Set threshold, when over do hs instead of a mainhand hit. Like you do ingame.

Another "error" I noticed when adding weapons. The predefined flieds like "Mainhand""and "Mainhandstats" in the Warrior_ITEMS.xls do not comply with the ranges which are made purple. Though not fixable it could cause problems with the lesser gods of spreadsheets.
The hit cap and crit cap cells reference to set 6 instead of their own set.
Additionally I would also prefer a single sheet instead of multiple.

Yes, it is a rather complex problem to solve. I've thought about first calculating each proc's estimated uptime independantly of each other to use as a starting point. Then applying them all together and recalculating a few times over to get final results. The recalculations would be used since many procs will cause a change in attack speed or rage generation, which will change the uptime of other procs. While this would provide a somewhat closer approximation, it still wouldn't quite do justice to to the human factors, where we try to use certain cooldowns in the presence of certain other temporary buffs. For example, using Bloodlust Brooch + Haste Pot when a shaman uses Heroism/Bloodlust. They scale each other greater than just calculating the average benefit of each buff.

Im not sure about in a spreadsheet, but from a programming language performance view, doing many recalculations should still be fast, because programming languages are generally extremely fast at doing math, which is the majority of whats going on here.

I think for random, non user controlled procs like flurry, weapon enchants, trinkets etc... it could be feasible to calc an estimated initial uptime for each. Based on that, calculate the average overlap time between each proc(because they scale each other when overlapped). Calculate each different permutation's dps, and estimated uptime % of each permutation. These could all be combined, and then recalculated a few times to find a final dps number. The amount of different permutations as you add more of the non controlled random procs, would grow to very large numbers quickly though. Im not sure whether a spreadsheet would be suitable for such calculations.

Some example permutations:
flurry
mongoose
executioner
mongoose + executioner
flurry + mongoose
flurry + executioner
flurry + executioner
flurry + mongoose + executioner

Each permutation would need a percentage uptime. The sum of all permutations would need to add to 100%.

Personally, im not sure the complexity is worth the slight accuracy gain when even better accuracy can be achieved using a model like Doc's new addon.

Yeah for most things I don't think its worth the extra time as I don't think it will change much from the estimates. The only one I do think would have significant impact is the rage system itself, which is nothing more than a proc off white hits. But I dont know how to calculate the odds of not having enough rage for a cycle/dump.

update:addon
You could test talents+buffs, if you want to help. That means, see if you can get an error while clicking around them or other strange things that happen.

Hi, I'm a new french here so sorry for my english.

First of all I would like to thanks all people who works on the sheet it very help me to my stuff orientation, gameplay etc

That said, I post to bring my little stone to the edifice for the addon :

I got a minor error (warriorsetup.lua:655: attempt to index field '?' (a nil value)) if I click straight to the fury tree, also there is 2 little mistake for imp.slam and precision : we can spend 5 points instead of 2 and 3 respectively.

Now question :
-Is the DPS supposed to change when we active some buffs/talents or is there only the interface for the moment?

-When we calculate the dps, it's never the same (i suppose this is because it's a random calculation between the min-max weapon damage?) but my question is, is it possible to make an average to always have the same dps?


Sorry if it's newbie issue/question but i don't have so much knowledge in this domain.


PS to grim : Is it possible to add an option in the sheet to enable/disable the EXECUTE mode ?

Ah, now I see. Thanks for explaining. It looks like it is correct for 2h and not accounting for the talented buff for DW. I've added the spec multiplier to the DW WF.


The only way I know to do something like what you suggest is with macros, and I am trying to get away from them for compatibility reasons. It's already going to add a TON of extra bulk just doing SEP and separate execute weapons. A brute force method isn't really something I am considering for this issue. I'll think on it and see if I can come up with anything to increase the accuracy, but I am not hopeful.


OK, this part is really just a stop-gap. I was rather dissatisfied with the way execute was handled, as it broke down rather severely under certain conditions, and was never terribly accurate under the best of terms. What I put in is a formula that is more....dynamic, for lack of a better term. The basic principle behind it is that the difference in weapon speeds is the more important factor, not absolute weapon speed. ie: a 2.6+2.7 combo will certainly generate more executes than a 2.6+2.6 combo, or any other > 2.0s matched pair. Previously execute CD was a simple average of the two, and before that, a constant 1.5s.


