Frost Cataclysm Discussion [4.0 thread] - Page 12

Incoherent · 10/22/10, 1:04 AM

Thanks for that Lhivera.

How many iterations was this? The numbers look very much less RNG affected this time. Very consistent accross the three runs. What does DPE and DPET stand for? Damage Per Event?
I like that there are 3.62 more total spells cast for the 1/2 run, 5.7 for the 2/2 run, in the same amount of time. This suggests your sim is working far closer to how I think it should. Your increased Frostbolt count is there, but not all those extra spells are Frostbolts, rather there is a definite trend towards more Ice lances, possibly more Brain Freezes, which is in line with what it should be I think. Looking at your output I think I finally grasp exactly what you mean by the ratio of Frostbolts cast.

Until my models evolve a bit I am very happy with those numbers. But I would lay money on the GCD being lower. What I see testing this is unmistakable, subjective yes, but repeatable. The GCD is dramatically reduced when casting Early Frost. You can even see it when watching the spell GCD rotating shadow timer graphic, it is clearly (to me, wishful thinking?) rotating much faster under Early Frost. Too fast to see at my framerates almost. Another anecdotal piece of evidence, the GCD timer in Quartz does not appear when casting Frostbolt under Early Frost.

I think it is a case of retrospectively we will see that this was obvious. "Of course Blizzard are not going to implement a talent that has no effect.". Waiting anxiously for someone else to confirm it.

Yes, but you're sure not going to drop FFO to improve the value of EF

Definitely not. But I still still not convinced that FFO doesn't actually enhance EF. I would leap onto my next hobby horse and postulate that it definitely reduces the value of Improved Freeze if it's cast while FFO is doing it's thing. Not to the point where you would drop it, but does it make casting freeze while FFO is zapping less of a DPS increase than when it's not? Thus altering your priority queue.

Zaldinar · 10/22/10, 3:15 AM

On EF and the GCD...

Combat logs *suck* for measuring time deltas between effects. The UI API has all the information you need, you just need an addon set up to record it... Which I happen to have made for several such questions. Macros work too, they're just more annoying. Anyway:

EF reduces the GCD for the Frostbolt cast that consumes the effect to 1.0 seconds. Haste doesn't change it any further (Meaning the 1.0 hard lower limit is still enforced), the cast bar time can be reduced below 1.0 seconds as expected with haste effects.

An easy macro to make this effect visible to you as a user is:

/cast Frostbolt
/script start, duration, enabled = GetSpellCooldown(GetSpellInfo(116));
/script DEFAULT_CHAT_FRAME:AddMessage("GCD of "..duration.." Seconds");

Spam that and watch the chat window. It will tell you your current cooldown on Frostbolt in seconds.

Incoherent · 10/22/10, 4:00 AM

Originally Posted by Zaldinar

EF reduces the GCD for the Frostbolt cast that consumes the effect to 1.0 seconds.

Zaldinar, is this statement based on real tests of what your macro is showing you? I would love to see an copy/paste output of your chatlog. Is there any rounding?

I understand what the objection is to using Combat logs for measuring time. But it can support an observation. I'd accept that my observed <0.9s time from one cast start to another is by itself shaky, but the total time between three casts firms up the number. More testing is needed. I am not in a position to test this at the moment, being at work.

NateB · 10/22/10, 4:07 AM

Originally Posted by Zaldinar

On EF and the GCD...

Combat logs *suck* for measuring time deltas between effects. The UI API has all the information you need, you just need an addon set up to record it... Which I happen to have made for several such questions. Macros work too, they're just more annoying. Anyway:

EF reduces the GCD for the Frostbolt cast that consumes the effect to 1.0 seconds. Haste doesn't change it any further (Meaning the 1.0 hard lower limit is still enforced), the cast bar time can be reduced below 1.0 seconds as expected with haste effects.

An easy macro to make this effect visible to you as a user is:

/cast Frostbolt
/script start, duration, enabled = GetSpellCooldown(GetSpellInfo(116));
/script DEFAULT_CHAT_FRAME:AddMessage("GCD of "..duration.." Seconds");

Spam that and watch the chat window. It will tell you your current cooldown on Frostbolt in seconds.

Using this script on live, the GCD appears to be reduced BELOW 1 second when early frost (2/2) is up:

Yfrog Image : yfrog.com/jbwowscrnshot102210005314j - Uploaded by Nb83

That said, the numbers appear slightly off. I have 28.64% haste and NW presence, which should mean a 1.139s gcd, and this shows 1.132s, perhaps a rounding error?
It is accurate for the effect of 2pcT10, bringing the GCD to 1.01s

Testing further, whether or not the 2pcT10 buff is up, the GCD for a 2/2 early frost Frostbolt is .8s, which is unsurprising, as it reduces the frostbolt by .7 seconds. However, it appears unaffected by haste (tested both naked and fully geared, GCD remained .8s)

In short, it looks like this will be a very valuable talent after all?

