oceanic night time which started at 11am GMT and finished at 1pm GMT. My ping seemed to be quite ordinary (300-400 ave) but I couldn't activate skills and the screen was freezing quite a lot. It was really impossible to play. We are only ~ 20 men squad.I have not experienced this issue for many years. As oceanic player I have always been playing with higher ping but it is manageable as long as the ping is consistent without spike. But what happen today it wasn't spike or high ping, its rubber banding and delayed skills activation. When I try to turn the view around it seems like it takes longer for one frame at a time delay.. its horrible. Instead of sending a support ticket I think a report here may help to provide you a feedback for the server fix.FYI - this is FoW server linked with GunnarI am calling from Australia.I have a short video of the frame lag here I can send it over if it helps.

Thanks.RegardsTalindra

Rubberbanding with delayed skill activation is network latency and/or packet loss, not the server struggling to calculate everyone's skill usage as this topic is about. Network issues can occur at any point between you and the server. It is not necessarily something fixable by Anet. The only thing they can do is ensure that their network and upstream connection are copasetic.

It could be a network problem or it could be server a problem (server not able to push its message queues fast enough or looses messages internally). We (as players) can not ne sure about the root problem. But I do know that even rubberbanding can happen to several players at exactely the same time (some time ago, not recently, it happend to all players in my squad and shortly after this the map crashed and we were all disconnected).

simplified rule of thumb: If it happens only to one player -> player problem, if it happens to several players -> Anet problem.

Rubberbanding doesn't really have anything to do with the server because the server doesn't enforce player position - it only reports player's last known position to other clients (Edit: I'll add that rubberbanding tends to be what other players see, not what the person who is rubberbanding sees). If it did, there wouldn't be teleport hacks. It's the primary reason the orb mechanic was removed from WvW.

Rubberbanding is when the position(s) of player (s) are out-of-sync in the client and in the server. This position-out-of-sync can happenbecause of a network problem (connection hangs/waits for retransmit because of a packet loss of the TCP-connection) but it can also happen because of a server problem (and sometimes there is a server crash / map wide disconnect later)

When rubberbanding happens you sometimes also have skill-lag (no skill activation). Because the activation of skills is also reported back from the server to the client and when the connection hangs, this information can also not be transmitted.

Yes I know, that rubberbanding is not the initial topic of this thread. That part of my response was to correct your wrong statement, that rubberbanding is always a network problem. It is a communication problem (no data gets transmitted) which is often because of a network problem, but it is not always the result of a network problem.

"skill lag" is when there is a noticible (for the player) delay betwen the pressed skill button in the client and the "skill activation response" from the server.The reason could be long round-trip-times (network latency in both directions), network congestion or server overload/congestion.

Sometimes, it first starts with skill lag (short delays/ small connection waits) and later there is rubberbanding (long delays, connection waits long time and sometimes resets)

Because skill lag is a Anet/server problem when it happens to several players at the same time (rule of thumb) Anet could see the reasons why the servers are in congestion/overload if they look at their server monitoring. They could easily just look at their own software and see what parts of the software are using the most ressources (network bandwidth, message processing capacity, CPU, RAM) and where the bottlenecks are.

So there would be no need to use WvW players as (unreliable) guinea pigs.

Link to comment
Share on other sites

@Zok.4956 said:

@Zok.4956 said:

@"Cal Cohen.2358" said:Hi Everyone,

We know skill lag is an issue in WvW. We’re actively investigating performance improvements and optimizations, but this is not a quick process and is going to take some time. While these investigations are ongoing, we want to see if there are other opportunities for some short-term improvements.

We’re planning to run some tests to see what performance gains we can get by reducing the total number of players on the map. The current target for the initial test is with EU reset on this Friday, July 10, where we’ll be temporarily reducing the map caps in WvW for both NA and EU by 21 (7 per team) and monitoring the impact on performance over the next week. The key feedback that we’ll need from all of you is to get a sense of how the maps feel with fewer players. We still want WvW to feel like WvW, and we realize that any reduction to player counts will have a negative impact on the overall gameplay experience. We will be weighing that negative impact against the positives of any performance improvements to determine if it makes sense to move forward with lower map caps beyond this initial test. If we do move forward with lower map caps, we will also be keeping a close eye on queue times and will increase the number of tiers as needed.

