[Warrior] DPS Spreadsheet 2.3 and beyond - Page 51

landsoul · 03/09/08, 1:11 AM

Originally Posted by Gruntle

The statement "converting additional rage beyond rotation into additional heroic strike damage to white, and then including the gain in that benefit into other functions, which contributes back into rage generation and additional rage to heroic strike completing the cycle one time" strikes me as a bit strange. The extra damage from HS will be yellow and thus not give you any extra rage. It will also not affect flurry since the HS hit eats up a white hit. The current sheets already converts excess rage to HS. I can't really think of any other effect the extra HS damage would have that would force you to use some kind of circular calculations.

When you are using Heroic strike, it removes the chance of a miss effect occuring on the main hand and lowering the chances to active PPMs and % chances to proc on trinkets, enchants, windfuries, etc. When you use HS, you are effectively minorly increasing the uptime of these procs and effects. While the net of using heroic strike is an overall loss in rage per second over rotation, it changes areas that contribute to the Rage Per Second model, where the values have to be recalculated in multiple passes until it does not change to a certain degree. The only time Heroic strike will not modify these is if you are rolling with 100% white miss recovery.

landsoul · 03/09/08, 8:35 AM

Well anyways, this is what I have been working on off and on for the last 4 months. I cant make a new thread because I don't have permission so I just put it here. It doesn't have all the goobly gob yet its very comprehensive for DW 17/44 and its variants at least.

Landsoul's Fury DPS 1.0 for patch 2.4; Feb 2008

Written in Open Office

sheet
FuryDPS1.0.ods - FileFront.com

prog
download: OpenOffice.org Downloads

I plan to incorporate Arms/Fury and Prot DPS later. It's got a few minor bugs but its very easy to get over them. Please DL it and give any feedback of what you think, want, or if there are any data errors.

Mithgaur · 03/09/08, 6:47 PM

Hi, I was hoping someone could help me with a question

I'm currently using Talon of the Phoenix on MH and Blade of Infamy OH.

To estimate my DPS I use Grim13 stylesheet. What surprises me is that Swiftsteel Bludgeon is giving me higher dps then what I'm currently using. I.e Swiftsteel Bludgeon MH and Blade of Infamy on OH.

I also get lower DPS with the mentioned above if I switch them around. I.e Blade of Infamy MH and Swiftsteel Bludgeon OH.

My question is, how can this be correct? A fast weapon on MH with almost equal stats to Talon of the Pheonix which is a slow weapon, is showing up superior dps.

Talon of the Phoenix(MH) and Blade of Infamy(OH) = 1374,1 DPS
Swiftsteel Bludgeon (MH) and Blade of Infamy (OH) = 1381,7 DPS
Blade of Infamy (MH) and Swiftsteel Bludgeon (OH) = 1373,4 DPS

Any help would be welcome.

Corden · 03/09/08, 7:16 PM

I was wondering if anyone has modeled the stats for the new fist MH weapons - Vanir's Right Fist of Brutality vs. Dragonstrike.

Basically - I am curious if the static haste increase is a bigger boost to DPS than the proc from Dragonstrike, the other stats notwithstanding.

Cheers,

C

gemmanite · 03/09/08, 7:20 PM

Originally Posted by Mithgaur

Hi, I was hoping someone could help me with a question

I'm currently using Talon of the Phoenix on MH and Blade of Infamy OH.

To estimate my DPS I use Grim13 stylesheet. What surprises me is that Swiftsteel Bludgeon is giving me higher dps then what I'm currently using. I.e Swiftsteel Bludgeon MH and Blade of Infamy on OH.

I also get lower DPS with the mentioned above if I switch them around. I.e Blade of Infamy MH and Swiftsteel Bludgeon OH.

My question is, how can this be correct? A fast weapon on MH with almost equal stats to Talon of the Pheonix which is a slow weapon, is showing up superior dps.

Talon of the Phoenix(MH) and Blade of Infamy(OH) = 1374,1 DPS
Swiftsteel Bludgeon (MH) and Blade of Infamy (OH) = 1381,7 DPS
Blade of Infamy (MH) and Swiftsteel Bludgeon (OH) = 1373,4 DPS

Any help would be welcome.

