[Elemental] Rotations - Page 13

Binkenstein · 01/23/09, 3:11 AM

Another question, now that we have a short duration self buff, is whether we should sync up EM with trinket use.

It works out as a comparison like this (for a 6 minute fight), where EM = Elemental Mastery, T = Trinket, Sync is waiting to use both every 3 min, and Use is as cooldowns are up. The difference is Sync - Use

	Combo	Sync	Use	Difference
A	EM + Trinket	0:40	0:20	+0:20
B	EM	0:20	0:40	-0:20
C	T	0:00	0:40	-0:40
D	Base	5:00	4:20	+0:40

Which results in a 20A + 40D vs 20C + 40D comparison, and when assuming D = 0 and the others are the gain from said conditions, we end up with a break even of A = C+2D.

Generating some values from SEIC, it is clearly weighed towards using cooldowns asap, while a 2+2 fight ends up being mostly even. Conclusion: use cooldowns when they are up, don't sync them.

Cragen · 01/23/09, 8:27 AM

Originally Posted by Binkenstein

Another question, now that we have a short duration self buff, is whether we should sync up EM with trinket use.

It works out as a comparison like this (for a 6 minute fight), where EM = Elemental Mastery, T = Trinket, Sync is waiting to use both every 3 min, and Use is as cooldowns are up. The difference is Sync - Use

	Combo	Sync	Use	Difference
A	EM + Trinket	0:40	0:20	+0:20
B	EM	0:20	0:40	-0:20
C	T	0:00	0:40	-0:40
D	Base	5:00	4:20	+0:40

Which results in a 20A + 40D vs 20C + 40D comparison, and when assuming D = 0 and the others are the gain from said conditions, we end up with a break even of A = C+2D.

Generating some values from SEIC, it is clearly weighed towards using cooldowns asap, while a 2+2 fight ends up being mostly even. Conclusion: use cooldowns when they are up, don't sync them.

That is what I've been doing with my Elemental Mastery and Blood Fury. Since I just kept my old macro that used both of them, I just do that at the start of a fight or I know we're using BL early and wait for BL to be up. Then depending on the duration of the fight I'll use my Blood Fury as soon it's off cooldown.

Other than that there isn't much more I can do since both my trinkets are proc based. Very nice when I get trinket procs and EM+Blood Fury, but afraid it rarely happens.

Shaâden · 01/23/09, 9:49 AM

Originally Posted by Harmiona

This rotation is good when Bloodlust is active?
FS LvB LB LB LB LB LB Lvb

There is not any good fixed rotation.
Point.

This is a priority system, so the number of LB between LvB will highly depend on your passive haste, your/raid haste procs, bloodlust, but also lag, movement,....

Therefore: DO NOT USE castsequence macros.

Tzarr · 01/23/09, 10:16 AM

On MMO I read, that LO only procs with a probability of 6.6 % with CL, if CL hits 1 target. With 2 hits it procs with 13,3% and 20% with 3 (6,5% per hit).
I wasn't aware of that and thought it would proc with 20%, independent from the number of hits. Did some quick tests which confirmed this theory.

50 LBs: 10 LOs (20%)
100 CLs on 1 dummy: 8 LOs (8%)
50 CLs on 3 dummies: 9 LOs (18%)

Didn't hit the percentages exactly, but I think I did at least "proof", that on 1 target, CL doesn't overload with 20%.

So the question once again: How often should I use CL on single targets? Did you guys include that in your calculations for the DPSC of CLs?
I am not as far into TC as some of you guys here, but I can image, with the new Glyph of Lightning Bolt and this (at least for me) new thing with LO not affecting CL as much as expected, LB gets ahead of CL in terms of DPSC on single targets.

Gerant · 01/23/09, 12:07 PM

Originally Posted by Tzarr

On MMO I read, that LO only procs with a probability of 6.6 % with CL, if CL hits 1 target. With 2 hits it procs with 13,3% and 20% with 3 (6,5% per hit).
I wasn't aware of that and thought it would proc with 20%, independent from the number of hits. Did some quick tests which confirmed this theory.

50 LBs: 10 LOs (20%)
100 CLs on 1 dummy: 8 LOs (8%)
50 CLs on 3 dummies: 9 LOs (18%)

Didn't hit the percentages exactly, but I think I did at least "proof", that on 1 target, CL doesn't overload with 20%.