This is not a long-term solution. Even if we find that reducing map caps leads to big performance improvements and the impact on gameplay is minimal, we will be continuing our investigations into performance with the plan of incrementally restoring map caps as improvements are made.

Hi CalIt was unplayable for me today off peak hours. We played during oceanic night time which started at 11am GMT and finished at 1pm GMT. My ping seemed to be quite ordinary (300-400 ave) but I couldn't activate skills and the screen was freezing quite a lot. It was really impossible to play. We are only ~ 20 men squad.I have not experienced this issue for many years. As oceanic player I have always been playing with higher ping but it is manageable as long as the ping is consistent without spike. But what happen today it wasn't spike or high ping, its rubber banding and delayed skills activation. When I try to turn the view around it seems like it takes longer for one frame at a time delay.. its horrible. Instead of sending a support ticket I think a report here may help to provide you a feedback for the server fix.FYI - this is FoW server linked with GunnarI am calling from Australia.I have a short video of the frame lag here I can send it over if it helps.

Thanks.RegardsTalindra

Rubberbanding with delayed skill activation is network latency and/or packet loss, not the server struggling to calculate everyone's skill usage as this topic is about. Network issues can occur at any point between you and the server. It is not necessarily something fixable by Anet. The only thing they can do is ensure that their network and upstream connection are copasetic.

It could be a network problem or it could be server a problem (server not able to push its message queues fast enough or looses messages internally). We (as players) can not ne sure about the root problem. But I do know that even rubberbanding can happen to several players at exactely the same time (some time ago, not recently, it happend to all players in my squad and shortly after this the map crashed and we were all disconnected).

simplified rule of thumb: If it happens only to one player -> player problem, if it happens to several players -> Anet problem.

Rubberbanding doesn't really have anything to do with the server because the server doesn't enforce player position - it only reports player's last known position to other clients (Edit: I'll add that rubberbanding tends to be what other players see, not what the person who is rubberbanding sees). If it did, there wouldn't be teleport hacks. It's the primary reason the orb mechanic was removed from WvW.

Rubberbanding is when the position(s) of player (s) are out-of-sync in the client and in the server. This position-out-of-sync can happenbecause of a network problem (connection hangs/waits for retransmit because of a packet loss of the TCP-connection) but it can also happen because of a server problem (and sometimes there is a server crash / map wide disconnect later)

When rubberbanding happens you sometimes also have skill-lag (no skill activation). Because the activation of skills is also reported back from the server to the client and when the connection hangs, this information can also not be transmitted.

Yes I know, that rubberbanding is not the initial topic of this thread. That part of my response was to correct your wrong statement, that rubberbanding is always a network problem. It is a communication problem (no data gets transmitted) which is often because of a network problem, but it is not always the result of a network problem.

"skill lag" is when there is a noticible (for the player) delay betwen the pressed skill button in the client and the "skill activation response" from the server.The reason could be long round-trip-times (network latency in both directions), network congestion or server overload/congestion.

Sometimes, it first starts with skill lag (short delays/ small connection waits) and later there is rubberbanding (long delays, connection waits long time and sometimes resets)

Because skill lag is a Anet/server problem when it happens to several players at the same time (rule of thumb) Anet could see the reasons why the servers are in congestion/overload if they look at their server monitoring. They could easily just look at their own software and see what parts of the software are using the most ressources (network bandwidth, message processing capacity, CPU, RAM) and where the bottlenecks are.

So there would be no need to use WvW players as (unreliable) guinea pigs.

You still have the incorrect idea about rubberbanding because you don't seem to understand how this game uses the position data. It is an asynchronous positioning system. The client reports its position to the server. The server then reports the last known position of that client to other clients. The server does not enforce client position. There is no "in sync" positioning. That's also why teleport hacks work in this game. Your description of rubberbanding sounds like other games with collision detection enforced by the server.

When you see a few people rubberbanding, that is their client's hiccup of reports to the server and not the server being unable to transmit. The server is transmitting what it knows about that player to you which then gets rendered as rubberbanding. The transmission issue is short connection delays with that other client to the server.

What one sees when not receiving data from the server during a long connection delay is all characters all moving off in the last direction they were headed as your client attempts a best guess render of position for them all. Then suddenly everything speeds up to catch up if the connection recovers or you timeout and disconnect. That is not rubberbanding at all.