The spreadsheet simulates your DPS from 100% to 0% and assumes you are not switching to two fast weapons to execute under 20%. So the proper way to calculate your normal cycle DPS above 20% is by adding white+cycle DPS. Similarly, switch to your execute weapons and add white+execute DPS to find your DPS from 20% to 0%.

I think that of the three combinations that you mentioned, yours is correct.

This might be a little misleading. I remember it being suggested that there be an option included in the spreadsheet to model switching weapons to execute.

Mithgaur · 03/09/08, 8:01 PM

Originally Posted by gemmanite

The spreadsheet simulates your DPS from 100% to 0% and assumes you are not switching to two fast weapons to execute under 20%. So the proper way to calculate your normal cycle DPS above 20% is by adding white+cycle DPS. Similarly, switch to your execute weapons and add white+execute DPS to find your DPS from 20% to 0%.

I think that of the three combinations that you mentioned, yours is correct.

This might be a little misleading. I remember it being suggested that there be an option included in the spreadsheet to model switching weapons to execute.

Thank you. That explains it. I have been trying to find where I can modify the fight from 100% to 20% and 20% to 0%. If you know, I would appreciate if you told me.

Kaan · 03/09/08, 11:34 PM

Originally Posted by landsoul

Well anyways, this is what I have been working on off and on for the last 4 months. I cant make a new thread because I don't have permission so I just put it here. It doesn't have all the goobly gob yet its very comprehensive for DW 17/44 and its variants at least.

Landsoul's Fury DPS 1.0 for patch 2.4; Feb 2008

Written in Open Office

sheet
FuryDPS1.0.ods - FileFront.com

prog
download: OpenOffice.org Downloads

I plan to incorporate Arms/Fury and Prot DPS later. It's got a few minor bugs but its very easy to get over them. Please DL it and give any feedback of what you think, want, or if there are any data errors.

In advance, thanks for the work you have done in making this new sheet.

However, I don't use OO and won't, so is there a way you can make a Excel variant of this?

Jorak · 03/10/08, 1:19 AM

Originally Posted by Kaan

In advance, thanks for the work you have done in making this new sheet.

However, I don't use OO and won't, so is there a way you can make a Excel variant of this?

Asking for a switch from free software using an OpenDocument format (whose spec is also free) to expensive proprietary software using a proprietary closed format seems a little unreasonable, don't you think? You know you could get an OpenDocument plugin for MS Office if you're so insistent on using it.

Machinator · 03/10/08, 2:49 AM

Feedback for landsoul's sheet.

Locked spreadsheets make it a pain to review how things work or add anything. Protected so people dont break it I can understand, passwords on the protection I dont. Also a manual Other section on gear would be good.

You will probably go crazy if you determine gear stats by giant IF statements. OO can handle vlookup cant it?
Just note that weapon DPS is rounded from actual (min+max)/2/Speed.
Executioner and mongoose are not as high as 1.5 PPM, testing put them around 1.1-1.15 PPM.
You dont allow Executioner on the OH for some reason.
I am not sure what your flurry uptime is doing exactly. Is 1-(1-crit%)^(3+Instants per 3 hits) not accurate?
I dont get the same DPS output if I change the gear and change it back.

On a side note, my pet peeve with sheets is counting procs as a value * uptime and using the result as a constant in the sheet. Most of the time this is close enough but with certain things, rage primarily, I am not sure it is. Do you know if it possible to model such things outside a simulation? I have been trying to come up with a model for rage starvation/saturation for my own sheet I am working on but I think I either dont know enough statistics, or its outside the scope of a sheet. I think this is something especially important for execute modeling.

landsoul · 03/10/08, 3:30 AM

Well if you really want to know.... grab a drink and start reading.

Flurry Uptime

Next we will tackle the most complicated of all baseline fury warrior functions, your average flurry uptime. Flurry uptime can never be 100% unless you have 100% chance to crit, which is impossible. To find flurry uptime, we must find loss in flurry and subtract it from 1. Loss in flurry is equal to the chance to not crit with specials and white attacks during the duration of the actual flurry buff. The Duration of the buff lasts for 3 landed non-crit white attacks, or 12 seconds. Unless we're weilding 6.00 speed weapons or couldnt hit the broad side of a barn, the 12 seconds never comes first.

