Because of how a Warlock plays, we're arguably one of the most spam-y characters in the game, and our most sensitive time to parse is our most spam-y. When focused casting time comes around, we're spamming our charms, Ancient Fury, sending a macro command to tell our chanters to mimic us, and then blowing through 3 insta cast spells and a bunch of really fast-cast spells, all while lining that up so our best stuff all lands during Time Warp. Latency loss absolutely cripples a Warlock, arguably moreso than any other class.
No. Combat arts have a 1/2 s base cast time. Mage spells have 0.5~3 s base cast time. Adding a flat delay across the board hinders CAs far more than spells because of their frequency. Yes, three spells are instant @ max increments, and yes focus casting//myth clicky//AE dehate are always instant. Yet, those spells do not contribute more than half of the spells cast during a typical encounter. Not even close. ='( CA users have the opportunity to hit more buttons, frequently do, and are "more spam-y" Cripples yes. More than any other class...? Movement//mob position//stealth//mob direction, which dictate which abilities can/cannot be used, are non issues. Mages, who are not crippled by these issues and fare better with longer cast times. Sorcerers are likely the hardest hit of the mages, and I spent the majority of the latency increase playing one, but it is vastly preferable to playing a scout, fighter, priest, and especially a beastlord. If you're unfamiliar with beastlord mechanics, pretend that sorcerer increments decayed very quickly during combat (3/second) and were not generated at an increased rate by AE abilities. Simple things such as "a bard not having swift voices for allegro" is... unacceptable and greatly changes encounter pacing. This is simply not true for a sorcerer.
I made a few more tweaks that are currently live at the time of this post. Update frequency is increased another 20%.
Ok, so I'm speaking in a raid context. I get the impression you probably don't raid in EQ2 based on your posts, and that is fine. In a raid context, a few things to consider: for a scout, you could lose the worse half your CA's and your parse would go down... 15%? It would not matter that much outside of a VC recording and/or a FFU chain for an Assassin. Beastlords don't exist in serious raiding guilds, unless they have a stubborn player who won't change and lack the ability to recruit a replacement/use a rare merc over the beastlord. - Now... intense intense latency issues to the extend where Auto Attacks **stop going off altogether** are pretty big for scouts. I didn't hear of latency issues that got THAT severe. Fighters can mostly stand there and afk as long as they hit the one or two abilities they need to hit. Right now playing a fighter is mostly boring unless you're a try-hard who is pushing dps (which hey, thats cool if you are) Priests too, can do their job, but I can definitely see cure delays, etc. However, this isn't TSO, so unless we're talking someone getting delayed while attempting to cure a last-second-fail-script-curse, I don't see it being crippling. Sorcerers? Ability timing/failure to chain abilities quickly at **specifically the right time** completely destroys their ability to be effective. Example of how this would play out in the real flow of a raid. Let's say on Mob ABC, I generally place first at 34million dps with our Assassin coming in second with generally around 30million dps. Now we add latency issues, and they spike so severe that the Assassin casts 0 abilities in their FFU, but I also get 0 abilities off in my Focused Casting rotation. The assassin's dps drops from 30million to 26million. Instead of 34million dps, I do 15million. Not really an exaggeration.
Weird, I didn't notice any problems at all, not even a DPS drop with the original tweaks last week. As a matter of fact running the cellar raid and then aom/ossuary tues-thur was some of the more latency free raids I have been in for a long time. Although it wouldn't surprise me if third-party programs weren't responding as fast with slower updates being sent from the servers.
The issues were sporadic. When they happened, they crippled me. Also the zone crashed altogether, but I think that was unrelated and I think Gninja is fixing that bit.
Still, I can't even recall it spiking to some unusual amount. Only time it slowed down is the typical 'room worth of trash pull' with 15+ mobs which it did before anyway. Really don't expect the game to perform well under such circumstances to begin with, kind of an unreasonable expectation.
I don't raid currently. Lag is a part of that (itemization//timeslot being the others). I appreciate your input and your perspective, as it provides valuable input to what raid guilds are doing now-a-days. Yes I agree that focus casting, during time warp, with increments is really cool and mechanically difficult, therefore "lag is really bad". I simply disagree that it is more difficult than a bard using abilities during combat mastery. I present a different perspective, what is possible if the game is truly pushed. A practical/impractical example relating to this thread is at low pings (100 ms), you can do a lot of really cool things. For example, you can (could?) get 4 CAs/2.4 AA rotation. Is this difficult? Absolutely, it practically demands that every skill be separately keybound. Other class examples include pre-curing an interrupting status (delay 0.1~0.2 s), snapping a memwiping mob before the auto attack goes off (delay 0.0~0.3 s), or interrupting an ability (0.25~1 s). Now, do you design your raid around consistently performing these tasks? Of course not, your raid is built to be great AFK, you said so yourself! Fighters don't need to push buttons to hold aggro. Priests don't need to cure on time. Scouts need only to auto attack to DPS. Mages don't care about spells save their burst. Most EQII raids are this way, and that is fine. Designing content for 24 super-high-end players who min/max everything would be catering to the 0.00001%. However, this content already exists. For example, is it possible to kill Neragul's/Getiar Razorwing's without a mage. Usually not, but by timing tranquility//precuring "perfectly" it is possible. I've done it, and every fight that doesn't require a perfect cure is easy by comparison. However, when I can't perfect cure because of lag well... that is something different entirely. It isn't me vs. the game anymore, I'm losing because of a force outside my control. That sucks. Latency was indeed bad enough to miss auto attacks. I couldn't achieve 4 CAs/rotation now, but its also high noon EST and I have ping far exceeding 100 ms so I expected that result. I'll try again this evening when the net is clear.