Also, your rule of thumb is just way too rudimentary to pursue root cause in a large interconnected system. You're making an assumption that everyone's packets do not ever travel through the same routes so therefore it's the server, which just isn't true when considering midstream routers.

Ultimately servers don't care about nor monitor the routes that their sent packets take to clients. They only care about the upstream connection, which must be fine since they're lowering map caps rather than doing network stuff. Anet cannot see the network issues of clients. All they see is connections in a TIME_WAIT state.

Link to comment
Share on other sites

Reduce the number of targets each weapon skill/utility skill hits (including heals) in WvW only.

Skills that hit 1 remain at 1Skills that hit 3 reduce to 2Skills that hit 4 reduce to 3Skills that hit 5 reduce to 4etc.

See how it goes.

Fairly sure you (The Developers) mentioned a while back that one of the biggest problem facing the game was reducing the load on the servers trying to calculate all the damage/what health the player should be at. Reducing the calculations needed may just make the game a little more playable. It might make fights last a little longer when it comes to more than 1v1, but that might actually be a little more enjoyable rather than being pummelled instantly into the ground.

Link to comment
Share on other sites

@"Heimskarl Ashfiend.9582" said:

I would rather they actually look at what causes it (it's boons and condis) than attempt the lowest effort possible "fix" (really, it's boons and condis) to attempt to make things work better (it's the boons and condis, look at the boons and condis) yes.

It's not boons and condis. It's the constant back and forth from the server, where are you? I'm here. What are you doing now? I'm doing this.

You can tell when a blob is coming well before you can see them just by the lag and they will have minimal boons on them and there are no condis flying around.

I disagree. Yes, when the framerate drops, that's not boons and condis. That's just the client needing to load in and handle additional models and effects, so you'll likely always sense a zerg nearby.

But the skill lag only occurs, when the zerg is actually fighting. And it only occurs, if one or both of the zergs put out a significant number of AoEs, Condis and Boons.

Link to comment
Share on other sites

Professional game developers with a publisher the size of NCSoft do not need to go to their communities for donations to fund a renderer update. If they see value in it, they'll assign or hire people for the job. If they don't, they won't. At this stage, if they're not already working on a renderer for DX11/12/Vulkan or planning such an upgrade, it's because they intend to retire the engine before the investment would pay off.

The situation in WvW is rather different and isn't strongly related to client-side hardware. The original design for GW was based around 1-16 player instances, and limited skill interactions. GW2 took that engine and expanded the number of interactions substantially, so it's possible that some efficiency problems arose that would never have mattered in the more limited scope that Mike O'Brien had envisioned back in GW1.

However it's also possible that, at the upper bounds of WvW combat, there are so many effects interacting with each other that it would slow down any server, no matter how fast the code runs.

The only solution to that problem is to reduce the number of calculations, but there are several ways you can do that. Lowering player limits is one, and probably the easiest. In order of increasing difficulty and efficacy, other options are: a balance pass on skills to reduce the number of targets etc.; optimizing skill interactions at the code level to reduce the number of calculations performed when interactions happen; or rewriting a significant part of the engine to approximate most of the calculations under high load, when a loss of accuracy won't be noticable.

(I'm not holding my breath on that last one. I'd expect something like PoE or D4 to solve that first.)

Link to comment
Share on other sites

@Helicity.3416 said:

@Helicity.3416 said:Nice.

Even longer queues for what will
almost certainly
not give a measurable performance increase.

so you would rather them do nothing and people to continue to complain about it?

I would rather they actually look at what causes it (it's boons and condis) than attempt the lowest effort possible "fix" (really, it's boons and condis) to attempt to make things work better (it's the boons and condis, look at the boons and condis) yes.

This right here is it.

WvW always had some lag, but it got noticeably worse when they removed the boon and condi caps. While that is fine for encouraging styles of play in PvE, it has no place in a competitive mode. As much as people want to claim they want ALL kinds of builds, the simple fact is the more options, the harder it is to balance a PvP mode.

@nthmetal.9652 said:

@"Heimskarl Ashfiend.9582" said:

I would rather they actually look at what causes it (it's boons and condis) than attempt the lowest effort possible "fix" (really, it's boons and condis) to attempt to make things work better (it's the boons and condis, look at the boons and condis) yes.

It's not boons and condis. It's the constant back and forth from the server, where are you? I'm here. What are you doing now? I'm doing this.