Lets say for example, we had a crit chance of 30% or .3. To successfully "cannibalize" a flurry charge we have to land and not crit any of our attacks. Therefore only will our charge be eaten if we hit normally or glance. To fully remove the buff, we must do this three times in a row. If you are weilding two weapons of the same speed, then you must do it 4 times in a row. Also none of our instant attacks over this time can crit. We must multiply all of our chances to not crit together to arrive at our percent time flurry is down.

Percent Flurry Downtime = (chance to hit or glance with white)^{3 or 4} * (Chance to not crit instant)^(instants per 3 or 4 normal swings)

[%flurryDOWN] = ([%Nhit] + [%glance]) ^ [HitsPerFlurry] * (1 - [%crit]) ^ [SpecPerFlurry]

[%flurryDOWN] = (1 - [%avoid] - [%glance] - [%crit] + [%glance]) ^ [HitsPerFlurry] * (1 - [%crit]) ^ [SpecPerFlurry]

[%flurryDOWN] = (1 - [%misswhite] - [%dodge] - [%crit]) ^ [HitsPerFlurry] * (1 - [%crit]) ^ [SpecPerFlurry]

[%FlurryUP] = 1 - [%FlurryDOWN]

[%FlurryUP] = 1 - ((1 - [%misswhite] - [%dodge] - [%crit]) ^ [HitsPerFlurry] * (1 - [%crit]) ^ [SpecPerFlurry])

Okay, there's a function in there we dont know yet, [SpecPerFlurry]. Basically its the amount of specials you will hit in your ideal rotation over the maximum duration that the flurry will last. To find this we need to calculate additional pieces of information. We can setup this equation:

(specials/second)(swings/flurry)(average seconds/swing) = specials/flurry

If specced correctly, Every 18 seconds 3 BT and 2 WW are used, yielding 7 special attacks every 18 seconds as shown in the cooldown cycle section that will refresh flurry. And we already know that we can get 3 or 4 attacks for each flurry. We need to calculate average seconds per attack, which also includes the flurry uptime function. Again, circular reasoning.. However this time we have an assumption that we will only be calculating this when we are flurried, so we can set the flurry uptime function to 100% there. Lets go ahead and find out how frequently we attack.

Swings per Second and hits per second

With differing speed weapons, frequency of heroic strikes, windfury totem, modeling this is not easy. Imagine a time table of your attacks when they are flurried and hasted. On an infinite time span, there are a specific number of offhanded attacks per main hand attack, even with windfury totem and heroic strike, assuming you have the same hit table factor for both weapons.

What we can do though is find the ratio of offhand to main hand attacks and then we can come up with several forms of the hits per second function that we can use accurately in other functions. Here are some things we can find:

Main hand white only swings landed per second
Main hand white and heroic strike landed swings per second
Main hand white and heroic strike and windfury landed swings per second
Main hand white and heroic strike and windfury and special attacks that use the main hand swings per second
Off hand white only swings per second
Off hand white and whirlwind swings per second

Main hand white only swings landed per second:
(1 - [%dodge] - [%misswhite])((MHWepSPD/((1 + .25[%FlurryUP])(1 + [%hasteAVG]))

Main hand white and heroic strike landed swings per second:
Using Heroic strike will completely eliminate your chance to miss, Yet we only hit heroic strike a percentage of the time, we call this [HSfrequency] and is found in a different section.
(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1 / (MHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG])))

Main hand white and heroic strike and windfury landed swings per second
(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1 / (MHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG]))) + [WFperSec]

Main hand white and heroic strike and windfury and special attacks that use the main hand swings per second
(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1 / (MHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG]))) + [WFperSec] + 5/18

Off hand white only swings per second
(1 - [%dodge] - [%misswhite])(1 / (OHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG])))

Off hand white and whirlwind swings per second
(1 - [%dodge] - [%misswhite])(1 / (OHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG]))) + 2/18

Windfury Extra Attack Frequency

What is a 20% chance to grant an extra attack off of Main Hand white and heroic strike landed swings that lands 1 - [%dodge] percent of the time? A windfury Extra attack. Windfury attacks per second is equal to this:

(1 - [%dodge]).2(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1 / (MHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG])))

Total Swings per second that can consume flurry

Based on experience, we know that Main hand white and heroic strike and windfury landed swings and Off hand white only swings can consume flurry. now that we have this information, we can setup the total swings per second that we needed for our flurry uptime model.

