I'm getting slightly tired of the drama surrounding Lava Burst at the moment, so I want to do some testing. This is to make sure that we know exactly how Lava Burst works, and can apply that information to see if this has any relation to the "decrease" in DPS some have observed since 3.1 was released.
In this testing, we will need to test the following
Glyph of Lava - confirm exactly how it works
T7 4pc - Is it adding 10% to the crit modifier like we expect, or doing something else?
Crit multipliers - are they working exactly how we expect?
Co-Efficient - make sure it's correct
Damage gain talents - make sure Concussion/Call of Flame stack additively
These tests will need to following these conditions:
Elemental Oath cannot be present when LvB is cast. Either setup a cancel aura macro, or respec to be without the talent
Temporary buffs like trinket procs, elemental mastery, totems, auras, etc should not be present/used.
Accurate recording of spellpower figures, talent builds and glyphs need to be noted, as well as whether you have CSD or not
Combat logs should be submitted with the following names: [Test subject][Name][Date] and separated into With/Without sections in the text file
Samples need to have at least 50 hits, and 50 crits (this will allow for better statistical confidence in results, and also more information for testing crit multipliers) for each test (this will result in 200 casts per test). Obviously you can use FS to improve the amount of crit casts later in the test.
Also note whether you have Dual Spec enabled or not
1) There is some bug related to dual-spec. We are fairly confident this is not a problem. It did have some bugs when 3.1 first launched, but we think we have caught them all. Therefore we think this issue is just a red herring complicating the rest of the discussion.
2) Elementals suffered a lot more from losing 5% crit from the Improved Scorch et al. change than did other casters. However, balance druids and some warlocks, also reasonably crit dependent, did not seem to drop.
3) Ulduar fights require a lot more movement, so Elementals can't cast as many spells. (I'm talking about direct damage spells here, not totems.)
4) There is some bug with Lava Burst or another spell that we cannot yet locate. We have looked at this a few times and everything seemed to be working correctly, but we can't completely eliminate this as a problem.
As I have said in the other, longer thread, we need to understand the cause of the problem before we go making any shotgun style changes that may or may not fix it.
I will also point out that the Ulduar set for Elementals is a pretty decent upgrade. You should be in a better relative position with full Ulduar gear than you were with full Naxx gear.
Once we've made sure that Lava Burst works how we expect, we can take a look at Clearcasting, Flame Shock, and finally Lightning Bolt (and the T8 4pc bonus)
In this testing, we will need to test the following
Glyph of Lava - confirm exactly how it works
T7 4pc - Is it adding 10% to the crit modifier like we expect, or doing something else?
Crit multipliers - are they working exactly how we expect?
Co-Efficient - make sure it's correct
Damage gain talents - make sure Concussion/Call of Flame stack additively
You probably want to specify which target dummy to test it on for consistency, and to make sure that any external buffs from other players are not present at the time.
Personally I am 99.9% sure that these tests will not find anything wrong, at least in terms of mechanics, and the max crit obtainable. Credit to you for suggesting a decent methodology, but I just don't think whatever problem there may be is going to show itself at a target dummy. Myself and Tufy have both tested LvB on dummies and found it to be fine, and then got suspiciously low crits when actually raiding.
Anyway, no argument that this is the right way to go about things, and I hope I am wrong, and that there is a bug thats fairly easily identifiable in this way.
This is more for establishing a baseline so that when one of those "suspiciously low crits" occur, we can look at it and say "ok, based on stats/buffs I would have expected X, but I got Y instead" and we can look at where they occur (is is some boss mechanic, dual spec bug, or something else?) and try to figure out why they occur. Then we can take the information to Blizzard and say "We've confirmed odd behaviour under these conditions".
The alternative to this is that everybody just shuts up about "my lava burst crits are low", because if you don't want to help test & confirm the problem, there's really no point in continuing to complain, is there?
The alternative to this is that everybody just shuts up about "my lava burst crits are low", because if you don't want to help test & confirm the problem, there's really no point in continuing to complain, is there?
