I have been getting questions regarding what to choose a lower quality but faster offhand or a higher quality but slower one. While the simsheet currently does not allow you to choose a different dagger to put DP on (it assumes WP/IP and DP daggers are exactly the same), I used the simsheet to back out some relevant numbers that can help make the decision. I did it for 245 ilvl gear so that it's more relevant to the majority of raiders.

Step 1: Calculate an estimate of the value of 0.1 speed by doing the following:
a. Estimate dps with 1.5 speed offhand without swapping
b. Estimate dps with 1.4 speed offhand, holding dps constant, without swapping and measure the dps increase
c. Repeat the first two steps with weapon swapping and calculate the marginal increase in dps value of swapping due 0.1 speed increase

For combat, changing speed from 1.5 to 1.4 increases DPS value of weapon swapping by 46 dps.
For mutilate, changing speed from 1.4 to 1.3 increases DPS value of weapon swapping by 22 dps.


Step 2: Calculate uptime of DP dagger for each spec

For combat, the uptime is 45%, and for mutilate the uptime is 30% (using 1 second reaction time and 3 second swap-back time. These numbers take into account all side-effects of weapon swapping for real combat, such as, global cooldowns and restacking of DP if it falls off.

Given the uptimes, and a rough estimate of 0.9 dps per point of AP with that level of gear, I can back out how much extra attack power is necessary to overcome the dps advantage of 0.1 speed. For combat, it would take about 114 AP to overcome 0.1 speed difference. For mutilate, it would take about 81 AP to overcome that difference. Obviously stats can differ by more than attack power, but this is at least a good starting point.

Just to give you an idea, going from Kinetic Reaper (ilvl 219) to Golem-Shard Sticker (ilvl 232), using 3.2 dps/point values for AG-based mutilate, is 35 dps worth of stats difference. Going from 245 to 258 steel bladebreaker is about 60 dps worth of stats difference. So it seems like 1 item level is enough to overcome the speed difference, especially at later levels. On topof that, if you are considering a slower offhand with more stats for your DP dagger, if it has stats like hit, expertise, and haste, the value of these stats for poison swapping is even higher because they help restack DP poison faster, so it's even easier to overcome the speed difference.

I have a question about the newest setup with AP gems + dual mongoose. I found this to be the highest setup for my mutilate build, I ran the sim several times and kept seeing AP gems + mongoose > other combinations. What I can't figure out is WHY mongoose overtakes berserking. AP gems makes sense to me - more of my damage is poison and poison damage favors ap, not agi. Before weapon swapping agi was BARELY > AP so the gem change is logical to me. If this is the case, what puts berserking behind?

Playing with some EP values quickly here...

Mongoose = 120 agi + 2% (66 rating) haste
Berserking = 400 ap

Assuming the EP values of both AGI and haste are 2.0, (anything higher means I'd gem for that instead of AP) it puts mongoose at 120*2 + 66*2 = 372, which is below berserking. Help? I thought I read that the procs were each 1ppm.

There is a multiplicator effect of the haste component of mongoose proc. The frequency of mongoose procs is higher if your haste is higher. That's the nature of the new ppm mechanics discovered by Aldriana and Latito. Which means that the 2% haste proc increases the chance a new mongoose proc will occur, thus the uptime of lightning speed from dual mongoose is higher than the uptime of berserk from dual berserking. So to do a proper comparison, you need to compare 400*Ub vs 372*Um, where Ub is berserking uptime and Um is mongoose uptime. The finding that mongoose comes ahead of berserking tells me that Um/Ub is more than 400/372=1.075.

That said I remember a discussion in Aldriana's thread about gemming haste vs crit. If you get too much haste, it's EP value diminishes, and you get a recommendation to start gemming crit. Perhaps putting a few haste gems instead of AP gems would yield a slight dps increase. I have not checked that. As soon as we are done with polishing the swapping mechanics, I will run a script to determine EP values and it will become clearer.

EDIT: just tried replacing 20ap/10 haste gems with 20 haste and noticed a 10 dps decrease, significant at 10% so it does seem like EP value of haste is just shy of 2. So the answer is in the uptime mechanics I described above.

@Aarcani, yes that a similar issue, and I think the answer there too is to use the lower quality 1.5 speed dagger as the DP dagger because 0.1 speed usually cannot overpower one ilvl worth of stats for mutilate.

Um... not entirely. The uptime difference between Mongoose and Berserking is actually relatively trivial; the point you're missing is that 2% attack speed from Mongoose is not the same thing as 66 haste, as the two bonuses apply multiplicatively to each other. Thus, if you have, say, 328 haste rating (10%), the 2% attack speed bonus is actually worth a little over 72 haste rating. If you have more haste rating, the Mongoose Haste will be worth even more.

That said: if Agi is running below 2 EP, it should be pretty hard for Mongoose to catch Berserking, without a) a simulator bug or b) the value of haste truly going through the roof - I'd expect it would need to be over 2 before Mongoose would catch Berserking with Agi EP < 2. So if b) is not the case, I'd take a look around for a), myself.

