Pet proc rates revisited

[Coprolith]

1. Introduction.

For those that remember, a long time ago I did parsings of our L62 pet's SoFlame and SoStorm proc rates and the effect of dexterity buffs on the proc rate. In this report, I'll expand on those results in order to better understand the game mechanics of pet procs by looking at haste effects, adding proccing weapons and a detailed examination of when procs occur in relation to combat rounds.


2. Parse details.

All parses were done against the Katta banker, Heracus Helsin, using our DPoC focused Spirit of Sorsha warder with the Spirit of Rellic proc. The banker is a very green con mob of course, but that's irrelevant since im only interested in how often procs fire off and not often they land. I varied pet haste with Arag's Celerity (85%). We don't know the base dexterity of our pet, but the added dexterity was varied by using combinations of IoS (+55 dex) and Dexterity (+50 dex). For pet weapons I used Fist of Ixiblat (FoI) and Blade of the Kedge (BotK).

Determining the proc rate was real easy. The number of procs is equal to the number of "Coprolith`s warder begins to cast a spell." messages in the log. Since pet stop attacking after exactly 10 minutes if they don't get agro'd, the fight time is simply equal the number of times I had to tell my pet to attack multiplied by 10 minutes.

In addition to this i took a detailed look at the logs to see when procs occurred in relation to the pet's combat rounds.


3. Parse results.

Test	+DEX	Haste%	time (m)	# procs	ppm	error
1	0	85	100	634	6.34	0.48
2	55	85	100	726	7.26	0.51
3	105	85	100	752	7.52	0.52
4	0	0	100	627	6.27	0.46
5	105	0	100	715	7.15	0.48
6	55	0	100	699	6.99	0.48

Test 7:
Pet with FoI and BotK (equipped in that order)
added DEX: +105
haste: 0
time (m): 164  (went LD ;-) )
# Rellic procs: 1066
ppm: 6.50+/-0.36
# FoI procs: 360
ppm: 2.20+/-0.22
# BotK procs: 122
ppm: 0.74+/-0.11


4. Discussion and conclusions.

The observed proc rates of Spirit of Rellic are graphically displayed in the following graph:



What we see is that hasting the pet makes no difference to the proc rate. Consequentially, we can average the result obtained at the same +DEX leading to:

+DEX <=> ppm
0 <=> 6.31+/-0.33
55 <=> 7.13+/-0.35
105 <=> 7.34+/-0.35

The effect of adding dexterity buffs is exactly the same as I observed in my original parses of Spirit of Flame using the L62 warder: the proc rate caps out at or around the dexterity added by IoS, and adding more dexterity has no effect that can be observed within the statistical variation of the parses.

I did the test with pet weapons equipped primarily to see if this would lead to a lowering of the spell proc. This occurs with PC, but its hard to notice it all unless you have a very long delay. For this reason i performed the test with an unhasted pet. We can see that the proc rate of Rellic does lower with weapons equipped, but the effect is small and just barely significant; 6.50+/-0.36 vs 7.34+/-0.35 w/o pet weapons. I'll get back to this shortly.

The proc rates of the pet weapons themselves seem to follow the same rules as PCs as well, roughly 2 ppm MH and 1 ppm OH.

When we look at when the proc occur, the following can be observed:
- Not once in all 7 tests did Rellic proc twice within the same second.
- Neither the FoI nor the BotK proc'ed twice in the same second either.
- There were 13 instances however of a FoI and BotK proc in the same second.
Since the pets delay is larger then 1 second, this means that the individual procs cannot go off more then once per combat round per hand. It also suggests that Rellic only fires of from the mainhand. I cannot recall if Rangers can get spell procs of their OH weapon, but everything else points to the conclusion that the mechanics of pet procs are exactly the same as for PC procs. The chance that these observations are accidental is effectively zero. The chance of finding 0 double-FoI procs when the chance of that occuring would be at least twice the chance of a FoI+BotK event is already remote, and the chance of finding double-Rellic procs would be 20 times larger at least.

Having established that pet procs work no differently then PC procs, we can now estimate the effect that equipping the pet with weapons has on the spell proc rate as well. I calculate it to a reduction in proc rate of 92% unhasted and 96% hasted. The parsed reduction for the unhasted situation is 0.89+/-0.07. That agrees well but of course thats easy to say with such large uncertainty in the data. Still its very unlikely that the calculated numbers are off by more then a few percentpoints. What this means in a normal situation where the pet is hasted he looses about 4% * 7ppm ~ 0.3 Rellic procs per minute. Since you get 3 weapon procs per minute in return for that's nothing to be concerned about.

The paper is now open for discussion  :wink:

/hugs

[Grymlok]

You are my hero   :shock: