Can we fix the lag instead of just "addressing" it?

Dropping swarm pets does not make laggy events/zones any less laggy. We've tested that multiple times. It is 100% the casters and all of their dozens of passive procs, combined with sheer number of people logged into a server during prime time.

I think there have been plenty of complaints about each of those forms of lag / game disruption, this is just a separate thread.

To clarify, some are making the claim that the calculation congestion/overload occurs across all serverwide instances (maybe EQ2 is even involved). In other words, even if one guild stopped all swarm pets (and limited procs), congestion is created by other guilds. Lag is behaving as a sort of cascade (snow-ball effect across multiple instances) or contagion. There's some assumptions here for sure. With this assumption, the treatment would need to be applied server-wide (whether that means testing an experimental raid or fully revamping combat - I suggest the former first for economy).

Seems just like the claim that making a single raid without any pets is going to have any noticeable impact.

Again, I would target the new expansion's T1 raid for maximum effect. (Raiders would less likely to be dispersed among T1, T2, and so forth). Ultimately, Darkpaw would draft the experimental design.

And again a single raid isn't going to give much data or have much of an impact. Just because you are targeting a raid in the new expansion doesn't mean that raiders are still not hitting other raids. There is a decent amount of raiders on TLP servers that will not be hitting that raid and will be raiding other things not to mention live servers still on older content.

Do guilds on TLP's in earlier expansions ever see raid lag with 72 ppl in instance but much less 'stuff' being calculated/going on?

Very seldomly, typically if a TLP raid sees lag it is due to just very high server load during primetime. There are occasions where you get 200+ players in the same zone attacking the same NPC where input starts to get super delayed and NPCs might skip swings and not fire abilities as often as they are supposed to.

Frank or Sieger could probably speak more about Selo's experience with lag as the server approached more recent expansions.

The presumption is that NPC Pathing is a lot more computationally expensive than adding a couple numbers.

Pathing is one of those 'hard' problems for computers.

https://en.wikipedia.org/wiki/A*_search_algorithm

"A* (pronounced "A-star") is a graph traversal and path search algorithm, which is used in many fields of computer science due to its completeness, optimality, and optimal efficiency.[1] One major practical drawback is its space complexity, as it stores all generated nodes in memory."

Maybe now that EQ is 64 bit, they can tweak pathing to better utilize the available RAM. The problem with using more RAM is more CPU cache misses... so tough to optimize.

Which is why I suggested replacing spawning 3d models for many pets with a single nimbus on the mob. Far less graphic footprint and far fewer calculations.

And it is a lot more then a couple of numbers being computed when you compare all the swings/spells/procs that are being calculated every second.

Some people say swarm pets create lag. Some say they don't.

Some say procs create lag. Some say they don't.

Some say prime time creates lag. Some say they don't.

I think it would be nice, instead of reading another six pages of a back and forth between two people, to have some red name come out and talk about some ideas and suggestions on how to fix lag. And to come out and specifically state what they know causes lag, and what they are doing about it.

No.

22 year magician. There is no reason to change Host of Elements in its current state. It has practical uses in the group game, and should be left alone.

What I suggest is just not allowing Host of Elements to be cast in a raid zone. And Bard Song of Stone. And any other class swarm AA that is similar. Leave them alone so they can be used in group. Just disable them in raid zones.

And I like having the different levels of Of Many. It at least adds.... some flavor. It is not just some generic DD. And if it just went off for max damage every time, that would make group mages do even more damage. As it stands, the solo mage generally cannot get to max damage of Of Many unless they pull out Host AA. And on a raid, even taking away all the AA swarm pets, majority of raids still are getting 15 pets on a mob. So no, leave the different levels of Of Many as is.

So what happens if it is discovered that all pets cause lag? You just destroyed magicians as a worthwhile raid class. Keep one mage on a raid for mod rods, rest of you can stay home.

No. Raids that prohibit ALL pets is not a good decision at all. You pulled out the "Do not call for nerfs" card a few pages back, and then sit there and want raids to disallow ALL pets? Unbelievable.

Did you read my posts?

I said the anti-pet mechanic would exist for 1 raid to test the widest margin of the hypothesis that pets are a major contributor to lag. I also said I support streamlining pets (compensatory adjustments) if this is confirmed. I prefer a test before any changes are made because time/labor-intensive change would be futile or in vain if something more fundamental such as hardware or code is the actual cause of lag.

Do you realize how much damage Host does during a burn? Just curious.

No. Are we talking 4% of a magician's parse? 6%? Or something more like 0.2%?

From the top of my memory, and it is foggy at the moment... I think I did 1.17 million on Aten and like 1.9 something on Shei. Don't quote me.

Lot's of good suggestions have been tossed around. For years. I wonder when something will actually get tested?

It's not like we're talking rocket science here. It's an ancient game! Turn this knob, see what happens. Put that one back and twist a different dial. Repeat until you learn something.

It's not like we won't have hundreds of players willing to try out new things in a massively multiplayer mode in just a few weeks. BETA could also be used to test lag reduction strategies the same time we are trying out event mechanics.

Here's a (not serious) idea but given some of the comments, maybe some will find it interesting. What about limiting the number of guilds raiding (or creating instances of a particular zone) to some number, say 5 or 10 at any one time. Or have a shared calendar of which nights guilds can raid. Would this help the server load? Would be interesting to test if guilds were to volunteer to move their Sunday raid to another night since Sunday seems to be when a lot of guilds raid.

But seriously, why not have the swarm pet ability just be a buff applied to the person, like COP or auspice, etc. Then no creation of flappies, no pathing occurs, and no melee swings from them. I started this idea with just thinking why do mage's swarm pets run and attack a mob, instead maybe they should swarm around the mage and boost their casting for the time they are alive, which then lead me to why not just a buff applied to the caster. Imagine if COP and auspice generated 10 pets for every person on which they land!

This idea could be similarly applied to procs which could instead just boost the person's defense, or casting power or whatever the proc gives you.

I would try to run raid instances run on separate beefier hardware.
DP dev, Have you heard of on demand infrastructure?