When I try to download your spreadsheet, it gives me the error the server was not found. It could be time of day or something, but I thought I'd mention it.

I'm getting results I don't quite understand with mutilate and weapon swapping active. My current gear t8 4pc but I get 0.6% less dps using rupture. This is rather hard for me to believe considering the set bonus.

The only things I change between rupture and no rupture is 2 pts from blood spatter to fleet footed, boot enchant from cat's swiftness to icewalker, glyph from rupture to tricks and of course turning off rupture on the DPS strat tab. Is there something I'm missing or does the higher envenom uptime with dual IPs really overpower 4pcT8 rupture?

I was doing this with no murder.

It seems to me rupture might be a dps gain with that set bonus if it was only used if envenom was already up, I don't quite see how to set a rule for that. Lowering the priority of rupture or set it to only be used if say 4 seconds remain on envenom. This way I think it would have less impact on envenom uptime.

You made a good point about multiplicative nature of the haste buff from mongoose. However it only reinforces my argument about the nature of mongoose buff. I do not think there is an error in the simulator with these mechanics, you just oversimplify the effect of haste proc on uptime table. It's actually quite interesting and involved.

First thing I am going to say is that applying average method to figure out average speed, assuming 100% snd uptime, 21% CT uptime, 45 seconds of heroism, and 1 haste pot, I get 2.39 (using 51/18/2 spec). In reality it is probably slightly higher due to multiplicative nature of buffs and stacking of speed pot and heroism in the simsheet, but let's work with the 2.39 value. Based on this value, 2% extra haste is equivalent to 77 haste rating. When you have both mongoose buffs up, that's equivalent to 156 haste rating.

Now I oversimplified when I was talking about a single number for uptime of mongoose and berserking. Obviously the full distribution is comprised of uptimes of 1 buff, 2 buffs and 0 buffs. The uptime of no buff is in fact trivially lower for mongoose, 30.5% vs 30.8% for berserking. However, due to the mechanics I described in my previous post, 1 proc increases the chance of another proc, so the probabilities of 1 and 2 buffs up are drastically different for two enchants. Thus uptime of 1 buff is 29.2% for mongoose vs 48.6% for berserking. At the same time probability of 2 buffs is 40.3% for mongoose and 20.6% for berserking. I get all these numbers from the simulation, I have not thought of a more reliable way to model the chain nature of the proc uptime.

Therefore, mongoose is worth on average 77*29.2%+156*40.3%=85.3 haste rating and 120*29.2%+240*156*40.3%=131.7 agility. At the same time berserking is worth 400*48.6%+800*20.6%=359.1 attack power. If agility were at 2 EP and haste were at 2EP, mongoose would be worth 434 EP vs 359 EP for berserking. However, I know from testing that agility falls short of AP gemming when you do weapon swapping in BiS gear, and I also tested that haste is below AP as well, so mongoose is somewhere below 434 but there is plenty of room for it to be above 359. I will know better EP value of mongoose when I run EP value estimation.

So, to be clear: you're suggesting that a 2% haste buff - thus increasing the number of procs in a given amount of time by 2% - is responsible for a 20% swing in the rate of double-uptime periods, and you *don't* think that's indicative of a bug? I mean, it depends to some extent on the rate of yellow attacks in addition to white, but given that the expected uptime of the OH Mongoose proc is only about 40% to start with, advocating that both will be up a full 40.3% of the time (100-30.5-29.2) implies that OH Mongoose (virtually) never procs without MH Mongoose already being up - and when it does, MH Mongoose always procs within seconds. And to me, that doesn't even pass the most basic of sanity tests.

Again: they are going to be up together a hair more often that you'd expect from looking at either by itself, but there's no conceivable way it could or should be of the magnitude you advocate.

Unless I'm missing something, deadly brew doesn't add any dps, it just makes your poisons also procc crippling.

Yea I see that I got wrong numbers last night trying to calculate uptimes of 1 buff, 2 buffs, and 0 buffs. But at the same time I doublechecked the proc mechanics and there is absolutely nothing different between mongoose and berserking. It's same attacks that proc them (all landed yellow damage, both mainhand and offhand for mutilate, and all landed autoattcks) and same chance to proc. Berserking procs 400 AP that goes through its proper multipliers, and mongoose procs 120 agility that goes through its multipliers to give AP and melee crit and 1.02 haste for 15 seconds.

I would still calculate the uptimes and check the uptime-weighted dps of each proc before making any conclusions.


EDIT: I rerun uptimes. With these uptimes, the two enchants come out with same EP, but I think the haste equivalent of 1 buff and 2 buffs is slightly higher than 77 and 156 because ofstacking of haste, and minor synergies between ag and haste procs. So it's consistent with haste and agility being just below AP in terms of EP and mongoose being slightly ahead of berserking.

I did some testing to see if weapon swapping could proc (possibly) 4 times the berserking buff (it doesnt work), and this got me thinking that if weapon swapping could get 2x Mongoose buff and 2x berserking buffs. Anyone tested this yet? (i dont have enough gold to test it!) Please and thank you to whoever gets this sorted out