Incoherent · 10/22/10, 4:23 AM

Originally Posted by NateB

In short, it looks like this will be a very valuable talent after all?

That is my opinion.
Exactly how valuable it actually is will depend on how often FoF and BF let you cast EF frostbolt . Assuming a 0.8 second Frostbolt is not higher DPS than IL or BF of course, in that case EF Frostbolt becomes absolutely your highest priority in the absence of DF or FFO. I do not have final numbers for Frostbolt vs IL or BF yet.

This has interesting consequences if the sub 1 second GCD is confirmed. At very high haste levels, EF Frostbolt might become very attractive if the instants are hitting the 1 second limit.

That said, the numbers appear slightly off. I have 28.64% haste and NW presence, which should mean a 1.139s gcd, and this shows 1.132s, perhaps a rounding error?

1.2864 x 1.03 = 1.32499  (32.499% haste total)
1.5secs/1.32499 = 1.1320823 seconds.

It looks ok to me

Zaldinar · 10/22/10, 4:33 AM

Originally Posted by NateB

Using this script on live, the GCD appears to be reduced BELOW 1 second when early frost (2/2) is up:

Yfrog Image : yfrog.com/jbwowscrnshot102210005314j - Uploaded by Nb83

YIKES! That's huge. I can confirm it also works this way on Live, but does *NOT* reduce below the 1.0 second floor value on Beta. I would not anticipate the Live behavior to stick.

Lhivera · 10/22/10, 8:16 AM

I've adjusted my sim to set the GCD to 1.0 seconds when EF is active as per the current Beta behavior. This increases the total value of 2/2 EF to about 2.34%. I don't have time now to do a run with 1/2 EF, but I think it's safe to assume that most of the value is still in the first point. I'll update when I know for sure tonight.

Incoherent · 10/22/10, 9:53 AM

There is an easy way to look at Early Frost, independently of specific DPS considerations.

Assuming the GCD effects we are seeing recently mean we almost always get the full benefit (We always do actually, but we might not get the full benefit of our haste with EF if we are held to 1 second GCD. EF time is removed before haste calculation )

There are three parameters involved.

Fight Length = L
Average Time between EF casts = P (Minimum 15, but 18 typical perhaps)
Time of first Early Frostbolt cast = D (When do you cast your first Frostbolt)

Now.

(((F - D)/P + 1) x 0.7)/ F = Time gained as a fraction of Fight Length. In my world, the DPS improvement.

It is not this simple, because we only get an integer number of EFs per fight. So instead it's:

(rounddown((F - D)/P + 1) x 0.7)/ F = Time gained.

With a First Frostbolt D at 5 seconds, this is the table.

The columns are P, the time between Early Frost bolts. The Rows are F, Fight length.

For First Frostbolt at 5 seconds.

	No. EFs	
						
F\P	15	16	17	18	19	20	21	22
20	2	1	1	1	1	1	1	1
40	3	3	3	2	2	2	2	2
60	4	4	4	4	3	3	3	3
90	6	6	6	5	5	5	5	4
120	8	8	7	7	7	6	6	6
150	10	10	9	9	8	8	7	7
180	12	11	11	10	10	9	9	8
210	14	13	13	12	11	11	10	10
240	16	15	14	14	13	12	12	11
270	18	17	16	15	14	14	13	13
300	20	19	18	17	16	15	15	14
330	22	21	20	19	18	17	16	15
360	24	23	21	20	19	18	17	17

	% of Fight time gained	
						
F\P	15	16	17	18	19	20	21	22
20	7.000%	3.500%	3.500%	3.500%	3.500%	3.500%	3.500%	3.500%
40	5.250%	5.250%	5.250%	3.500%	3.500%	3.500%	3.500%	3.500%
60	4.667%	4.667%	4.667%	4.667%	3.500%	3.500%	3.500%	3.500%
90	4.667%	4.667%	4.667%	3.889%	3.889%	3.889%	3.889%	3.111%
120	4.667%	4.667%	4.083%	4.083%	4.083%	3.500%	3.500%	3.500%
150	4.667%	4.667%	4.200%	4.200%	3.733%	3.733%	3.267%	3.267%
180	4.667%	4.278%	4.278%	3.889%	3.889%	3.500%	3.500%	3.111%
210	4.667%	4.333%	4.333%	4.000%	3.667%	3.667%	3.333%	3.333%
240	4.667%	4.375%	4.083%	4.083%	3.792%	3.500%	3.500%	3.208%
270	4.667%	4.407%	4.148%	3.889%	3.630%	3.630%	3.370%	3.370%
300	4.667%	4.433%	4.200%	3.967%	3.733%	3.500%	3.500%	3.267%
330	4.667%	4.455%	4.242%	4.030%	3.818%	3.606%	3.394%	3.182%
360	4.667%	4.472%	4.083%	3.889%	3.694%	3.500%	3.306%	3.306%