You can tell when a blob is coming well before you can see them just by the lag and they will have minimal boons on them and there are no condis flying around.

I disagree. Yes, when the framerate drops, that's not boons and condis. That's just the client needing to load in and handle additional models and effects, so you'll likely always sense a zerg nearby.

But the skill lag only occurs, when the zerg is actually fighting. And it only occurs, if one or both of the zergs put out a significant number of AoEs, Condis and Boons.

Amen.

Link to comment
Share on other sites

@"jishi.7568" said:What's your thoughts? If Anet did a Fundme/backers fund to help start the new game engine build. Would you help them?

What's your thoughts on their current mitigation for the problem?

They're not going to invest any more than the bare minimum to keep their low-expectation playerbase logging in. And that's not much. Are you asking if players would be willing to donate their own money (like a charity) to "help" Anet rebuild their game engine?? I respectfully decine.?

Link to comment
Share on other sites

@ANet

Do we have any ANet approved "in-game tool" that would be able to report a game client's "status" if a player decides & wants to report poor performance?

If a player who installed the base game without any client "mods" & sets everything to low or disabled can choose to let ANet know the details of what might be causing their poor performance.

Or, ANet might be able to hand pick players to perhaps participate in a "limited project" determining poor performance using this "in-game tool".

Need to be careful that this "in-game tool" needs to be limited in scope to reduce the amount of exposure.


This "in-game tool" reporting could be used to diagnose if it's client-based or server-based.

Players get to figure out that maybe an unsupported "client mod" might be impacting performance, or ANet gets to observe patterns of client performance while monitoring server tweaks.

Need to be careful that this "in-game" tool is not abused to compromise the integrity of a players computer, or ANet's servers.

Link to comment
Share on other sites

For the technically minded: I thought the player amount and required computations would scale by square (power of 2), but it seems to grow cubic (power of 3) when 3 blobs are near each other.

Let me illustrate:if the previous map cap for borderlands was 50 players, the worst case for computations is 50 3 = 125000 (imaginary computational units)if the new cap for borderland maps is 7 players less or 43 players, the worst case for computations is 43 3 = 79507

Let's imagine the servers can do maximum of 100 000 computational units per frame of calculation. With the previous map cap of 50, this would mean constantly going above what the servers can process and causing massive lags and drop outs. Reducing just 7 players from each side would reduce the amount of computations needed by about 37 % and the amount of amount would be always 20 % lower than the server capacity. So there would be rarely ever any noticable lag.

I might be wrong in my calculations as I have no insider information of the algorithms and data structures used.

Link to comment
Share on other sites

@Diku.2546 said:@ANet

Do we have any ANet approved "in-game tool" that would be able to report a game client's "status" if a player decides & wants to report poor performance?

If a player who installed the base game without any client "mods" & sets everything to low or disabled can choose to let ANet know the details of what might be causing their poor performance.

Or, ANet might be able to hand pick players to perhaps participate in a "limited project" determining poor performance using this "in-game tool".

Need to be careful that this "in-game tool" needs to be limited in scope to reduce the amount of exposure.


This "in-game tool" reporting could be used to diagnose if it's client-based or server-based.

Players get to figure out that maybe an unsupported "client mod" might be impacting performance, or ANet gets to observe patterns of client performance while monitoring server tweaks.

Need to be careful that this "in-game" tool is not abused to compromise the integrity of a players computer, or ANet's servers.

https://help.guildwars2.com/hc/en-us/articles/231273708-Collecting-Diagnostic-Information-for-Support-Reporting-Tech-Issues-

(see -diag command line parameter for example)

Link to comment
Share on other sites

I'll just add this old discussion here due to lots of technical information: https://forum-en.gw2archive.eu/forum/game/gw2/Official-state-of-skill-lag-and-server-optimizations/first

Also: "As far as skills just not executing (noticed some people claiming utilities are more susceptible to this), its mostly just a race-condition as far as processing on the server. You’ll notice that your auto-attack skill seems to process more reliably than other skills. This is mostly due to the fact that we process things like auto-attack timers before player input. Obviously that sounds like a bug (and honestly now that I think about it, I want to look into doing something about it), but the reality is that under normal circumstances, the player input would process before the auto-attack timer triggers. Something you can try to verify this is disabling your auto-attack and see if your other skills become more responsive."

TL;DR: Turn off your auto-attack! Haha!

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...