So the question once again: How often should I use CL on single targets? Did you guys include that in your calculations for the DPSC of CLs?
I am not as far into TC as some of you guys here, but I can image, with the new Glyph of Lightning Bolt and this (at least for me) new thing with LO not affecting CL as much as expected, LB gets ahead of CL in terms of DPSC on single targets.

I'm not sure if 50-100 casts is enough to get statistically significant results out of the RNG in game.

Tzarr · 01/23/09, 1:21 PM

Ok, I did further 400 CLs on one dummy:

First 100 CLs: 6 LOs
Second 100 CLs: 12 LOs
Third 100 CLs: 7 LOs
Fourth 100 CLs: 7 LOs

Combined with den 100 CLs I made earlier this day, I get 40 LOs after 500 CLs -> 8% proc chance. I have to admit, for now I can't proof we have a 6.6% proc chance, but I am pretty sure its no 20% proc chance either.

"Kettenblitzschlag" is german for Chain Lightning.
The counter (Anzahl) shows CLs + Overloads (400 CLs + 32 OLs).

Mmootimus · 01/23/09, 2:23 PM

Originally Posted by Binkenstein

Just to clarify things....

Running fixed rotations is not the best way to maximise dps whether it be by macro or whatever.

I would in no way disagree with this, and /castsequnce macros are particularly bad, however I would be interested to know thoughts on the following:

I am currently trying a fixed rotation of

FlS, LvB, CL, LBx4, LvB, LBx4 ... and repeat. (This is for non-BL, with BL it obviously doesn't work).

The idea of this is that it gives me a very tight rotation of just a fraction over 8secs between LvB casts.

There is probably no easy answer to this, but my question would be how much DPS is this costing me, and is a priority system the way to go, given that I am not convinced the extra damage (mainly from CL) will make up for the fact my LvBs are unlikely to be as tight with me making it up on the fly.

I'm not disagreeing that priority rotations are best on paper, but in the situation that you can make a very tight fixed roation, and with CL now being relatively less appealing, are priority rotations always going to be a meaningful improvement?

LITTLER0TT · 01/23/09, 2:45 PM

What I've been doing for a while is figure out with your particular haste what is best to cram into those 8 seconds between Lava bursts.

What I ended up having was about the same exact rotation you have there

It ends up being 1 chain lightning between every lava then fill the rest with Lightning bolt

My haste has since increased and now i just LvB and Lightning bolt. No need for chain lightning because if i cast it i just end up sitting there for .5 seconds waiting for LvB.

In a sense a Priority rotation will eventually turn into a fixed rotation. Its all maximizing damage between LvB. It's just calling it something else. Obviously once you get different haste buffs it changes completely.

I've been theorycrafting a rotation every time I notice my old rotation seeming...off.

I'm sure my opinion is primarily mine own though.

Ghando · 01/23/09, 2:57 PM

Mmootimus, I am using the exact same rotation (and using FS in the place of CL for the second half of the cycle). Having a fixed rotation with a given set of gear and buffs is fine; if you think about it, with a given amount of haste in a stand-and-nuke situation there will be an objectively ideal rotation. When people say fixed rotations are bad they're really just saying you should remain flexible. If you have a rotation with your current setup that is tight and efficient, stick with that. Just be prepared to change it as your gear and situation change.

Montegomery · 01/23/09, 3:16 PM

Originally Posted by Mmootimus

I'm not disagreeing that priority rotations are best on paper, but in the situation that you can make a very tight fixed roation, and with CL now being relatively less appealing, are priority rotations always going to be a meaningful improvement?

It's not a bad thing to know what will be the tightest rotation when a fight allows, but it's bad to take that knowledge any further than a guideline. There are disruptions to a tight rotation even on a Patchwerk-esque tank and spank; the moment you get a haste proc or Bloodlust is cast the tight rotation is shot.

To boot, there aren't that many tank and spank fights. Even if a fixed rotation were preferable in such situations the majority of fights are going to disrupt that rotation fairly consistently. In the end a priority system and a fixed rotation will look almost identical on Patchwerk, but the latter will suffer everywhere else.

Ezareth · 01/23/09, 3:40 PM

Originally Posted by Shaâden

There is not any good fixed rotation.
Point.

This is a priority system, so the number of LB between LvB will highly depend on your passive haste, your/raid haste procs, bloodlust, but also lag, movement,....

Therefore: DO NOT USE castsequence macros.

During Bloodlust you should never use chain lightning if you are above 10% haste unbuffed unless there are more than 1 targets.