As to you specific inquiries, I'm going to keep it fairly general. I take [fight length]*0.2/1.5 to see ~ how many sub 20% GCDs there are. I then subtract the avg # of MH swings in that time. I take the remaining GCDs and multiply that # by OH crit chance. I then take the sum of that # and # MH swings and divide execute time by that.
[fight length]/(MHs+(tGCD-MHs)*OHcrit)
It's a hack-job, but it is a reasonably close match to the real numbers. There is a lot that hasn't been accounted for with it, but it's a big project to fully model it, and easier done from scratch, in a sheet I wrote from the ground up, which is what I am working on. Just want to get the current sheet as error-free as I can, then I will concentrate on the re-write.

Something like this will be available in the re-written version. For now, in the upper-right, you will see output for white, cycle and execute DPS. You should be able to use that data to figure out what you need to know, I hope. Please keep the ideas and requests coming.


The current SEP calc is using a similar principle. It adds in that particular stat, records the result, removes the stat and repeats for each stat. Just that simple action takes a good five seconds to execute. I don't want to know how much time this would take. Plus, it requires macros, which I am staying away from moving forward. I am no Excel ace though....I never even used it until a couple months ago when I first came across the original spreadsheet, so everything I know is what I've learned from reverse-engineering Doc's work.


My thoughts exactly. When it is ready, Doc's sim will be the last word in accuracy, as lua is much more of a programming language, while Excel is more just a way to do a few neat tricks with math. Excel just doesn't have the flexibility to do some of what needs to be done. It does have it's uses though, and I hope to be able to maintain the project at an acceptable level of quality.


I'm aware of this, but there are some weird things going on when I attempt to change it. I'm trying a few things to deal with it though.
*edit* Finally figured it out and expanded all categories to 100 rows. Think it fixed the


/sigh... I did it again. The flag for the hourglass was triggering both procs, and the flag for talisman was triggering neither. Should be good to go now.


My apologies for the continued issues. I hope to have an error free version out so you all don't have to keep d/ling the thing for awhile.


Warrior DPS v312


Oh, and as to distance to cap all showing set 6, that was corrected previously and should be good in the last few versions. I just checked and it looks right, unless I'm overlooking something.

This either never happened or was reverted, it's still "One-Hand", if anyone wants to update the spreadsheet.

edit: misread his post, he is correct, but I still can't select The Decapitator as my off hand.

couple things with the spreadsheet:

i think cell K42 should be empty (+haste rating for helm socket bonus).

maybe add "T6" under set for the Onslaught bracers/belt/boots, so they will count towards the T6 set bonuses.

Thanks for the nice work once again, Grim.
I just downloaded the sheet and started updating the Warriors_ITEMS.xls with some copy and pasting from an older version and noticed that [Hard Khorium Choker] was listed with 160 ArP instead of 150.

Just thought I'd mention that for completeness sake. <_<

Bug with the proc of "Band of the Eternal Champion", 0 AP.

This is my first post, and as such I figured I would use it where I spend most of my time when I come to the forum.

There are a few things that I was wondering as far as the item sheet goes. On trinkets, or other items that have a proc of attack power or other ability, is there a way to average out the proc per minute; and/or use with cooldown, and in turn adding in a static number of the stat that procs? For example, Icon of Unyielding Courage has 30 hit rating and a use to ignore 600 armor for 20 seconds, can it be averaged out to where the armor ignored would be equal to the armor ignore that would be seen over time? I was also wondering if that could be applied to the Hyjal exalted ring as well as other AP trinkets such as Berserker's Call and Bloodlust Brooch. It's easier to manually type in things like Solarian's Sapphire, where it adds 70 extra attack power to battle shout, since that is up 100% of the time, its a static 70 attack power increase, or 77 attack power for fury.

Things like Dragonspine Trophy that show they are very low dps trinkets would benefit greatly from an averaging out of proc's and give an overall better image and idea of different gear ideas and uses.

I try to use your spreadsheets and MaxDPS.com - News to get a general idea of gear and gemming and what gear that I can use that will let me do the most efficient dps.

Thanks for your time.

The spreadsheet seems to be quite accurate, according to what I'm getting from my simulator. It's just difficult to calculate SEP, since you have to run it for a looong time to get a constant dps value.

I tested the simulator and the results vary a lot. This makes it a little difficult to estimate what gear comes out up front.
Does anybody has an idea to that? I don't think the results are incorrect but as I stated it's hard to make a decision based on the simulator right now.