	Time gained over whole fight
							
F\P	15	16	17	18	19	20	21	22
20	1.4	0.7	0.7	0.7	0.7	0.7	0.7	0.7
40	2.1	2.1	2.1	1.4	1.4	1.4	1.4	1.4
60	2.8	2.8	2.8	2.8	2.1	2.1	2.1	2.1
90	4.2	4.2	4.2	3.5	3.5	3.5	3.5	2.8
120	5.6	5.6	4.9	4.9	4.9	4.2	4.2	4.2
150	7	7	6.3	6.3	5.6	5.6	4.9	4.9
180	8.4	7.7	7.7	7	7	6.3	6.3	5.6
210	9.8	9.1	9.1	8.4	7.7	7.7	7	7
240	11.2	10.5	9.8	9.8	9.1	8.4	8.4	7.7
270	12.6	11.9	11.2	10.5	9.8	9.8	9.1	9.1
300	14	13.3	12.6	11.9	11.2	10.5	10.5	9.8
330	15.4	14.7	14	13.3	12.6	11.9	11.2	10.5
360	16.8	16.1	14.7	14	13.3	12.6	11.9	11.9

I see percentage time gained as roughly equal to a DPS increase. Just like Haste. I am not making the decision about whether prioritising an EF is a gain over casting DF, FFO, BF or IL, but I think it enters the equation as a possible option. There is a possible 1-1.5% to be gained over not caring, which might tip the scales compared to the other spells.

What you gain from EF rests mostly on P

Edit: P, not D
Edit: Bracket in wrong place eqn 2

Lhivera · 10/22/10, 10:05 AM

The problem is that none of that extra time will get you more Deep Freezes, Frostfire Orbs, Pet Freezes or
Waterbolts, all of which add up to a very significant portion of your DPS. That's why the actual DPS gain is much smaller than the extra casting time (though obviously also considerably better than my earlier estimates).

Zeldyrr · 10/22/10, 10:13 AM

Very interesting find on the EF/GCD interactions. The finding made me rethink the results of my Waterbolt tests (here). I changed Zaldinar's macro to cast Waterbolt (and changed the spellID) and Waterbolt in my gear has a 1.21 second GCD.

However, the comments that the combatlog is a bad means by which to test timings made me doubt the other numbers determined in the link, comparing the two different waterbolt macros. Zaldinar, is there a way to write a macro to more accurately test the time between casts of a spammed spell? (As opposed to the GCD time, since Waterbolt cast time > GCD time.)

Incoherent · 10/22/10, 10:51 AM

Originally Posted by Lhivera

The problem is that none of that extra time will get you more Deep Freezes, Frostfire Orbs, Pet Freezes or Waterbolts.

Indeed. That's where it diminishes. But that applies equally to Haste, with the exception that the Elemental gains a bit from that. Or?

If it's 10 seconds you gain, that is 10 seconds worth of FrB, BF, IL DPS which you would not otherwise have done.

Lhivera · 10/22/10, 11:14 AM

Yes. More time equals more casts. I don't believe anyone has ever argued that the talent would reduce DPS (unless the potential gain was so small that human reaction to the change in rhythm messed things up). The only issue has been how much it was worth.

The GCD change on the talent (good find, by the way) changes the equation, making it worth considerably more than my earlier estimates. I don't think there's any doubt that the first point is worth more than NP or Ignite at this point. I should have a good estimate of the second point's value soon.

Edit:

Ran 500-iteration runs remotely.

1/2 EF = +1.07%
2/2 EF = +1.92% over 0/2, +0.84% over 1/2

I'll need to do some runs this weekend to compare the per-point values of NP and Ignite.

ETA: The reduction from my earlier result of over 2.3% is because I reduced the simulation time from 7 minutes to 6 and forgot to shift the start time of Time Warp, so this morning's run didn't include any Time Warp haste.

Lhivera · 10/22/10, 5:36 PM

(double post, my apologies.)

Rennicus · 10/23/10, 2:36 PM

I have two questions, looked in the thread but couldn't find it.

Is it a dps gain to use frostbolt right before icelance/DF when FoF procs?
Should we reforge Haste into crit if we haven't reached 33% yet?

Thanks in advance, great thread.

Serjery · 10/23/10, 8:43 PM

I believe earlier the stat priority was stated as follows:

1) Hit to cap
2) Crit to 33%...as in this is the second most important thing for mages which makes us at pretty much 100% crit chance whenever FoF is up, i.e. no dps loss from non-critting DF. I personally have reforged all my haste into mastery and crit
3) Mastery -> Considering how often we have FoF activated, our mastery is heavily beneficial to our overall dps. I have reforged any of my haste (unless spirit was an option) into mastery after I had my crit @ 33%
4) Crit. Yes crit is still better than haste even after getting to 33%
5) Haste