I have spent rather a large amount of time trying to put this point of view across on the EU forums, and had a fairly predictable response. I probably should have known better.
Personally, having spent a considerable time testing LvB in the past, and several hours testing it again after 3.1, I am both reluctant to spend any more time on it, and personally convinced that there are not any problems to be found from target dummy testing. Of course, I am just one person with my opinion, hopefully there are lots of willing volunteers to go out, do some tests and prove me wrong.
If I may suggest, this thread could also be used in parallel to forensically pick over any "dodgy" LvB crits from actual raiding environment? And I don't just mean anything thats "a bit low", I mean something identifiably below where it should be, from a player who understands the generally assumed mechanics of the spell, and has a combatlog to back it up. I realise this is slightly getting ahead of ourselves, but as LvB is not a new spell, and everyone who has looked so far hasn't found any hard evidence its mechanics have changed (including assurances from Blizzard who checked their change logs etc), would it not also be useful to look at actual data?
Other than the 2 damage increasing raid buffs, special boss mechanics, and any spellpower increase caused by a Demo lock in excess of ToW, I can't think of any other variables that should exist outside of an actual bug. If accurate details were given of spellpower, trinkets etc. at the time, this would have at least half a chance to find something.
I do realise the base mechanics of the spell need to be tested in time to prove anything, and do not wish to discourage anyone from doing that.
Been busy today doing the test in parts since I didn't have enough time to test all the parameters, will update this post with logs tomorrow.
I agree with Mmootimus, I have only encountered questionable numbers in raids so far and not on dummy's.
However, while testing for possible differences from the assumed mechanics you need a base to build your assumptions on. So I hope more people are willing to spend some time to get some logs.
Edit:
been completely swamped this weekend, so a tad slow ;( first 3
I think both methodologies are valuable. Let’s, once and for all, check the baseline functioning on the target dummies (I would suggest a level 80 dummy to avoid misses as people won’t have the +3% raid buff). At the same time let’s start gathering raid combat log data. Not I hasten to add, unsupported comments that I did a 6XXX Lava Burst and should have done more, but WWS reports or the like where we can check what debuffs were on the boss and what buffs were on the player at the exact time the spell was cast/landed.
Mmootimus – do you have any data saved for the tests you have done so me might use those as part of our baseline? I'll get my data together tomorrow and add it to the thread.
I applaud the effort - it's very important to be sure of how the mechanics work before you can look for possible bugs. I will attempt to get some logs as soon as possible on the dummies. Binkenstein, which dummy did you want us to use, heroic or boss?
I started to hear of this "low crits" problem some time ago, but never really thought much of it, didn't feel like I was having any problem with my Lava Bursts. But today on Ignis, I had ~4k SP (trinket procs) and my LvB crit for just 10k. This was in the heat of the moment, so I didn't really check if there was a reason to it, like partial resist or even missing CoE, but my LvB had generally seemed low the whole raid.
Recently, I've been working on this addon here. In it, I've added some debug code to it which will calculate the expected damage range a spell should hit for, and then report if the actual spell hit outside this range. Could be interesting to turn this debug code on during a raid, just to see if there are any problems.
I applaud the effort - it's very important to be sure of how the mechanics work before you can look for possible bugs. I will attempt to get some logs as soon as possible on the dummies. Binkenstein, which dummy did you want us to use, heroic or boss?
Heroic. Eliminates miss and resists affecting results.
Originally Posted by Ryn
I started to hear of this "low crits" problem some time ago, but never really thought much of it, didn't feel like I was having any problem with my Lava Bursts. But today on Ignis, I had ~4k SP (trinket procs) and my LvB crit for just 10k. This was in the heat of the moment, so I didn't really check if there was a reason to it, like partial resist or even missing CoE, but my LvB had generally seemed low the whole raid.
Recently, I've been working on this addon here. In it, I've added some debug code to it which will calculate the expected damage range a spell should hit for, and then report if the actual spell hit outside this range. Could be interesting to turn this debug code on during a raid, just to see if there are any problems.
That seems awfully similar to DrDamage.