(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1 / (MHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG]))) + [WFperSec] + (1 - [%dodge] - [%misswhite])(1 / (OHWepSPD / ((1 + .25[%FlurryUP])(1 + [%hasteAVG])))

We can do a little cleaning up here. Remember we are setting flurry uptime here to 100% because we are flurried when we want this number.

(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1 / (MHWepSPD / (1.25(1 + [%hasteAVG]))) + [WFperSec] + (1 - [%dodge] - [%misswhite])(1 / (OHWepSPD / (1.25(1 + [%hasteAVG])))

(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1.25(1 + [%hasteAVG]) / MHWepSPD) + [WFperSec] + (1 - [%dodge] - [%misswhite])(1.25(1 + [%hasteAVG]) / OHWepSPD)

(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])(1.25(1 + [%hasteAVG]) / MHWepSPD) + (1 - [%dodge]).2(1 - [%dodge] - (1 - [HSFrequency])[%misswhite])((1.25(1 + [%hasteAVG])) / MHWepSPD) + (1 - [%dodge] - [%misswhite])(1.25(1 + [%hasteAVG]) / OHWepSPD)

Now, going back in to our
(specials/second)(swings/flurry)(average seconds/swing) = specials/flurry problem, we have the swings per second part of it. So we need to inverse it to make it average seconds per swing that will consume flurry. Instead of flopping that cumbersome model, I would just rearrange the initial equation to this:

(specials/second)(swings/flurry) / (swings/sec that will consume flurry) = specials/flurry

Remember that if your attack is avoided, it will not consume a flurry charge. Flurry Uptime will increase as you add more crit chance, incorporate same speed weapons, and are able to land instant attacks more efficiently. Flurry Uptime will decrease when adding hit, expertise, and haste, however adding these stats more than makes up for this loss in other places on the board. We will multiply this Uptime value to the 25% increase in attacks made while flurried.

Again, here is the Flurry Uptime Equation.

[%FlurryUP] = 1 - ((1 - [%misswhite] - [%dodge] - [%crit]) ^ [HitsPerFlurry] * (1 - [%crit]) ^ [SpecPerFlurry])

Yeah some of the PPMs might be wrong, but I haven't seen any concrete proof on it. Just a bunch of people claiming to get this and that and always coming up with different numbers. I have not uncovered the problem that requires you to make a gear change two to three times to get the correct value yet.

Making If statements requires no effort by me. all I have to do is copy and paste from a generated text file. What this does for me is severely cut down on extra information.

Also, ferocious inspiration is not working. I'll get it fixed in the next upload.

I also didnt include Executioner in offhand because its pointless and would require too much work to make a model to where it only refreshes the duration.

Overcoming rage saturation and starvation comes from the skill of the player in my opinion. Watch your instant timers, rampage duration, and rage pool to make the right judgement call. I find very few times my rage hits 100. Not having 75 total for a triple GCD use for BT + WW + Rampage gets me sometimes but making a model of that would get interrupted by the actions of the player.

Gruntle · 03/10/08, 6:34 AM

Originally Posted by landsoul

When you are using Heroic strike, it removes the chance of a miss effect occuring on the main hand and lowering the chances to active PPMs and % chances to proc on trinkets, enchants, windfuries, etc. When you use HS, you are effectively minorly increasing the uptime of these procs and effects. While the net of using heroic strike is an overall loss in rage per second over rotation, it changes areas that contribute to the Rage Per Second model, where the values have to be recalculated in multiple passes until it does not change to a certain degree. The only time Heroic strike will not modify these is if you are rolling with 100% white miss recovery.

You're of course right about this, I think that it is a very minor effect though. It requires even more circular references (the ones in place already make the sheets quite unstable).

