Open a new issue on Rawr site and upload the xml where this happens and I'll take a look.

Does it do a complete recalc though? Try turning off incremental optimizations.

I wish I had better news, but the reality is that in b14.1 the reconstruction only gives consistently good results for non-CS arcane builds. Hopefully for b14.2 I can say the same for CS arcane builds. It looks like in your specific case the main problem is with drums. I haven't done much testing with drums so far since there's many other things to get working first, but it's on the list to do.

For fire one of the main issues is that having partial uses of mana gems/pot and caps/destro pots isn't really conducive for converting into a real executable sequence. Having them partial is good for comparisons since it gives you the indication of trend even though the actual increases come in more discrete units. But for making a sequence half a gem and half a cap is neither here nor there. The current idea that I'm planning to implement is to add another option in SMP that forces integral consumables. My guess is that it should work much better for sequence reconstruction purposes. Most likely for b15 release.

Here's how incremental optimizations work since it looks like quite a few are confused about its purpose. There are two phases of computation always. One is computing things for currently equipped gear. Then all computations for gear comparison charts are done. The computation for currently equipped gear is always the same, regardless of whether incremental optimizations are enabled and should always give an accurate result. The computation for gear comparisons however depend on incremental optimizations. If they are enabled the computations are done assuming that spell cycles and cooldown stacking will be similar to the one that was computed for the currently equipped gear.

I am looking at adding swipe info to the bear threat model and I am coming up with some surprising numbers that don't seem to agree with a lot of posts regarding swipe vs. lacerate.

Lacerate Threat:

Lacerate threat has 3 components:

• Initial attack damage - damage is 31 + ap/20 + idol/t5 bonus. It is modified by mangle and and naturalist, can crit and is not mitigated by armor (but can missed/dodged/blocked/parried). Finally, the damage is modified by bear form threat modifier (with talents) of 1.45
• Innate threat - flat amount of threat (285), modified by bear form threat modifier of 1.45 Can be missed/dodged/parried. I am not modeling block currently, but is it true that the innate threat still applies on a block?
• Bleed damage - Bleed damage 31+ 5*(ap/20 + idol/t5 bonus) every three seconds (modified by mangle and naturalist). Threat is 20% of the damage * bear form threat modifier.

Swipe Threat (One target):

Swipe Threat is all damage (plus a small amount of threat gained through crits generating rage). Damage is .077*ap + 92 + idol. Swipe can crit, is modified by naturalist, 4t6, mitigated by armor and can miss/dodge/block/parry. The overall damage is modified by bear form threat modifier.

Damage portions for both abilities are modified by ferocious inspiration and blood frenzy.

In my current armory gear (including by charm of swift flight, lol) I am calculating Lacerate spam as 421 TPS and swipe spam as 255 TPS (against a single target). Here is am considering "spam" to be 3 attacks every mangle cooldown or 1 attack every 2 seconds.

Sounds about right. Swipe gains a ton from raid buffs, lacerate doesn't. And you don't have 4T6. Sunder is huge too.

I've noted the TPS numbers displayed in Rawr to be very far off my estimates. There are a few differences I see from your assumptions.

Lacerate initial damage = (31+AP/100+idol) * damageMod * bossArmor
Lacerate bleed damage = (31+AP/100+idol) * 5 * damageMod * mangle
Swipe damage = (AP/14+84+idol) * 4t6 * damageMod * bossArmor

damageMod = naturalist * bloodFrenzy * sanctityAura * ferociousInspiration

I've also seen threat modifier reported as both additive (1.45) and multiplicative (1.495). From warrior threads I thought it was confirmed to be multiplicative, but it looks like Omen implements it as additive also.

Given these numbers in my current gear (6t6), I estimate Lacerate spam at 204 TPS, Swipe at 223, and (1xLacerate/2xSwipe) at 246.

Yes, your threat calculations were the same as mine. I was only refering to the base damage calcs.

There were discussions in the Feral Megathread about Lacerate threat some time ago. I confirmed the AP/100 number based on in game testing using various AP values on the ogres above Shatt. I'm not 100% sure, but I thought it tested armor as well. It should be easy to test when i can get back in game if the iniital damage is increased by Mangle.

The Lacerate AP/20 number you might have seen is the same as 5*(AP/100), which is correct for a 5 stack of the bleed. However, the static 31 value (and idol increase) is also multiplied by 5 also.

Yeah, i did some simple testing and it is looking like you are correct. The initial damage is mitigated by armor. I did think it curious that ap/20 is ap/100 * 5

Okay, I have made some changes to the bear threat (and dps) model and I think it is a lot more accurate. I was only able to test it against some level 70 elites in Shadow Labs Heroic, but the observed values were coming out very close the the predicted values.

I am not taking into account glancing blows, but I have added it to my list so I won't forget it. Block is still not modeled. I would also like to have a way of configuring the dodge/block/parry/armor of the target.

Coming soon, actually.