As I've said, once we establish the base mechanics, we should be able to reliably predict how LvB scales with % gains and extra spellpower. This means that if someone goes "I had X sp (minimum) for this fight, and got crits for Y, what's going on?" we can easily calculate what the base damage should be, multiply it out for crits, and then see how close this comes to the observed figures. It may be that these are all products of ClearCasting not being up when Ebon Plague/Curse of Elements/Earth & Moon happen to be down, along with everyone switching from Glyph of Lava to Glyph of Totem of Wrath.
If you don't want to help with this, so be it. I'm not forcing anyone to, and I will try to do my own testing (I don't have dual specs though, and I don't see any problems myself, so I'll probably just end up confirming mechanics), but I think it would be fair to say if you don't want to participate in identifying what the "problem" actually is, you should stop complaining about said "problem".
Wouldn't you think the level 80 "Grandmaster's Training Dummy" would produce more accurate results to account for solo hit percentages without +hit debuffs that would normally be on a Boss level mob/dummy?
And also I'd love to lend my assistance in this research but I've never uploaded combat logs before. Could anyone give me a quick "How -To" on what site I could use that allows hosting of Training Dummy combat logs?
Wouldn't you think the level 80 "Grandmaster's Training Dummy" would produce more accurate results to account for solo hit percentages without +hit debuffs that would normally be on a Boss level mob/dummy?
Thinking simply, if people are only seeing problems with Lava Burst crits (it always crits). It would be logically to assume that you are fighting bosses with a certain amount of fire resistance.
I was actually discussing this bug tonight in my guilds caster channel and cheking the numbers in recount all night intentionally not putting up FS to check crit vs non-crit numbers. I was pointing out to fellow casters that I'd see 67k non-crit LvB's and 8-9k crit LvB's on the same mobs. To rule out Fire Resist it was on mobs from every corner of Ulduar. I also lead PuG Naxx25's with friends and guildies to get alts gear and I notice the same ranges there on its damage.
Anyways my money's on one of the crit modifiers of Lava Burst. If it crits for as much as it hits for, or at least it does for me. I also bought dual spec late as I really had no reason to have it since I only play Ele. I was noticing the same problem with and without having dual spec enabled.
Another point of interest before I get to bed... I was sitting at a training dummy for about 5 minutes as I didn't have time to do full tests with variables tonight and decided to shoot of a few Rank 1 LvB's. I got an 8100 crit with Rank 1. Immediately after I shot a Rank 2 and got a 7800 crit. Would it be worth anything to bother with Rank 1 v Rank 2 to verify base damages?
Did some testing with some spare time I had this morning.
All gear should be the same as listed in my armory, no weapon buffs or totems down and I cancelaura'd every trinket proc and elemental mastery.
I was actually discussing this bug tonight in my guilds caster channel and cheking the numbers in recount all night intentionally not putting up FS to check crit vs non-crit numbers. I was pointing out to fellow casters that I'd see 67k non-crit LvB's and 8-9k crit LvB's on the same mobs. To rule out Fire Resist it was on mobs from every corner of Ulduar. I also lead PuG Naxx25's with friends and guildies to get alts gear and I notice the same ranges there on its damage.
Anyways my money's on one of the crit modifiers of Lava Burst. If it crits for as much as it hits for, or at least it does for me. I also bought dual spec late as I really had no reason to have it since I only play Ele. I was noticing the same problem with and without having dual spec enabled.
Another point of interest before I get to bed... I was sitting at a training dummy for about 5 minutes as I didn't have time to do full tests with variables tonight and decided to shoot of a few Rank 1 LvB's. I got an 8100 crit with Rank 1. Immediately after I shot a Rank 2 and got a 7800 crit. Would it be worth anything to bother with Rank 1 v Rank 2 to verify base damages?
6.7k is within the expected range of LvB hits assuming 2 trinket procs, EO/CC up and fully buffed/debuffed. 8-9k is within the minimum range of LvB assuming no trinket procs, no EO/CC and a debuff or 2 missing on the mob. This is why any such evidence should be supported by a WWS report or the like so we can see if there is something amiss or if it is just a combination of factors.
