Theck's MATLAB TPS analysis - WotLK/3.x

If that were true, then the test I just performed wouldn't have worked.

I hit holy shield, and waited till the GCD was 3/4 done, then equipped my libram. The GCD continued and finished as normal. Equipping only incurs a GCD if one isn't already up, but it doesn't appear to extend existing ones.

Just tested it a couple of minutes ago.

The GCD extra only happens if you are in combat.

Edit: Just went up to a dummy, started swinging, casted holy shield waited until about half way through, then swaped librams, it reset the GCD for me.

If I may steer the conversation back towards SoV, I'd like some additional combat log confirmation on what components can and cannot crit/dodge/miss/parry, if someone could be so kind.

Grehn|Skipjack wrote:If I may steer the conversation back towards SoV, I'd like some additional combat log confirmation on what components can and cannot crit/dodge/miss/parry, if someone could be so kind.

Based on the combat log posted earlier in this thread...

The 5 stack proc cannot be avoided and it can crit.

The application of the DoT can miss and be dodged/parried.

The tick of the DoT itself cannot be avoided but it also cannot crit.

Just some initial PTR testing. I don't have time to do indepth at the moment:

Shield of Righteousness: There is no extra threat on it. It does threat=damage*1.9*1.43, the same as all other holy abilities. So this is a tooltip error, probably just left over.

Seal of Vengeance: The DoT wasn't nerfed, they fixed the tooltip. It always did that much damage, the tooltip was never adjusted when they nerfed the duration from 18 seconds to 15. The coefficients for a 5 stack on the PTR are 6.5% SP and 12.5% AP on the DoT portion. The melee hit is indeed 33% of your weapon damage and incorporates talents, buffs, and AP changes (so 33% of your actual swing damage, not base weapon damage). Though it isn't 33% of the current swing's damage, just 33% of your general damage range.

EDIT:
Seals of the Pure: Does not affect the 5 stack damage of SoV. I have 445-618 weapon damage ==> 146-204 SoV 5stack damage. Seals of the pure changes that range to 168-234 if it affects the 5 stack damage. So if SotP worked, then I would never see a hit below 168 and I would see hits above 204. I received hits in the 150's from SoV 5-stack and never saw one hit for above 204 in a half hour spent on the training dummy. My spec was 5/0/0 and the points were learned (verified). I had no set bonuses (unequipped my gloves to be safe).

A little disappointed that the SoV attack can't be avoided. Would have been a good place to make Expertise more appealing.

Well saying it couldn't be avoided is a bit of a misnomer. The attack won't go unless your melee swing lands. I for one would not like it if the boss's chance to dodge/parry our attacks got to double dip on our SoV 5 stack attacks

True enough.

Hope things work out for you Theck. The information you put into this is why I play MMO games and RPS at all. Superb min/maxing from analysis. Its very interesting to me, reminds me of back in the D2 days of making ones merc a superior beast of mass murder.

Jasari, thanks for the stand-in, your info does this thread well. One very tiny thing in aesthetics though; png! <3

jere wrote:EDIT:
Seals of the Pure: Does not affect the 5 stack damage of SoV.

This is a pretty serious nerf to SotP. I just ran the code based on the updates that you and Jasari posted. Here's the current incarnation of the SoV code:

Code: Select all

Vengeance_DOT = (0.013.*SP + 0.025.*AP).*5.*5.*... %inherent for 5 ticks of a 5-stack
    (OHWS.*SotP.*Crusade.*GoSU)* ...    %talents
    (CoE.*SanctRet).* ...               %buffs/debuffs
    T82pc.* ...                         %T8 2-piece bonus
    boss_resist_reduce;                 %boss resist

Seal_of_Vengeance = 0.33.*(weapon_dps.*weapon_base_speed).* ...  %inherent
    (100+(melee_crit + HotC).*Crit_Meta - boss_mcrit_supp)./100.*... %crit
    (OHWS.*Crusade.*GoSU).* ...   %talents/glyphs
    (CoE.*SanctRet).* ...               %buffs/debuffs
    boss_resist_reduce;                 %boss resist


%Seal extended DPS for melee swings that's independent of rotation
Vengeance_DOT_DPS = Vengeance_DOT./15;

SoV_Melee_DPS = Seal_of_Vengeance.*(100-boss_avoid)./100./(player_swing_speed./(1+Reck));

SoV_HotR_Proc = Seal_of_Vengeance.*(100-boss_avoid)./100;

The only changes I made were the 5-stack proc (Seal_of_Vengeance) not being avoidable and not being affected by SotP.

