Moonkin Raiding - Page 173

Passthesugar · 12/09/09, 4:27 PM

Prior to 3.3, the window for chaining Eclipse was relatively large. Following SAA, any Starfire landing within 0.3 secs (roughly) of Lunar Eclipse ending could cause the next Eclipse to proc. Now, with 60-70ms, the window is (as far as I can tell) nonexistant. Therefore I've gone from chaining 75%+ of Lunar-->Solar eclipses to 0%, which (in addition to the 2pT8 nerf) is a pretty big hit.

• Hamlet · 12/09/09, 4:33 PM

Originally Posted by Passthesugar

Prior to 3.3, the window for chaining Eclipse was relatively large. Following SAA, any Starfire landing within 0.3 secs (roughly) of Lunar Eclipse ending could cause the next Eclipse to proc. Now, with 60-70ms, the window is (as far as I can tell) nonexistant. Therefore I've gone from chaining 75%+ of Lunar-->Solar eclipses to 0%, which (in addition to the 2pT8 nerf) is a pretty big hit.

Yes, it's intentional that chaining Eclipse was made impossible. We've talked about this a lot. It was compensated by the 10% Eclipse buff.

Passthesugar · 12/09/09, 4:35 PM

heavy edit: Ignore my posts, was totally premature in posting some ill conceived thoughts. My numbers became more normal as I got used to the different approach needed.

smallrabbit · 12/09/09, 5:42 PM

funktionatin SSA from Adoriele

http://elitistjerks.com/attachments/...dawev1.4.0.zip

Spink · 12/09/09, 6:12 PM

Originally Posted by smallrabbit

funktionatin SSA from Adoriele

http://elitistjerks.com/attachments/...dawev1.4.0.zip

For anyone looking for a convenient link to double check this is safe and from Adoriele : SquawkAndAwe

feior · 12/10/09, 5:49 AM

Originally Posted by Librarybook

Ok. Let us know how much of a dps gain you can get by causing massive amounts of lag. At the very least the results of your testing could be of benefit of our Australian moonkins. I'm especially curious how solar eclipse will play out.

The sole reason for the suggestion was to test out the inner mechanics of the new eclipse buff. If eclipse could be triggered by a WE styled eclipse with half a second lag, then it supports the constant 15.1 second eclipse cooldown. Otherwise it supports the idea that the eclipse cooldown is something a lot more complicated. I guess the general gist is this:

0-12 sec during eclipse: cast starfire - make sure you stop casting if you can't get
12-15 sec: cast starfire and have the client start sending the message before eclipse is over
15-15.5 sec: lag
15.5 sec: receive crit or hit (if crit and procs solar, then it supports 15.1 second eclipse)

Kind of a hard test to do with all the lag I'm suggesting, but it just needs to work once.

By the way librarybook, you still haven't answered my previous question. Just pointing it out, as the answer would be very useful information for me.

diratheman · 12/10/09, 1:09 PM

Originally Posted by feior

The sole reason for the suggestion was to test out the inner mechanics of the new eclipse buff. If eclipse could be triggered by a WE styled eclipse with half a second lag, then it supports the constant 15.1 second eclipse cooldown. Otherwise it supports the idea that the eclipse cooldown is something a lot more complicated. I guess the general gist is this:

0-12 sec during eclipse: cast starfire - make sure you stop casting if you can't get
12-15 sec: cast starfire and have the client start sending the message before eclipse is over
15-15.5 sec: lag
15.5 sec: receive crit or hit (if crit and procs solar, then it supports 15.1 second eclipse)

Kind of a hard test to do with all the lag I'm suggesting, but it just needs to work once.

By the way librarybook, you still haven't answered my previous question. Just pointing it out, as the answer would be very useful information for me.

I'm Australian player so i can test the proc on lag to see if can still be triggered like WE did. (i sit on 250 -350ms in raids)

ill post back after i have done it.

Blades · 12/10/09, 1:37 PM

Does anyone know the value of the proc on Nibelung for moonkin? Apparently it can proc off of periodic dmg; 80% certain I saw it proc while I was channelling hurricane, and had a MF up.

Orin · 12/10/09, 2:03 PM

Originally Posted by Blades

Does anyone know the value of the proc on Nibelung for moonkin? Apparently it can proc off of periodic dmg; 80% certain I saw it proc while I was channelling hurricane, and had a MF up.

1% according to Wowhead. The real question is if it procs from IS and MF tics or not. From there you could calculate the DPS value of the proc, but it will take some work and more testing than what has been posted in the wowhead comments so far.

Strize · 12/10/09, 3:01 PM

Hey guys, what it your stance on the new heroic 5man trinket, [Nevermelting Ice Crystal]

I'm currently using abysal rune and the sundial, but i am about 120 haste over the haste cap.

It makes sense for me to use this, as the passive 111sp is much better.

The real question is when should a trinket like this be used. With 2pcT9 this proc affects moonfire ticks(these ticks dont count down the critrating as well!), so ideally it would be best to hit this then refresh MF/IS. Theres 3 spots that i could see this useful in the rotation. Pre-Lunar to try to proc lunar, pre-solar to try to proc solar(and get a couple good wraths in), or at the begginning of Solar Eclipse. What do you guys think?

• Hamlet · 12/10/09, 4:26 PM

Hard to be totally sure on numbers yet, but I think NIC is worse than both Abyssal and Sundial.

Librarybook · 12/10/09, 4:35 PM

Originally Posted by feior

The sole reason for the suggestion was to test out the inner mechanics of the new eclipse buff. If eclipse could be triggered by a WE styled eclipse with half a second lag, then it supports the constant 15.1 second eclipse cooldown. Otherwise it supports the idea that the eclipse cooldown is something a lot more complicated. I guess the general gist is this:

0-12 sec during eclipse: cast starfire - make sure you stop casting if you can't get
12-15 sec: cast starfire and have the client start sending the message before eclipse is over
15-15.5 sec: lag
15.5 sec: receive crit or hit (if crit and procs solar, then it supports 15.1 second eclipse)

Kind of a hard test to do with all the lag I'm suggesting, but it just needs to work once.

By the way librarybook, you still haven't answered my previous question. Just pointing it out, as the answer would be very useful information for me.

Feior,
You're still talking about a subject that was proven not to work two pages ago, and you're demanding code from random forum members? Furthermore you're doing this not to try and up dps, but to be right and prove me wrong. you are a complete idiot.

Nibelung dropped for us last night, the damage from a single proc is between 27 to 30k and i've read reports of two Valkyries being up at the same time, so there doesn't appear to be a cooldown.

Crook · 12/11/09, 5:45 AM

Does anyone know if the new idol is worth getting first? Or should you wait to get tier gear now that its all itemized better?

feior · 12/11/09, 6:27 AM

Chance of [Sundial of the Exiled] proc = 10% with 45 seconds of IC
(chance to not proc)^(x spell casts) = chance to not proc after x spell casts

(1-0.1)^x = 0.5
x = log0.5 / log0.90 ~= 6.5 spell casts on average to have a 50% chance to not proc
(i.e., after 6.5 spell casts, you're expected to have had a 50% chance to proc the trinket - so average proc is after 6.5 spells)

supposing you have an average of 1.1 second casting on wrath, 2.2 second casting on starfire, and 50/50 for cast time on both spells, then (1.1+2.2)/2 = 1.65 average cast time.

6.5*1.65 ~= 10.7 seconds average time to proc after IC -> expect to proc it every 55.7 seconds

proc lasts for 10 seconds, so we're expecting about (590)*(10/55.7) or average of 106 SP

[Sundial of the Exiled] ~= 84 crit + 106 SP

[Nevermelting Ice Crystal] already has 111 SP, so the SP upgrade is very minor over the Sundial.

If we were to suppose that NIC has 920 crit for 20 seconds every 3 minutes or 180 seconds (so take 1/9'th of the proc's worth for an average), then NIC has 102 average crit which is an up on sundial in this scenario. Unfortunately, this crit buff can only stay up at 920 for 20 seconds if you don't crit (assuming 1.2 sec uncrit wraths, 20/1.2 ~= 16.67 -> 0.5^16.67 ~= 0.0007% chance of not critting after 16.67 casts)

Using this during lunar eclipse is not generally recommended, so let's assume you use this as soon as you come out of lunar. Let us suppose you're a boomkin completely buffed at 34% on sheet, then 3% from pallies, 4% from talents, 4% from idol, 5% from mages/locks gives you about 50% crit chance for wrath.

I will attempt to calculate an average best case for this trinket.

The trinket gives 20% off the start, and -4% every crit -> so that means after 5 crits, we lose the buff. For now let's assume that the trinket gives a buff for 5 crits (ignoring the fact that the the trinket's proc is crit related). How many casts, on average, will it take take to consume the buff? Well Wrath crits for 50%, so there's an equal chance of critting or not critting.