Rank 1 max base damage = 1290
Rank 2 min base damage = 1192
So it is quite possible for Rank 1 to crit for more than a rank 2.
Addressing an early point concerning Fire resist - I imagine we would have seen a little bit of fuss from the Fire mages if that had been the cause. As we have seen none, I think it can probably be discounted unless we have firm evidence.
Edit: turns out I won´t be able to do the testing after all. Sorry!
Also
Originally Posted by Cryonic
I was actually discussing this bug tonight in my guilds caster channel and cheking the numbers in recount all night intentionally not putting up FS to check crit vs non-crit numbers. I was pointing out to fellow casters that I'd see 6-7k non-crit LvB's and 8-9k crit LvB's on the same mobs. To rule out Fire Resist it was on mobs from every corner of Ulduar. I also lead PuG Naxx25's with friends and guildies to get alts gear and I notice the same ranges there on its damage.
If this holds up under scrutiny (e.g. no trinket procs messing things up etc.), this sounds awfully like Elemental Fury not working properly...
What about your average hit/crit for the entire instance? (excluding Thaddius)
Assuming your Armoury is still right with (2 piece T7/T8) and 1927 SP it appears you are actually hitting too hard! Using existing models to analyse the data, I would suggest that your min/ave/max values under those conditions should be:
LvB Max Hit : (1518+0.7714*1927)*1.11 = 3335.0
LvB Ave Hit : (1355+0.7714*1927)*1.11 = 3154.0
LvB Min Hit : (1192+0.7714*1927)*1.11 = 2973.1
In fact you got:
LvB Max Crit : 8326 (+12.4%)
LvB Ave Crit : 7513 (+7.3%)
LvB Min Crit : 6608 (+0.1%)
LvB Max Hit : 3650 (+9.4%)
LvB Ave Hit : 3239 (+2.7%)
LvB Min Hit : 2976 (+0.1%)
It looks like your minimums are exactly where they should be but your average and maximum values are very high. It would take a better expert than me to analyse the combat logs precisely but looking at them, there appear to be times where your trinket procs or CC buff were not removed and this could be effecting the results. For example, the first 6 lines of your log are:
0:00'21.110 Ashye Lava Burst hits Grandmaster’s Training Dummy for 3278 Fire.
0:00'26.250 Ashye Clearcasting was removed from Ashye.
0:00'30.875 Ashye Lava Burst hits Grandmaster’s Training Dummy for 3206 Fire.
0:00'40.219 Ashye gains Dying Curse.
0:00'41.125 Ashye Lava Burst hits Grandmaster’s Training Dummy for 3323 Fire.
0:00'42.938 Ashye Dying Curse was removed from Ashye.
So at first glance it appears that 2 out of those 3 LvBs were effected by CC or Trinket procs. My understandng is that a trinket will proc on cast AND effect that cast which means that the buff cannot be cancelled for that cast (damage is calculated on cast, not landing I believe). The only way to do this test reliably is to remove any proccing trinkets. Finally, although you have been fairly religious in removing the EO buff before casts, it is actually the CC buff that gives the bonus damage and that you have often left on.
As a point of interest, all casts that hit for harder than anticipated were effected by either Trinkets, CC or Lightweave.
In a nutshell, without disecting the combatlog in extreme detail, this test shows that the minimum values are where expected but does not tell us much about the average and top end behaviour of the spell.
I’ve run 3 tests on the level 80 target dummy. All tests were performed with no procs, no CC, no totems and no weapon buffs. No tests included Glyphs that effect the results. All results are given in the format: No. of Casts : Min/Ave/Max.
Test 1 – No Elemental Talents, 1958 SP, 4 Pc T7 bonus
Based on Test 3, we'd expect the following with raid buffs:
Just 13% damage debuff: 7542 - 9294
13% debuff & 3% buff: 7768 - 9573
And that's about where we've been seeing things post 3.1. At this stage I'm thinking the only difference is that people swapped Lava out for ToW, and were expecting to see an extra 500-900 damage on LvB.