Heres the output of the talent point simulation:
Setup:
Rotation: 969
Seals: SoV
Glyphs: V only
Talents: Various, in combinations to isolate the benefit of each talent individually.
Gear: Theck's "WTF Double Vanq again?" set

Code: Select all

  DPS   TPS
Reck   17.0379   31.9251
SotP   13.3321   36.9477 (vs  23.5697   65.3195 if it affected the proc)
Conv   18.2894   43.6381
Crus   29.7968   74.4269
TbtL   49.6172  137.5060
OHWS   88.6307  221.5690

or in convenient graphical format:



The second set of numbers I gave in for SotP in the table are what you get if you let the 5-stack proc get the benefit of SotP. It's nearly a factor of two, and makes all the difference in whether SotP is better or worse than Conviction. Perhaps the most interesting thing here is that Reckoning and SotP only differ by around 5 TPS per point, and Conviction isn't far ahead (12 TPS ahead of Reckoning). This still means there's little reason to take reckoning over SotP, but the additional points required to unlock Conviction mean that overall it's less efficient than the other two talents if you aren't going to push for a Crusade build.

To see this more clearly, let's look at the plot of reasonable talent configurations. I've changed the default talent configs so that the sample is a little more representative of what we might see in practice. Since we now have two 2-point utility talents vying for attention with Conviction, I've included setups with 1, 3, and 5 points in the talent. In addition, I have a build that uses 1V+3C with 3 points in Reckoning.

Setup:
Rotation: 969
Seals: SoV
Glyphs: V only
Talents: Various combinations
Gear: Theck's "WTF Double Vanq again?" set



Some interesting notes about this graph:

• 5V > 5S, unsurprisingly, but it's close enough that 5S>4V. However, 5V takes 4 "wasted" points in T2 of Ret, so it really takes nine points to do the same job that 5 points in SotP manages.
• On the other hand, the 3V+3C build is roughly equivalent to the 5S+5R build. In this case, each build requires 10 points in total (6 in ret plus the 4 "wasted" ones), making them exactly equivalent in terms of efficiency. However, the ret build also unlocks PoJ or Vindication, giving you an extra utility talent.
• 5V+3C is still the best threat you can get (though I suppose technically 5V+3C+3R is higher).

I like the way Jasari divided the builds up by the number of utility talents they allow, and will likely re-do this plot later to try and incorporate that idea. I just didn't have time to get to it tonight.

One final plot for tonight: To compare SoV to the current implementation of SoR and the "expected" implementation of SoCom (i.e. exactly like SoB but weaker), here's the plot of each seal with its most useful threat glyph:

Setup:
Rotation: 969
Seals: SoV, SoCom, SoR
Glyphs: V for SoV, J for SoCom, R for SoR
Talents: 3V+3C (3/5 Conviction, 3/3 Crusade)
Gear: Theck's "WTF Double Vanq again?" set


As you can see, each of the other seals lags SoV by over 1k TPS. Without some major changes, I don't see anyone using either of these other seals for much of anything.

In fact, just to be fair, let's calculate TPS for SoC and SoR with no glyph and SoV with the Vengeance glyph for the 3V+3C spec. This represents a situation where you're glyphed V, but choose to use one of the other seals for some reason (maybe you don't want the DoT):

Setup:
Rotation: 969
Seals: SoV, SoCom, SoR
Glyphs: V for SoV, none for SoCom or SoR
Talents: 3V+3C (3/5 Conviction, 3/3 Crusade)
Gear: Theck's "WTF Double Vanq again?" set

Code: Select all

Seal    TPS
SoV       8.11k
SoR       6.82k
SoCom     6.80k

As you can see, SoR is still ahead of SoCom for a fast weapon. There certainly seems to be no reason to waste a point in SoCom, since SoR is going to be better in any reasonable situation.

Theck, one thing I noticed when coming up with my proposed viable specs was that there are not enough worthwhile talents to not spec deep enough into ret to reach Conviction.

For example, even picking up SotP, you have enough talents left over to get 3/5 Conviction: http://www.wowhead.com/?talent#sVZV0xA0uGusIufdxf0h

The only way this isn't the case is if you choose to pick up Divinity or spec deeper into holy to pick up imp LoH.

With this in mind, the questions I found always come down to:

3/5 Reckoning or Divine Sacrifice + Divine Guardian?

2/2 Vindication or 2 points in Conviction?

2/2 PoJ or 2 points in Conviction?

In the case that SotP begins benefiting the SoV 5-stack the situation really wouldn't change we'd just have to see which is better:

5/5 SotP + 1/5 Conviction vs 3/5 Conviction and 3/3 Crusade (if you're picking up one of either PoJ or Vindication)
or
4/5 SotP vs 1/5 Conviction and 3/3 Crusade (if you're picking up both PoJ and Vindication)

Yeah, I agree. There are still people that like to pick up some of the Prot utility talents though, and that could sink enough points to bring up the question of whether it's still worth pushing for Conviction rather than SotP+Reckoning.

I think I've managed to cover almost all of those situations in the graph as is by including every odd number of points in Conviction. Since Crusade is always better than anything else, it seems logical that it's only worth modeling all or nothing for that talent.

With the graph, it's really hard to fit more than 9 specs without them all blurring together. I can switch to bar plots and mess around with the axis labels to make them smaller, but I'll deal with that when we get to the point of legitimately needing more than 9 representative specs.