I know it's still in development so now is the time to find the right representation for the warrior dps. Do we need a graphical representation? Or some more avanced math?

I mentioned in the German-Warrior-Forum that a percentage value should be given, to simulate random events (walking 20 secs= 20 secs no dmg = flurry and other procs fade). This makes the dmg even harder to evaluate. I really think this needs some kind of solution. The spreadsheet only gives static results, but they are useful to some point. For now I see it's hard to come to a conclusion between similiar items when they differ only by a few ap/str and haste/crit.

Assuming the simulator is accurate and run enough times to get a statistical average, it would give useful SEP values. The sheet assumes you always get the average of everything, which can only be fixed by more advanced math or huge sheets, but it is probably close. I would use values you get and just fudge a bit towards hit and a little less on procs to account for luck.
Not sure what you mean by graphical representation, dps is a number. Though recount style graphs in the addon would be sweet.

Right now it would take ~7-8min to get one dps value, that's almost an hour total.
10000 hours (that's over a year) of combat per run, one for comparison and one for each SEP we'd like to know.
Maybe there is a more constant way to get an average of n values.

I realize this thread might be dead, but with the lack of a better thread to go to, I'll ask my question here.

When I use the dps spreadsheet/simulator, Hast has a higher benefit to arms than crit.
It would seem that logically just the opposite is true, as crit will increase flurry up time and and haste will mess with slam rotation.

Is this independent of other people's experience?

Hello. Seems i have same question as above. Why in this spreedsheet haste is so good? I was always thinking that haste and hit are pretty similar and those stats aren't that good for fury warrior.
Thanks

PS: Sry for my english.

I think the quote "haste will mess with slam rotation" is a bit of an overgeneralization. Before I begin, here's a link to a nice (and short) read on the slam rotation ( http://www.honorbound.se/Slamdps2.pdf ). There, it explains that the theoretical minimum weapon speed for a slam-rotation is 2.5 seconds. Even with flurry up, a 3.5-speed weapon has an effective speed of 3.0 (assuming you have 3 pts in flurry -- a standard 2h build).

On the other hand, crit provides diminishing returns (w/respect to flurry uptime) as crit increases. So, the spreadsheet is likely valuing haste over crit because you already have a reasonable amount of crit, and because your effective weapon speed (inherent weapon speed - flurry) is still above 2.5 seconds. When your haste drops your effective weapon speed below 2.5, the spreadsheet should start severely devaluing it. (also, make sure the spreadsheet knows you're doing a slam-rotation. I've made that mistake before)

Hope that helps

A friend just pointed out to me that the spreadsheet doesn't appear to make any calculations for deep wounds. Changing the talent points in deep wounds doesn't change DPS at all, and the cells don't appear to reference it anywhere.

solarian saphire is missing the whole point of having the trinket in its stats, 70ap :-)

Or is it moddeled somewhere deep down inside the sheet if equiped?

I put in a cap on how fast a slam rotation can be, based upon more or less exactly what you are saying. Once +haste (in it's varying forms) gets you to that speed, the sheet will not allow your rotation to get any faster, therefore seriously handicapping the DPS increase from +haste. (in effect, it would then increase your white DPS only, kind of like the opposite of going beyond 9% +hit in a DW build)

To answer why haste is shown to be so valuable in SEP, there is a simple answer. It takes 22.08 crit rating to get 1% more crit, while it only takes 15.76 haste rating to gain 1% of haste. This means that even though 1% crit is more valuable than 1% haste, it does NOT follow that 1 crit rating is more valuable than 1 haste rating because 1 crit rating is more "expensive" in terms of SEP. Just remember, SEP is one rating point, not a percentage point. There is a very big difference.

This is already how these things are handled. At present, the SEP column in the item stats does not reflect any on-use or proc DPS, but the actual DPS output on the spreadsheet does take this into account. In the next full-scale release this is one of the things I intend to resolve.


And....since I don't see any fresh errors brought to light while I was away, I'm going to post a "final" version of the sheet in it's current form. Work continues on the new version, and I think I am getting close to being able to release an alpha quality version, suitable for inspection by those who would be interested in checking my math/offering suggestions. Maybe by the end of this week.


WarriorDPS_v313



Oh, btw, anyone who has d/l'd the alpha quality tanking sheet.....any input on it?