Kishkegelt · 01/26/09, 12:23 PM

I guess the question is whether there is a formula for determining the dps cost of waiting or going over the 8-second LvB timer. So, for example, is it better to throw CL LB LB CL LB for a total of 7.97sec (including casting delay) or CL LBx4 for a total of 8.37sec.

For that you need to know what your expected damage from LvB, LB, and CL are. In the SEIC, is there such thing as calculated average damage for each of the spells?

lrdx · 01/27/09, 8:18 AM

Originally Posted by Tzarr

The counter (Anzahl) shows CLs + Overloads (400 CLs + 32 OLs).

LO will not proc on misses, that makes 32 LO proc from 382 CLs (8.37% proc chanse). With 95% confidience, the margin of error of this test was 2.77% ( 1.96*sqrt(0.08*(1-0.08)/n) = 0.0277), so I take the theory of LO not proccing 20% for single target CLs proven. Whether it's 6% or 8% is still uncertain, to get the confidience interval down to 1%, we need a sample of 2800 CLs.

Tzarr · 01/27/09, 8:52 AM

Originally Posted by lrdx

LO will not proc on misses, that makes 32 LO proc from 382 CLs (8.37% proc chanse). With 95% confidience, the margin of error of this test was 2.77% ( 1.96*sqrt(0.08*(1-0.08)/n) = 0.0277), so I take the theory of LO not proccing 20% for single target CLs proven. Whether it's 6% or 8% is still uncertain, to get the confidience interval down to 1%, we need a sample of 2800 CLs.

Unfortunately I did the test at the lvl 80 dummy and didn't use my complete hit gear... so you are right, I forgot to include the misses in my calculations for the proc chance. Thanks for that

.

I still wonder, if the TC cracks here already knew about this reduced proc chance on single target and if they included it in their spreadsheets and calculations for rotations and DSPC values.

Agash · 01/27/09, 8:56 AM

Originally Posted by lrdx

LO will not proc on misses, that makes 32 LO proc from 382 CLs (8.37% proc chanse). With 95% confidience, the margin of error of this test was 2.77% ( 1.96*sqrt(0.08*(1-0.08)/n) = 0.0277), so I take the theory of LO not proccing 20% for single target CLs proven. Whether it's 6% or 8% is still uncertain, to get the confidience interval down to 1%, we need a sample of 2800 CLs.

From the limited testing I have done (155 casts, 11 LO procs, 7.10% proc chance), it seems clear that CL is not a 20% proc on single target. Assuming that it is not then the next most obvious mechanic would be that it procs 20% of the time assuming all 3 jumps hit. This would require a 7.36% proc chance on each jump which would fall within a reasonable margin of error of both sets of results.

The other possible mechanic might be that it has a 20% chance of proccing exactly 1 LO which would require a 7.85% proc chance on each jump and is again within reasonable margins of both sets of data. This seems to me to be really stretching the spell description though.

I'll continue gathering data when I get the chance but no promises I can get to 2800 any time soon!

Phlis · 01/27/09, 11:49 AM

Originally Posted by Binkenstein

Another question, now that we have a short duration self buff, is whether we should sync up EM with trinket use.

It works out as a comparison like this (for a 6 minute fight), where EM = Elemental Mastery, T = Trinket, Sync is waiting to use both every 3 min, and Use is as cooldowns are up. The difference is Sync - Use

	Combo	Sync	Use	Difference
A	EM + Trinket	0:40	0:20	+0:20
B	EM	0:20	0:40	-0:20
C	T	0:00	0:40	-0:40
D	Base	5:00	4:20	+0:40

Which results in a 20A + 40D vs 20C + 40D comparison, and when assuming D = 0 and the others are the gain from said conditions, we end up with a break even of A = C+2D.

Generating some values from SEIC, it is clearly weighed towards using cooldowns asap, while a 2+2 fight ends up being mostly even. Conclusion: use cooldowns when they are up, don't sync them.

Yeah, I've always found using cooldowns as soon as they're up is the way to go, except, funnily enough, with the new EM, because of the bug. EM's cooldown doesn't start until after the buff wears off, so it takes 3:20, for it to cooldown from one cast to the next. I generally wait the extra 40 seconds so it lines up with my third trinket use, depending on how much health the boss has left(<20% there's no real reason to wait, >30%, I'll wait)

Subvertio · 01/27/09, 8:07 PM

Originally Posted by Phlis

Yeah, I've always found using cooldowns as soon as they're up is the way to go, except, funnily enough, with the new EM, because of the bug. EM's cooldown doesn't start until after the buff wears off, so it takes 3:20, for it to cooldown from one cast to the next.

What makes you think this is a bug, and not a feature?

Daenerys · 01/27/09, 8:28 PM

Originally Posted by Subvertio

What makes you think this is a bug, and not a feature?

Blues acknowledged it as a bug.

Curtis · 01/28/09, 5:52 AM

Originally Posted by Ezareth

During Bloodlust you should never use chain lightning if you are above 10% haste unbuffed unless there are more than 1 targets.

Could you please tell me more about that ?

I tried to do some maths with a pretty simple model (no crit, no overload, no 4% glyph), here is what i get :

CL : 1.5s cast (base cast is 2s)
LB : 2s cast (base cast is 2.5s)

base dmg CL = base dmg LB +200 = X +200

spell power CL = spell power*2/3.5
spell power LB = spell power*2.5/3.5

dmg total CL = base dmg CL + spell power dmg CL = X+200 + 0.57 spell power
dmg total LB = base dmg LB + spell power dmg LB= X+ 0.71 spell power

if you want to maximize your dps = dmg/time

dps CL = (X +200+0.57 spell power)/(1.5(1-haste))=(0.66X+133+0.38 spell power)/(1-haste)
dps LB = (X +0.71 spell power)/(2(1-haste))=(0.5X+0.355 spell power)/(1-haste)

with haste within 0->1 (ex : 30% is 0.3 )

dps CL-dps LB = (0.16X+133-0.025 spell power)/(1-haste)

->This model doesn't find haste having any influence on CL dps better than LB dps or not (which is not that a surprise the way i made it).

Did I miss something or is this model too simple or ...

Thanks in advance !

doogless · 01/28/09, 6:27 AM

It has to do with Bloodlust pushing Chain Lightning below a 1.0 second cast. Spells have an effective minimum cast time of 1.0 seconds, so some of the haste from Bloodlust is wasted by trying to push a spell down to 0.9something seconds like would be happening with Chain Lightning with 10% unbuffed haste. Lightning Bolt doesn't have that problem so the slightly lower damage is outweighed by getting the full benefit of Bloodlust's haste.

Curtis · 01/28/09, 6:29 AM

you're damn right, silly me !

thanks

Shaâden · 01/28/09, 6:34 AM

You forgot at least one important point which is the reason why Ezareth told "during BL":

The GCD is capped at a minimal value of 1 sec.
With 10% haste unbuffed, plus 8% from raid-buffs, plus bloodlust, your CL cast time is under this cap, and so you will waste some time waiting for the end of GCD after a CL cast.

[EDIT: Forget what i said below.... I checked my average LB, LvB and FS damages, and there's no way that pure LB spam could be better than include FS and LvB whatever your haste rate, except maybe with an insane amount of spell power, sorry for thsi stupid idea

]
By the way, FS and LvB being also 2 spells of 1.5 sec (or instant), there should be a cap of haste where spamming LB during Bloodlust would be better than keeping FS and LvB in the rotation (especially during elemental mastery as FS does not scale well with it and LvB does not scale at all of course, wheras LB scales very well with it).

Does anyone have an idea of this haste amount required so that pure LB spam would be the best DPS sequence under bloodlust and elemental mastery ?

Ezareth · 01/28/09, 3:48 PM

Originally Posted by Subvertio

What makes you think this is a bug, and not a feature?

I asked Caydiem on the Shaman forums of all places and she responded that it was indeed a bug.

Hopefully it is fixed in the next patch.

WWS Loading...

5641 DPS on patchwerk last night.

Nailing my rotation was the greatest reason for my increase in DPS.

It is interesting that Lava Burst is only 20% of my DPS and Lightning Bolt 70%.

The rotation of Chain Lightning in every other lava burst rotation seems to be optimal. I definately stopped putting CL in my normal rotation and the DPS increase is showing.

Binkenstein · 01/28/09, 6:31 PM

Has anyone tested/confirmed whether Glyph of Lightning Bolt is a 4% multiplier, or that it stacks additively with Concussion?

Mmootimus · 01/28/09, 6:51 PM

I did some testing yesterday and I think it stacks additively with concussion. However I got caught up in the fact that Totem of Hex appears to be bugged, and didn't reach a 100% solid conclusion.