Wednesday, March 31, 2010

Learning How to DPS the Heroic...Dummy

Last night, I didn't really learn anything from my time with the Heroic Dummy in regards to which spell rotation/priority works best, but I did learn some stuff on how to go about evaluating my DPS on the Heroic Dummy.

So, I am going to share with you, gentle readers, what I did and how it was all wrong.

What I did:
I decided to evaluate one minute of my rotation and stopped all casting right at a minute.

How it was all wrong:
I had wildly different numbers for the same rotations (3990 dps - 4500 dps). Unfortunately, a minute was not enough time for crits to even out to around my stated crit percent, so, for example, on one eval, I had 31% of my Mind Flay ticks as crits. On another eval, I had 46% of my Mind Flay ticks as crits. That was a significant difference.

What I'll do different:
I will cast until I'm Out of Mana on each eval. I didn't want to do this initially because it is a bit subjective - i.e. I don't have mana for the next priority spell, but I do have enough for this one over here. That's why I like the timed idea better, but I suppose, enough time will have elapsed and enough damage accumulated that the end of the session hopefully won't be significant.

-

What I did:
I wore my normal raid outfit.

How it was all wrong:
I have the Nibelung staff, which has a proc. On one of my sessions, it proc'd and dang if I didn't score 500 more DPS on that fight! doh!

What I'll do different:
Since I want to remove as many variables as possible, I'll equip a staff that doesn't have a proc. I'm not looking for raw numbers, but rather I'm looking to compare numbers to each other.

-

What I did:
I did exactly one eval per rotation.

How it was all wrong:
As I mentioned earlier, the time I DPS'd did not allow for crit normalization; add to that the fact that I only did one eval per rotation and you can easily see how my (Normal Rotation without self buffs) > (Normal Rotation with self buffs). Yeah, I was scratching my head at that one also - What, you mean that extra 174 spell power doesn't mean a thing?? yeah, no.

What I'll do different:
Along with the increased session time, I need to run multiples of each test to normalize and average my data. I'm not sure if 5 is (good) enough, but I think that's what I'm going to start with.

-

What I did:
I didn't write down what I wanted to evalute. I had a fuzzy list in my head and wrote it down as I made my evals.

How it was all wrong:
It wasn't really wrong, just disorganized. I don't need to get the spreadsheet out, but hey, having a plan is a good idea.

What I'll do different:
Actually, a spreadsheet isn't such a bad idea. If I do make one, I'll post a screenshot of it here just so you guys can marvel at my Excel 1337ness. You were impressed by my weight chart? You ain't seen nuthin' yet!

Ok, now that confession time is over, let me tell you what I'll be looking at and evaluating:
SW:P - Shadow Word: Pain
VT - Vampiric Touch
MB - Mind Blast
DP - Devouring Plague
MF- Mind Flay
SF Shadow Fiend
SW:D - Shadow Word: Death

First up:
My Normal Rotation (SW:P, VT, MB, DP, MF. I clip my MF to cast MB when it's up, but I don't clip it to cast anything else). Also known as NR.

Verses

NR without clipping.

Verses

NR with clipping MF to cast any higher priority spell(VT, MB, or DP - SW:P is refreshed by MF) produces Winner1. The idea is to get the best rotation before adding spells.

(What follows is one path that I may take. Once I throw in new spells, I need to go back to the NR vs. NR w/o clipping vs. NR w/ clipping for DoTs stage to fully evaluate the DPS gain/loss I may be seeing. I don't have a lot of confidence that I'll be that thorough...)

Winner1 & SF (SF placed after MB and before MF) whenever it's off cooldown, produces Winner2. Note, this should only produce a DPS increase, but it remains to be seen.

The better of Winner1 and Winner2 & SW:Death (SW:D is placed after MB and before SF and MF), produces Winner3.

I'm not sure if I'll be able to hit the dummies before the raid tonight, but hopefully I can. If I can't, then during the raid, I'll make sure to pop my SF every cooldown as well as hit SW:D whenever I'm moving, which is what I should have been doing anyways if a DoT doesn't need refreshing.

Also, I will try very hard to NOT refresh DoTs BEFORE they expire. The argument for not refreshing before expiration is as follows:

Suppose DoTX ticks 5 times in 15 seconds; that's one tick every three seconds with the first tick coming in at 3 seconds, and subsequent ticks at 6, 9, 12, and 15 seconds.

If I refresh early, say just before the last tick, then for DoTX that was being refreshed, the last tick occurred at :12 and the next tick (tick #1 of the new application) will land at :18. I've just gone 6 seconds between ticks. Bad.

If I refresh just after the 5th tick, then I go the normal 3 seconds between ticks. At most, I should go 4 or 5 seconds if I'm finishing up a cast/channeled spell.

I'm going to have to pay special attention to VT in this regard, as I can start casting it before it has its last tick in order for it to land just after the last tick and maximize my up time. SW:P and DP have no such concerns as they are instant casts.

Wow. Looking back, this post actually resembles something useful. Dang, how did that happen?

No comments: