Koosai

Maurice you forgot one major source of armor pen, Warp-Spring Coil. The proc on that when combined with executioner will be very powerful. I wouldn't factor The Night Blade in, lack of dps and dagger spec woes should factor it out for anyone with access to s2 weapons.

Here is my attempt at sorting this out. I am splitting the benifits of mongoose into three parts, the speed, the crit and the ap. The attack speed will mostly only affect white damage, more combat potency and procs but I'm going to ignore those. White damage is around 60% of my damage, so 2%(Pretty sure that is mongoose haste) haste on my white damage means that my white damage is 61.2% of what it normally is, so I'm doing 101.2% of my normal damage. The attack speed is therefore a 1.2% total damage increase. Assuming I'm rupturing and not using eviscerate my crittable attacks are roughtly 85% of my damage (Couldn't remember exactly, 60% white and 25% SS sounded reasonable to me). If 3% more of my white attacks crit those 3% do twice the damage, ignoring RED. Assuming a 30% crit rate, 25% glancing and hit capped we have a 45% chance to hit, 3% of those will now crit. However the 30% crits will actually be 48.48% of the white damage, and hits will only be 36.36% of the white damage. Increasing the crit of my white damage increases my white damage by 1.09%. My white damage is now 61.87% of my unbuffed damage. Assuming the same 30% crit rate, 3% of my non-crit SSs will now crit for 230%. SS crits are 49.64% of my damage. Having 3% of my SS hits by 230% increases my SS damage by 1.96%. SS damage is now 25.49% of my unbuffed damage. With the crit and attack speed from mongoose I do 102.36% of my normal damage, a 2.36% increase in damage. NOTE: let me know if I got sloppy with my math, I assumed that glancing blows do 75% damage.
Doing the calculations for AP taking white damage, SS and 5 pt rupture into account (rupture is 10% of damage here) using a s2 MH I get:
AP --- AP/14 - Mon+AP/14 - %+white - %+SS - %+Rupture - total increase (including crit and attack speed increase)
1400 100 ----- 108.57 ------ 4.34% ---- 3.47% - 2.16% ------- 6.14%
1800 128.57 - 137.14 ------ 3.79% ---- 3.11% - 2.01% ------- 5.69%
2200 157.14 - 165.71 ------ 3.36% ---- 2.82% - 1.88% ------- 5.34%
2600 185.71 - 194.29 ------ 3.03% ---- 2.58% - 1.77% ------- 5.07%
3000 214.29 - 222.86 ------ 2.75% ---- 2.37% - 1.67% ------- 4.83%
3400 242.86 - 251.43 ------ 2.52% ---- 2.20% - 1.58% ------- 4.64%
3800 271.43 - 280 ---------- 2.32% ---- 2.05% - 1.51% ------- 4.47%

If someone could thow the AP and total increase into excel and plot a best fit line for me, looks quadratic to me, I can't seem to find a way to do it in Open Office.

One other important note is that I did the calculations using a s2 MH, having a higher damage range will decrease the % SS damage was improved by, a lower damage range will increase the % SS damage was increased. Also because white damage is not normalized higher *dps* weapons will decrease the % white damage was increased by.

If Lith and my calculations are correct it looks like assuming the same PPM they become roughly equivalent at 3000 buffed AP against bosses with the lowest armor. While this is attainable, it is also only for the best case with executioner. Though it is important to note that mongoose will be slightly better than what I calculated because the haste will increase poison applications, procs and combat potency. The crit could also trigger crit based procs if you have any.

 

Nydus

Ok had a bit of curiosity in this thread so far.

What we know the proc's rates are identical, and so is the duration of the proc.

Now for a raid boss setting average boss 7700 armor brought down to 3600 from sunder,FF,COR

night blade = 1305
3600 armor / 14157.3 = 25.5% armor
2295 armor / 12852.3 = 17.9% armor
Increase of 7.7%

executioner = 840
3600 armor / 14157.3 = 25.5% armor
2760 armor / 13317.3 = 20% armor
Increase of 5.5%

Warpspring coil = 1000
3600 armor / 14157.3 = 25.5% armor
2600 armor / 13157.3 = 19.8% armor
Increase of 5.7%

Serrated Blades = 560
3600 armor / 14157.3 = 25.5% armor
3040 armor / 13597.3 = 22.4% armor
Increase of 3.14%

So quite clearly executioners effectiveness is roughly 5-6% for a raid setting

Now mongoose we have 120 agi and 2% haste
Effects on the classes
Rogue 3% crit and 120ap
Warrior 3.6% crit
Shaman/paladin 4.8% crit

So quite clearly executioner is a better enchant for all classes except a rogue.

As for a pvp setting executioner is the master of cloth..... And will yield more than a 10% damage increase on clothies.

Now there is some truth behind executioner being extremely powerful for mutilate builds, mainly due to the use of evis in rotation with rupture but it requires a very size able chunk of armor pen

 

Pitbuller

There is very powerful tool to shamans Crazy Shaman's DPS & AEP calculator (c) Yo
Thats not clear at all.
All test shows with reasonable gear that mongoose => exe.
Only with best gear in game + sunders + ff + cor and then exe can beat mongoose but not clearly.