I'd have to say the same about your proposed changes to how to model flurry. The formula you provided seem to be correct and is probably more accurate than using 1 - (1-crit)^(hits per sec) (where hits per sec includes both white, yellow and wf hits). But I severely doubt that you will gain much in accuracy going to the more complicated model. Especially since the flurry mechanics are still not properly understood (i.e. how flurry fades when using 2 weapons). Check earlier pages in this thread for a discussion (and tests done by drALLCOM3) on how flurry works in game. E.g. you say that using 2 weapons that are unmatched in speed will give you 3 flurried white hits and using matched speeds will give you 4. This is just not true. In the best scenario matched speeds will give you 4 hits, but that requires that the weapon swings stay synchronized for all of the fight, this is seldom the case since both parries and using special abilities to start your attack will throw of the synchronization. For unmatched speeds the number of flurried hits can vary depending on the exact speeds of the weapons and where the white hits fall during the dps cycle, the mean is probably close to 3 but I wouldn't count on it (perhaps it's even depedent on the latency of your connection and so on).

But sure, if you can make the spreadsheet work with these changes, they should give more accurate results.

I tried using your spreadsheet (great that you made it in OO from scratch) using OO 2.0. I changed a lot of gear and some buffs, but I cannot get any dps estimate out, instead I get ERR:509 errors in the output cells. Since the sheet is protected (I really don't see the point in that?) I cannot go in and check what's wrong. Maybe it's just that I'm using an old OO version, otherwise I suspect the errors have something to do with crit. Otherwise I like the sheet, it seems simple to use and contain most stuff that you'd want. I would like an option to add your own gear and bogus stats though.

landsoul · 03/10/08, 12:29 PM

Okay besides unprotecting the sheet, you want the option to add your own gear? I dont understand are you trying to make bogus gear? Can you not find the items you are currently wearing? The items included are in the game minus a few new releases. Allowing to add bogus stats would not be a problem but I don't see a reason why you would want to. Is there a reason you want to add bogus stats?

About the errors: try it with the latest version OO. What boxes in the output tab are displaying errors?

The reason to protect it is so that people who think they know what they are doing can go in and bust it all up. Its very sensitive to change because of how it is setup. There's no reason to change any of the formulas without posting the issue here. I can however let the cell contents be fully visible but not changeable.

About flurry mechanics:
We all know how flurry works. It increases your attack speed while the buff is up, until it goes away when you cannot crit again. The speed from it may be applied to a partial swing. When using paired speeds and your attacks are not interrupted or sped up by a parry, the flurry buff will apply to the full swings on the second hit of each weapon, essentially working as 4 hits instead of 3.

I personally have been using matched 2.6 weapons (Vengeful Sword and Blade of Infamy) for quite some time. In raids fighting bosses each swing happens simultaneously every single time, allowing it to use the 4 hit rule.

When using differing speed weapons the flurry mechanic gets extremely more complicated because no one really knows if the last cannibalize forces the speed to distribute over a partial swing or not. Who cares even? We're doing tangible estimations based on averages in spreadsheets anyways. If someone wishes to prove exactly how flurry works thats fine, I just choose not to slave over combat log data, server-client lag, or possibly inaccurate proc watchers or mods.

Apate · 03/10/08, 12:41 PM

Originally Posted by landsoul

I personally have been using matched 2.6 weapons (Vengeful Sword and Blade of Infamy) for quite some time. In raids fighting bosses each swing happens simultaneously every single time, allowing it to use the 4 hit rule.

It does? Every single time? Is this a new change?

Previously, you needed to toggle attack before moving in to the boss to align the weapon swings. It was still susceptible to desyncing, though. There should be some data and info on this in the flurry thread. There's also some info regarding your "4 hit rule."

Brissa · 03/10/08, 1:04 PM

Originally Posted by landsoul

Okay besides unprotecting the sheet, you want the option to add your own gear? I dont understand are you trying to make bogus gear? Can you not find the items you are currently wearing? The items included are in the game minus a few new releases. Allowing to add bogus stats would not be a problem but I don't see a reason why you would want to. Is there a reason you want to add bogus stats?

The obvious answer is that you want to be able to add any new items that show up on the PTR (to see if they will be upgrades for you or not) without having to wait on someone else to release a new version of the spreadsheet.

landsoul · 03/10/08, 1:17 PM

Can't some people just be a little patient wow lol..

I'll add some of the new items (I think I only missed one though, the crit/pen ring) if someone wishes to add more just let me know.. And a few new things in the next version. Possibly some corrections on various PPM effects also.