I think that is a possibility. The glyph swap and the loss of the 4pc T7 will reduce LvB damage by about 9.3% which is about 1400 damage off max LvB crits and I believe this may account for much of the initial reaction. The Totem of Hex is also more widely available now and that would account for another 400 loss over the LvB totem for people that were using that. However, there are also reports from people that I would usually consider less likely to make a hurried conclusion like that who are reporting a drop in overall DPS which this does not account for.
Some of the DPS loss can be attributed to the nerf to Elemental Mastery and part to the nerf to Scorch – about 200 DPS between them. Some more might be attributed to the more mobile nature of the fights, the more frequent target switching, the changes to rotations as haste ratings change and the wider spacing causing some buffs to be lost. Not least of course is the possibility that in harder content more raid members die and buffs are lost that way.
What we really need to see is some solid WWS reports that include LvBs hitting for less than they should considering all the buffs/debuffs present. Average numbers here will prove nothing as it is impossible to determine what was going on during the fight that might cause DPS to drop.
So if anyone has a report where a LvB (or LB) hits for less than it should, let’s see them and see if they can be explained. All it takes is one conclusive example to prove that there is a problem – it shouldn’t be too hard considering the number of people reporting the problem.
As I posted in the other active thread on this topic, we still cannot find any bugs with Elemental. We looked at a lot of numbers from before and after 3.1 and (except for the lower crit values) the numbers don't seem very different. It could be that other classes, notably rogues and Feral druids, just improved.
Nonetheless, we do think Elemental is a little low so there is likely to be a small buff coming. I feel obligated to add the standard no promises caveat.
I will be interested to see how exactly this is handled. I still feel that Ele is right up there with the best of the pure classes for single target nuke DPS, so it will be interesting what form this buff takes. (This is probably best discussed in another thread).
I do not think this should detract from trying to find any LvB bug however. If there is a bug, it is most likely to affect scaling, and so will not necessarily be permanently fixed by a small buff. I am also concerned generally by the sheer amount of attention this issue has received, and think it needs to be pinned down for the sake of the general sanity of the shaman community (seeing as it has got the attention of everyone from EJ to Blizzard, to say nothing of several thousand less enlightened posters on the various official forums).
Lost 5% crit for 30 seconds every 3 minutes + 5% crit -> roughly 262 crit rating, which would be about 157 dps, although I'm not sure if that would account for the lower ClearCasting uptime. (Obviously it's closer to 262 with both the T8 bonuses).
Gained direct damage bonus to flameshock (all of 20 dps for 2 points)
The other problem is that now that Demonology locks are getting close to 2800 spellpower as well as being viable dps, it may be that Totem of Wrath needs to be changed so that we get some personal bonus from the talent (unless they consider the glyph to be the personal bonus)
I think you have missed the reduction of EM from 30 to 15 secs in your calcs. That effectively reduces your crit by 15% for 15 secs per 3 mins over what you have listed for an average crit rating loss of about 318. Including for CC uptime losses and gains on FS damage, I think our overall DPS loss is around 212. When other classes/spec have received slight buffs (some, more than slight), it is not surprising that we see ourselves dropping down the DPS rankings.
Demonic Pact introduces some interesting possibilities. Of course it is a higher DPS boost than ToW but it is also a less reliable one and pre 3.1 could end up dropping off the target ~2% of the time (not precise as I’m no lock expert). The Scorch nerf actually increases this to maybe a 3% drop off. I am wondering if some of the reduced LvBs reported in raids are due to this drop off as guilds switch to using DP over ToW (especially as Demo locks have been buffed relative to Aff). It would account for a 750 damage drop on LvB crits.
What if this "bug" is not reproducable on a targeting dummy, maybe it requires a raid buff of some sort. Like a warlock's Demonic Pact overwriting totem of wrath, and for one second, neither buff is applied, just speculating here. Although testing on a target dummy does rule out a lot of variables, it may actually rule out the ones causing the bug.
05/07/09, 7:54 AM
Similar Threads