Slow, slower, shaman weapon.

 

Malan

I ran some sims today Pitbuller and found that my simulated DPS for Executioner/Mongoose was only 10 DPS higher than Executioner/Executioner, and almost on par to Mongoose/Mongoose, with 175 passive -Armor from my neck.

Shitting up every single thread on EJ since '06

 

◊ Shaker

I would highly question the simulation's model of Executioner/Executioner, then. Dual Executioner is really, really non-optimal in practice.

Though, I suppose that could be a ShamanThing™. :P

 

Pitbuller

The problem is that when you gain more -armor mongoose scale too. And point where exe > mongoose go futher and futher. I used malan gear to reproduce test. All buffs up and 3600boss armor value(sunders, FF, CoR)
Baseline only 175armor penetration from neck.
With kings values are:
AP: 1593 (trinket ap as static value 278/6)
crit: 29.37%
hit: 14.21%
haste: 4.69%
boss armor: 3600
100.2dps and 96.5dps 2.6 speed waepons

Easily reproduced test.

Slow, slower, shaman weapon.

 

Sarlunas

First off, 20.7% is generally rounded to 21% not 20%. Secondly going from 25.5% DR to 20% would not be a dps gain of 5.5%, it's actually (1-0.2)/(1-0.255) ~= 1.0738, roughly 7.4% dps gain.

This may or may not be true but you present no numbers whatsoever to support it. The gain from crit and haste is not straightforward, especially when considering talents like flurry etc.

I suppose it's understandable that a warlock doesn't know much about melee dps but then why post here in the first place?

 

• Aldriana

In the interest of checking the proc rate on Mongoose, I just broke my Blade of Savagery on one of the Dire Maul ogres, and can now conclude with a reasonably degree of certainty that Mongoose is not, in fact, 1 PPM.

I had 58 passive haste equipped, no procs other than Mongoose, and a base 1.4 speed MH.

I hit the poor unfortunate fellow 7781 times, generating 166 uptime periods including at least 220 procs; this works out to a proc rate of between 1.09 and 1.42 PPM (if it's PPM) or 2.46% and 3.20% (if it's not).

However, it occurred to me that there's a better way to estimate proc rates from such a data set, namely: only count the hits that occurred outside of Mongoose uptime, and compare to the actual number of uptime periods; that is, any uptime period starts with a hit that did not occur during Mongoose uptime, and, as such, taking total number of uptime periods divided by hits outside of mongoose gives an estimate of proc rate that removes the ambiguity around refreshing procs.

Grinding the numbers according to this new insight, we have 5460 eligible attacks, and 166 uptime periods; this means a proc rate of between 2.58% and 3.50%, or 1.15 PPM to 1.55 PPM.

This strikes me as sufficient data to say with reasonable certainty that it is not 1 PPM. What it is instead still needs to be determined, but I would say that one proc per 40 seconds looks reasonably good for it, which would resolve the issue of why Executioner seems to have a proc rate around 1.33 PPM and procs at about the same rate as Mongoose.

To confirm the proc rate of Mongoose, we will, of course, need more data of this form; if someone would like to perform a similar test with Mongoose on a slow 1H weapon (2.5+, ideally) I'll work on expanding this 1.4 (actually 1.35 due to passive haste) data set so as to see if we can pin the ranges down a bit more exactly.

 

Machinator

How long did it take to hit 7781 times, were there misses? How did you count uptime periods? What methodology did you use to get those numbers?

"Information is ammunition."

 

• Vulajin

I can do this with my Talon of Azshara. Excellent insight on taking total uptime periods relative to swings made without the buff, that greatly simplifies the analysis (and also means that I don't need to dig up that buff-canceling macro to get really accurate data). I'll edit my combat log into this post once I have it, later this afternoon probably.

(edit) Completely forgot it was Tuesday. Well, I'll get started as soon as the servers come back up, I have pretty much an entire day with nothing on my plate.

 

• Aldriana

I started attacking at 9:26 and my sword broke at 12:12 - so a bit under 3 hours, all told. I didn't miss. I counted all data by running the combat log through a python script I wrote to pull out the relevant numbers; it just counts total hits + crits, and looks for the "You gain Lightning Speed" message. Fundamentally it's very similar to the Executioner-analysis program I posted earlier in this thread, just adapted for Mongoose.

 

songster

Heh, I guess you could double check it by doing the reverse. Compare the number of refreshed procs (i.e uptime > 15 seconds) to the number of swings made while under the influence of Mongoose. You expect the number to be slightly lower due to missing triple-refreshes.

 

• Vandemar

I did some testing on a DM spirit last night for this- Blade of Infamy with Mongoose, 330 hit rating, DST equipped but no passive haste for 2.6 mainhand speed with 2.16 during Haste procs. Combat log is attached. I'm not sure how many hits were involved, but it went from 03:18 to 06:02 when the server went down for maintenance.

Attached Files

infamymongoosetest.txt (248.0 KB, 70 views)

 

Sarlunas

Pretty much the same test: 2.7 speed fist weapon (talon) with mongoose on DM spirit, no misses, no haste, no procs. Took 4.5 hours for it to break.