Now I want to find n casts such that sum(C(k,5),k=5..n)/(2^n) = 0.5 (number of combinations of more than 4 crits out of n casts, over the set of all types of cast, and we want this to occur half the time so we choose this to be equal to 0.5). We find that n = 10 (proof is along the lines of realizing that half of all permutations of a set of 2*x, where the set can only contain hits or crit elements, are exactly the same as the other half).

So an average of 10 casts will clear the buff. Let us say our average wrath time is 1.1 second (for simplicity), we find that the buff lasts for 11 seconds. This part is kinda pulled out of my ass, but the average value of the buff inside this 11 second window is 920/2=460 (which I'm fairly confident is true).

So for an average of 11 seconds out of 20 seconds, you're getting about 460 crit -> (11/20)*(460) = 253 crit per proc. Since this proc occurs at best 1/9 of the time, we get 253*(1/9) ~= 28 crit

I considered this to be the best case for NIC because the buff actually reduces the average number of casts before it clears itself.

Conclusion? Well it seems to suck a lot. The above is merely an approximation of the best case for NIC, so there are bound to be better. But I was a bit skeptical... 28 crit on average seems to be a little weak.

So I decided to program something in c# to verify.

static void Main(string[] args)
        {
            Random r = new Random();
            int totaltrials = 1000000;
            int totalcrits = 0, totalcasts = 0;
            int total_casts_before_proc_ends = 0;
            for (int k = 0; k < totaltrials; k++)
            {
                double extracritvalue = .2;
                double casttime = 0;
                double naturesgracetimer = 0;
                double roundcast = 0;
                int timescritted = 0;
                int numcasts = 0;
                bool calculated_procs_end = false;
                while (casttime < 18.8)
                {
                    if (naturesgracetimer <= 0)
                    {
                        roundcast = 1.2;
                    }
                    else
                    {
                        roundcast = 1.0;
                    }
                    if (r.NextDouble() < 0.5 - extracritvalue) //did not crit
                    {
                        naturesgracetimer -= roundcast;
                    }
                    else
                    {
                        naturesgracetimer = 3;
                        timescritted++;

                        extracritvalue = Math.Max(extracritvalue - 0.04, 0);
                        if (extracritvalue < 0.01 && !calculated_procs_end)
                        {
                            total_casts_before_proc_ends += numcasts + 1;
                            calculated_procs_end = true;
                        }
                    }
                    numcasts++;
                    casttime += roundcast;
                }
                totalcrits += timescritted;
                totalcasts += numcasts;
                if (!calculated_procs_end)
                    total_casts_before_proc_ends += numcasts + 1;
                //Console.WriteLine(Math.Round(timescritted * 1.0 / numcasts, 4));
            }
            Console.WriteLine("Percent Crit: " + Math.Round(totalcrits * 1.0 / totalcasts, 4) + " with average proc duration of " + total_casts_before_proc_ends * 1.0 / totaltrials + " spell casts.");
            Console.ReadKey();
        }

I first tested out the code so that it mimicked the trinket as a proc that doesn't modify crit, and I get the expected 50% crit with average of 10 spell casts before the proc dies.

The code above has values set so that the trinket actually affects the crit rating, and we get 54.94% crit (during the proc mind you) with an average of 8.13 spell casts before the proc dies.

4.94%/9 = 0.55% crit on average = 25.2 crit rating on average (which by the way is much less than 84 crit)

[Nevermelting Ice Crystal] ~= 25.2 crit + 111 SP

Strize · 12/11/09, 10:55 AM

Great rundown feior!

Based on your calc that gives a 59 crit rating difference.

With 2pcT9, are the MFs ticks crit rating calculated on a per tick basis, or at the beginning?

If fairly sure it is at the beginning. If so how valuable would the +20% crit to a full moonfire be?

A shadow priests refresh of SW:P using their Pain and Suffering talent keeps the crit rating buff on the SW:P.

Lets assume this works as well with glyph of starfire its being used pre-solar and you extend the moonfire by a full 9 seconds.

If so how valuable would the +20% crit to a full moonfire be?

Although all of this calculation might be useless with a probable incoming change to this trinket. Using Pain and Suffering, shadow priests can keep up the full +20% crit on their SW:P for a full fight by doing their standard rotation.