|
Originally Posted by http://elitistjerks.com/f47/t66856-moonkin_pve_dps_updated_3_3_2_a/
Lunar is slightly stronger than Solar (slightly faster to proc and slightly more damage per proc, and makes full use of Glyph of Starfire).
|
A question popped up in my head about Lunar being faster to proc – shouldn’t that be dependent of the spell haste? Especially when above the haste soft cap (i.e. when Wrath is locked at a one second cast time).
I started looking into this and after verifying the haste dependency I found that paired with 5/5 points in Starlight Wrath, Wrath (and thus Lunar proc) takes a severe penalty from latency. I have yet to fully understand this effect, but according to Rawr, when Starlight Wrath is maxed (i.e. when Wrath is on the same cooldown as the global cooldown (GCD)), the cast time of Wrath becomes GCD+latency. I’ve googled and asked on Rawr’s discussion page for an explanation to this, but have not gotten any answer yet. However, I’ve done some testing in the game that seems to verify this.
I thought I’d post my results of the subject here, in case anyone is interested.
If you don’t put any points in Improved Insect Swarm (IIS), the haste needed is independent of crit chance and only depends on latency. However, as you put points in IIS you get a crit chance dependence, and the haste needed to make Solar proc faster than Lunar behaves in a similar fashion to what is shown in this graph (this particular plot is for 3 points in IIS and 0 latency):
As you can see, the higher crit chance you have, the higher haste you need to make Solar proc faster than Lunar.
Now, taking latency into account, the haste needed to proc Solar before Lunar decreases quite drastically as latency increases. You can find the numbers in the table here
haste_needed_to_proc_solar_before_lunar.xls .
For instance, if you have 30% crit chance and expect 50 ms latency, you will need 56% haste to make Solar proc before Lunar. This is a bit beyond soft haste cap - however, during a Nature’s Grace you only need 30% haste to make Solar proc before Lunar under these circumstances.
The numbers are, once again, highly latency dependant, so you have to look at your own crit and haste to find out at what latency you will benefit from farming Solar instead of Lunar and vice versa. It might even be worth installing an addon which shows latency clearer than Blizzard’s original (e.g. Recount), so that if your latency crosses any relevant threshold, you can switch to the relevant nuke.
Remember that the table relies on 3 IIS, meaning you need to keep Moonfire up for the values to be correct. (I’m assuming most people run with 3 IIS, hence I haven’t bothered to do the table with 0/1/2 IIS. If anyone has any interest in this, let me know).
Finally, looking back at the quote:
|
Lunar is slightly stronger than Solar (slightly faster to proc and slightly more damage per proc, and makes full use of Glyph of Starfire).
|
“Slightly” is quite a vague term, and by the time I would’ve finish the calculations to take this into account (what with me being so slow and all), Blizzard will have released the talent overhaul patch. Therefore I have assumed that the increased damage per proc is negligible. If anyone feels like doing the math though, that would be great.
Does it work?
After doing a few tests I’ve found that while farming for eclipse, alternating Wrath/Starfire seems to boost my dps more than spamming the nuke that has fastest proc time. This is most likely because I gain one more eclipsed nuke at each proc. Further, it minimizes Glyph of Starfire’s downtime by limiting it to Wrath spamming during a Solar.
In conclusion, rather than starting fights with Lunar farming, I’d say it seems better to start fights by alternating Wrath/Starfire – beginning with the one that has the shortest time to proc, given in the attached table. Also, moving close to the target while applying dots (unless you can’t because of aoe or whatever), so that Wrath connects faster, gives you more breathing room when looking for the lunar proc.
It would be nice if someone could waste a few minutes to check if they notice any dps-change, either up or down, by casting like this - or if it’s simply my brain that wants it to be better so that I experience some placebo effect.
The calculations for the table has been done in MATLAB, you can find the .m file here
timetoproceclipse2.m
where I’ve tried making comments for each step, if anyone wants to check for faults or whatever (possible to open with regular text editor).