AMR Sim 26% lower than SimC/Bloodmallet Arms Warrior ilvl229 (BiS)

The Bloodmallet SimC based datasets show Venthry for example, being significantly further ahead of any other Covanent.


(set max renown 35)
Venthyr soulbinds: 6,400 - 6,500 dps
Night Fae: 5,800 - 6,000 dps

The item sets are listed that are used for bloodmallet, when these items are copied to the AMR simulation, so the items are identical, there is a substantial different in the values calculated when using the Venthyr Soulbinds.

60 Pandaren Arms Warrior
4,769 DPS

Considering they are both patchwerk fights, I’m not sure why these are so drastically different, even if AMR tries to add in player error rather then mathmatically perfect sims - this seems like far too large a difference. I’ve adjusted the fight length on the AMR simulation as I can’t seem to find the SimC/Bloodmallet fight length, but this does not appear to make a significant different in the DPS values.

There is probably some settings that are not matching up exactly - it can be very difficult to set up the exact same simulation in the two simulators. When they are set up the same, usually the results are very close.

Well, yes that is my thought as well. But does this not pose a problem of deciding what is reasonably accurate if two methods of analysis do not agree? (within a reasonable amount).

I wouldn’t mind if at least the normalised data would show similar trends, but this doesnt happen either.

For example if you normalise the SimC data for venthyr at 6450 (1.00), and use a value of 5900 for night fae then the relative value for NF is 0.914.

Using the base gear from AMR as an example (not the same gear as above):

Ven 4,719
NF 5,160

With AMR the NF value is 1.09 for a similar set of gear (normalising should take out the differences seen between fight length etc.). Normalised data between the two simulations seems pretty extreme in conclusions. In terms of trusting which simulation provides an accurate representation of output - without a 3rd simulation method, can either be trusted?

That output seems pretty over the top for venthyr that you linked from bloodmallet. I’m not sure how they are getting numbers that crazy-high.

If you can see anything wrong with our simulations, let me know. I do a lot of testing and making sure everything is working correctly. I would be very surprised if I was off that far. I really don’t think we’ll see arms warriors crushing it that hard in BiS gear.

I’ve been simming my own venthyr warrior on raidbots and AMR and the same big difference is a occuring, 4.4k to 3.9k difference. This is specifically with talents Sudden Death, Massacre, Avatar and Ravager combined with Signet of Tormented Kings legendary. Theres something going odd going on with the talents as AMR prioritises Dreadnaught over Ravager (Dreadnaught choice does same overall DPS), but with a condemn build, the legendary would set off bladestorm with this lego, meaning no condemns can be performed as opposed to ravager which can be cast simulatenously as Condemn. Picking Dreadnaught for SimC brings the Sims more in line, 3.9k vs. 4.1k. SImC is gaining a lot from Ravager.

  1. The overall breakdown of spells seems similar but a much higher proportion of damage is coming from Condemn for SimC, 47% compared with 38.5% for AMR. It looks like SimC is prioritising Condemn if colosal smash debuff is up and if sudden death procs over other spells, which might account for the extra counts, potentially? If not captured in AMR rotation.

300 second fight for both, Condemn is used a total of 77 times SimC, 58 times AMR. In comparison, Mortal strike is used 26 times SimC, 29 times AMR
Overpower 47 SimC, 54 AMR. AMR more resource starved? Or priority difference.
Spell priority AMR:
Spell Overpower
Spell MortalStrike if CanRefreshDot(DeepWounds)
Spell CovenantCondemn
Spell ExecuteArms
Spell MortalStrike

  1. Condemn Normal and Crit averages are ~4.5% higher for SimC (tooltip says 207 shadow dmg on AMR, SimC shows 2*103.4%*AP as shadow. Might be a tooltip thing but might be something? Checked the spells under Arms in AMR and there’s modifiers I can’t quite understand but that could be fine and just the coding used for ARM).

  2. In comparison, Mortal strike and Slam are <1% different between sims.

  3. Ashen Juggernaught has an uptime of 62% for AMR, 68% for SimC.

LICHTE (SimC)

4,428 DPS

Lichte

60 Tauren
Arms
Warrior
3,918 DPS

The best in bags gear selection also tends to not favour Kings legendary, which Bloodmallet has a 10% increase in DPS (for BiS anyway, obviously numbers will be different with lower gear levels) but there is something there considering the legendary option + ravager is giving big gains both from my personal sims with my own warr and the BiS comparison combined with the many more uses of Condemn for the same length fight in SimC vs. AMR. Without the lego + ravager combination, the sims are very close.

Hope this helps.

I downloaded the latest version of simc and did a direct 1:1 comparison.

I can make a few tweaks to the rotation to get a little more damage from execute/condemn - it won’t make up the huge differences you are seeing here, but should add something in the range of 2.5% DPS, so worth doing. I don’t think the gear suggestions are going to change noticeably, but we will recalculate to be thorough.

One big difference is probably that we do not have windfury totem enabled by default, simc does. You need to turn that on in the settings to do comparisons, since Arms benefits more from windfury than any other spec. That’s easily 8-9% damage difference. Our gearing strategies assume you do not have windfury totem.

The other difference comes from the scripts being used. Both AMR and Simc try to account for the start of a fight and end of a fight being “compressed” due to openers and execute phases. So an ability like Condemn which can be used while above 80% or below 35% (with talent) will have more or less value depending on how quickly your team takes the boss through those health percentages. A boss won’t be above 80% health for 20% of the fight, and won’t be below 35% health for 35% of the fight. It will be less than that. We are compressing it slightly more by default than simc, which results in less uses of Condemn and less uptime on Ashen Juggernaught. Which is more realistic? Who knows - depends on your team and the fight. It is configurable in the boss scripts.

I had a weapon drop since the last sim which accounts for the increase in overall dps compared to the previous post, but your recent changes seem to have done the trick.

Your WF suggestion I hadn’t considered in the AMR sims and I do personally always have WF for our raids.
4845 dps SimC - 77 casts
AMR reports
4,373 without WF (condemn 42.5%) - 64 casts
4,674 with WF (condemn 43%) - 70 casts, so that seems pretty close to me.

Condemn accounts for 46% of damage from tonights raid (real data from the patchwerk like devourer and shriek encounters) and is matching much better with the AMR sim now at 43%, Simc is 46%. Was this compressing the executes phases less or a modification to the script allowing more condemn casts with ravager? Thats one noteworthy thing from before that even with a ravager+King build the condemn casts were pretty much unchanged from unhinged/dreadnaught build for AMR. Ravager offers me almost unlimited rage for its duration and if kings procs ravager as well I can perform a heinous number of casts during >80%/<35%.

Your WF suggestion I hadn’t considered in the AMR sims and I do personally always have WF for our raids and its likely the biggest factor.
4845 dps SimC - 77 casts
AMR reports
4,373 without WF (condemn 42.5%) - 64 casts
4,674 with WF (condemn 43%) - 70 casts, so that seems pretty close to me.

Yeah raid composition will definitely alter the proportion of the fight spent in execute phases, MM hunters, other Arms warriors, rogues, retri palas etc. with executes. Hard to sim around.

Appreciate the changes, I prefer using AMR systems over Raidbots anyway!

Thanks.