Free file hosting by Savefile.com

 

Machinator

Well I wish I remembered enough of my C++ course to write something to parse these, perhaps over break. I would like to get something that everyone can use rather than posting logs and hoping someone parses it.

The reason I am interested in the time it takes is because of what I have been reading about haste and PPM style procs. Im not sure what it is, but there is evidence that special attacks use base weapon speed and normal attacks use current (hasted) weapon speed. So over the same time period with the same weapon, using haste would not increase the number of procs. For purposes of testing I think its best to not use any sort of haste or flurry, besides mongoose itself. I dont think that 2% haste will throw the numbers off that far.

Does anyone have any references to previous work on PPM? Im not coming up with much that is much more useful.

I just did a test last night on PTR, 176 mins, 189 gain lightning speed. I didnt count procs longer than 15s though.

PS. Is there a better way for a non-rogue to get to the DM ogres? :P

"Information is ammunition."

 

• Aldriana

If there's demand for it, I can try to clean up one of my python programs enough for public consumption; however it's a little bit tricky since you tend to be looking for different information for different procs. At some level you do just need to modify source code to get the exact information you want out rather than having a canned program.

 

Machinator

Well from a programming view, the needed inputs are what the proc shows up as in the log, proc length, and hidden cooldown if applicable. I think that the only hard part would be people that dont understand the limitations of the combat log, and therefor the program, and get wacky numbers.

Should this sort of thing be a new topic? I know I would use it for things other than Exe and mongoose.

"Information is ammunition."

 

• Aldriana

If you're looking at procs of a certain sort, that would be useful; but, for instance, if you wrote the program for Mongoose and Executioner, and then tried to use it on Tsunami Talisman (which triggers on crits but not hits) or Sword Spec (which has gains but not fades), it would break. The problem is that every proc is a little different.

However, given our experience here, it might be worth starting a new thread, of the form "check the proc rate on every proc in existence, since clearly we've made some bad assumptions" - which would be a natural place to put such tools.

 

• silentogre

I'm sure someone could whip up a nice C# app with a few xml config variables that would work on multiple situations.

Anyone interested in working out a definitive list of proc conditions/rules as the foundation for this?

//edit: should probably be a new topic

 

• Aldriana

So, running the numbers on the data set in post 339 using the new "uptime periods versus eligible attacks" method, we find 4580 total swings and 270 uptime periods, for a proc rate between 5.21 and 6.58%, or 1.16 and 1.46 PPM.

From this, I think we can safely say that it is indeed a PPM effect, and that it does indeed most likely lie between 1.16 and 1.46 PPM. To narrow that down further, we will of course need, you guessed it, more data. The most useful data would be data that expands one of the existing data sets; either 2.7 speed with no haste (or haste procs) equipped, or a 1.4 speed weapon with exactly 58 haste rating and no haste procs (I use Swiftstrike Cuffs and Band of Devastation). If you have such data sets, please post them and we'll try to get this figured out once and for all.

 

• Latito

I could do the 1.4 speed later.. but I'll have either 0, 31, 37 or 68 haste (I think).. I've got the ring and the belt. I'll have to wait until ~thursday though since BB -> Illidan is tomorrow. I could also do a 2.6 speed with any of the above mentioned hastes afterwards. Basically just setup the character and go do christmas shopping, right? The DM guy doesn't hit back so its just "make sure you don't have non-mongoose proc'd haste" and only equip one weapon. Any reasonable amount of hit rating should result in no misses.. not that it really matters a ton.

 

silendeath

so what would be best for a fury warrior, dual exectioner, duel mongoose, I would like to go exectioner on the mainhand and mongoose on the offhand. I just wana find out what would be best before I go and make a mistake with the enchant

 

• Aldriana

Right. Only thing to be careful of is that if you go too long without hitting a key, you get automatically logged out after a while. I generally start it and then watch a movie/read a book/browse the forums/work on the spreadsheet/whatever, and just tab in every 15 min or so and hit a few keys to make sure to stay online.

Come to think of it, it probably would make sense to ditch the haste rating and standardize on just 1.4 weapon speed - will make it easier for more people to test. So, lets say for purposes of testing this, we'll standardize on 1.4 and 2.7 weapon speed, no haste procs, no haste, and see if we can get the range small enough to see what's going on.

Edit: Based on this decision, I cut off tonight's set of testing at 1.35 and dropped haste to 1.4; hence my final numbers for 1.35 from last night and tonight are 6441 eligible hits and 199 uptime periods, for a proc rate between 1.18 and 1.56. I'm not starting a 1.4 set, which will be short due the fact that my weapon isn't starting at full, but it's a start.

Edit 2: First batch of data at 1.4: 98 uptime periods out of 3622 eligible hits; proc rate is between .93 and 1.39 PPM.

 

Machinator

Well I made a thread for determining proc rates here. Im sure I missed some stuff so feel free to add.

"Information is ammunition."

 

• Latito

Proc Rate Determination
Mongoose testing - total almost 10k attacks with a single weapon equipped, no static haste, no haste-procs