So the average of the samples before this were 1.29 CAs per second. A new small set of samples after this change: 17 CAs in 10s or 1.7 CA/s 17 CAs in 11s or 1.55 CA/s 18 CAs in 11s or 1.64 CA/s 13 CAs in 8s or 1.63 CA/s New average there is 1.63 CAs per second. So it is a significant improvement. Still not reaching the 2 CA/s that my stats say I should be able to though.
I'm speculative that ping goes a long way. To the best of my understanding, the client forbids ability cast until the server says it is okay. This effectively adds a "ping based delay" on everything. Obviously I don't have the source code, nor have I ever worked for SOE, DBG, or related companies. It is, however, the best explanation I have heard thus far. That said, you're doing far better than I am tonight. I am experiencing a 150~170 ms ping, therefore the expected range is 1.492~1.538 CAs/second. I'm averaging 1.511~1.518 which is very close to expected values. What is your ping/where are you stationed (I'm guessing 110 ms//west coast)? I've yet to have many detailed discussions about this sort of thing, and I could use your help! The last time I did testing of this nature was when I was located in Oklahoma, which is a good deal closer to California than Ontario. I could achieve 4/rotation there, but I honestly couldn't tell you what my ping was (just that it was better). Unfortunately, due to my new location, I'm not the best reference to how things were before on the super high end. I've been on priest since moving to Ontario, and it is significantly less important that I obtain 4 CAs/s (and more important that I time things adequately), and thus my focus has been elsewhere.
I am in Washington state and running on fast machine... looks like my ping is around 120ms right now. The accuracy of timestamps on logs and my method of measuring it may introduce a tiny bias in the favor of the server. For consistency with previous results I'll continue doing the same thing though. Here's what I'm doing: Activate CAs as fast as possible one after another (everything with a 0.25s cast and 0.25s recovery that gives savagery from Savage Spirit). Continue until I run out of things to click. Count the time between timestamps on the first occurrence and the last occurrence of "YOUR Savage Spirit emboldens YOU". Count the number of occurrences of "YOUR Savage Spirit emboldens YOU" in that log segment.
@120 ping the expected value is 1.61, which is consistent with your tests. The variance is higher, but your ping is much smaller so larger deviations are expected (i.e. 100~140 ms is 1.666 ~ 1.562 vs. 150~170 ms of 1.492~1.538). Also a fast machine, but I have disabled TCP Delay on this computer which reduces latency in some online games. A general rule of thumb is the older the game, the more likely the chance that it will provide a boost (e.g. it helped immensely in Ragnarok Online 2003, but League of Legends is coded with TCP in mind and thus has no influence). I have received mixed reports for whether it helps with EQII. Based on the alignment of our data with theory, however, it doesn't seem to influence it much if at all. I wrote a guide for disabling TCP Delay for Windows Professional 2007 in 2013, but there are guides for other operating systems elsewhere. Your counting method is a lot less annoying than mine, which involved exporting ACT data into a notepad file, importing that into an Excel file, then selectively parsing it to remove hit inconsistencies (dots, procs) on outgoing skill damage. I'll swap to your method, although they should provide identical results.
Well all the flavor of month warlock spots were taken and there was no more room on 'hemo- toxin' band- wagon either, sides I prefer to raid in 'non' serious guild where we kill trash for 3 hours despite the lag issues. Thanks to dev and others for trying to reduce it. Will test out tonight during our 'silly' raid.
Karsa above ^^^ is a pretty darned good beastlord, too. A definite asset to our "silly" raid group. I've noticed the slowdown and I honestly thought it was just me getting old, or my new glasses prescription messing with my brain, or a lack of caffeine, or something. I've felt like I've been playing through mud, and my DPS has suffered. Now that I've read this thread, I know why. It really does diminish my enjoyment of the game. Put it back the way it was...
If ping is affecting the number of CA/s someone can use, that says to me there is a bit of client/server coordination coding work to be done. Ability queuing is supposed to eliminate the impact of ping and other minor hiccups. It sounds like it might not be working completely correctly at the moment. If we were to actually "expect" to be able to sustain fewer CA/s when I have higher ping (even though I press the exact same buttons in the exact same way)... that basically says every beastlord needs to physically move to the west coast to be fully effective. Which is rather silly.
The only way to do that would be if the server knew of the queue'd ability, which would cause some very interesting events where that ability changed on the client but it was too late for the server to respond! I'm not advocating that this is the correct way, rather that the alternatives are just as problematic. Ideally, everyone would be on the west coast. Some classes have it worse off than others, but this isn't an EQII exclusive. Before I joined EQII, I played Ragnarok Online (think MMO-League of Legends) and it was commonplace for character calculators to account for server based delay. When I write spreadsheets for class comparisons, I account for server delay as well. In a competitive environment, classes are absolutely balanced around low ping. EQII is not that type of game. You don't see players moving their residence//location based classes that are common in competitive games (e.g. LoL, RO, DoTA2, CS). My best advice is to do what you can to work around it, e.g. if I'm stuck with 0.67~0.70 s CA delay, I should look into 6 s weapons and aim for 5 CAs/3.6 s rotation rather than the ideal 4/2.4 s. That said, lets see if there is a difference post update. Unfortunately, I suspect this is as good as its going to get... but I don't have a point comparison to say for sure.
From my testing just now, my dirge appears to be back to pre april 28th parse on epic training dummy! I haven't had a chance to run the assassin to see if the latency is completely resolved or not yet.
