PDA

View Full Version : Rage Generation and Haste



bagelbite
04-17-2008, 12:49 AM
Haste effects has been said to lower the rage per hit for quite a while now. This includes the haste effect from flurry, as well as haste from haste rating. I beleive this false, and I assert that haste does not affect rage per hit at all. This post is strictly about rage per hit, I'm not talking about rage per second at all.

I encourage all criticism of testing methods, and I discourage people from using my conclusions until others have tested and verfied.

Formulas:Rage generation - WoWWiki - Your guide to the World of Warcraft (http://www.wowwiki.com/Formulas:Rage_generation)
The current belief is that the hasted weapon speed is used in the rage formula when calculatiing rage per hit. This specific detail is what I beleive to be false, and I am asserting that the origional weapon speed found on an item tooltip is used.

I made a quick calculator to aid me in doing the math for how much rage I should receive per hit. It handles all the math of figuring out what your rage per hit should be given the parameters it accepts inputs for.
http://rehfeld.us/wow/rage-test.html

If you prefer to do the math yourself, the following links contain the information needed.
Formulas:Rage generation - WoWWiki - Your guide to the World of Warcraft (http://www.wowwiki.com/Formulas:Rage_generation)
Combat rating system - WoWWiki - Your guide to the World of Warcraft (http://www.wowwiki.com/Combat_Rating_System)
Haste - WoWWiki - Your guide to the World of Warcraft (http://www.wowwiki.com/Haste)

By studying the rage formula, rage per hit is comprised of 2 seperate calculations. One component is based on your level, and how hard you hit, nothing else. The other componenet is based on weapon speed. The weapon speed portion of the equation is what we care about, since this is the part which would be effected if haste were to alter your rage per hit. Unfortunately, in-game we cannot see how much rage comes from each component of the formula, so we must minimize the impact of the component which deals with the damage you do. We will do this by making sure we have a narrow damage range, which allows us to isolate the rage from the part of the equation we care about.

The narrowest damage range weapon is Battleworn Claymore - Items - World of Warcraft (http://www.wowhead.com/?item=23346)

And by virtue of our weapon with a narrow damage range(3 damage), our rage will only vary by 0.041 rage per hit, which is plenty accurate for testing. The rest of our gear doesn't matter much, so long as no gear can proc. Ideally we have low attack power, so we don't hurt our target too much and dont fill up on rage too fast.

I choose talents which don't interfere with damage or rage, only going up to flurry.


We need an accurate way to measure rage. Unfortunately, the game only provides visible rage in increments of whole integers, not fractions. Even UI script such as
/script DEFAULT_CHAT_FRAME:AddMessage(UnitMana("player"));

only provides rage in whole integers.

So, we first need to establish that the game, internally, does indeed keep rage values accurate to decimal precision, and that the game just rounds for display purposes. We establish this by auto attacking our target for an amount of damage which would yield an amount of rage which should result in a fraction. For example, using our example 2.9 speed 2h weapon, and my provided rage calculator, if I hit my target for 180 damage, I should be awarded 7.396 rage per hit. So the first time I attack them, my rage will increase by 7, but every 2nd or 3rd hit it will increase by 8. This proves the game internally keeps rage to at least 1 decimal precision, and rounds for client display purposes.

With that established, we can test difference in rage per hit, comparing hasted and non hasted hits.

I chose a priest as my beat bag. No buffs involved. I altered my attack power so to where my observed attacks were either 149 or 150 damage.
Plugging this into my calculator, I should receive:
7.123 rage per hit.
6.108 rage per flurry hasted hit, if flurry affects rage per hit.

The key thing to keep in mind here is that your rage will always increase by at least 7, and the occasional 8. The occasional increase of 8 is from when the fraction of rage(.123) adds up enough times to form a whole point of rage, or enough to get rounded up. The game display will not increase by less than 7 rage per hit, because we generate slightly more than 7 rage per hit. I tested this without yet taking the flurry talent, and It proved true.

Now, we add 5/5 flurry and attack our target again. The key thing we are looking for here is that if haste decreases rage per hit, we will see most hasted attacks provide an increase of less than 7 rage. In particular, most hits would be expected to increase rage by 6, and occasionally 7 rage when the fraction adds up enough times. However, I did not come across a rage increase of less than 7, and instead I still got the occasional increase of 8 rage, which is exactly what happened without flurry. I tested for a few minutes, across a good amount of flurry charges.

Now, you may be thinking its possible the first flurry charge which gets used maybe for some reason hasn't yet applied its new speed into the rage formula. Even if we assume this to be true, at least 1 of the following 2 flurried hits would cause an increase of only 6 rage, which didnt happen.

I repeated the test without flurry, but used Abacus of Violent Odds - Items - World of Warcraft (http://www.wowhead.com/?item=28288) instead and again found no evidence that haste reduces rage per hit.


I did another test, this time in blasted lands againts the invulnerable mobs. This time I allowed rage from crits to be included in my data. I hit for 165, this would give me the following expected rage per hit numbers:
7.327 rage per hit
6.312 rage per flurried hit

Remember, a crit is exactly double the rage of a hit. No, I didnt have a +crit dmg meta gem equipped or anything other factor/talent which affects crit damage.

I would attack the mob, saving rage, until he went invulnerable. At this time I would rotate my character to the side, as there is an angle(about 90 degrees) where you will not auto attack, yet you can still use special abilities. I would spam hamstring until flurry procced, at which point I would execute. This would leave me with 3 flurry charges, zero rage, and no auto attacks firing. Then I could liesurely rotate my character back towards the mob, where I could easily observe how much rage I generated from 3 successive flurry attacks. There was no data tainting from damage taken, as when the mob is invulnerable, he doesn't attack me. I repeated this a few times, and each time, it supported that I was getting normal rage per hit. To give an example where i ended up with 36 rage after the 3rd attack:
165 hit
330 crit
330 crit
and the expected rage per hit of each, respectively
7.327
14.654
14.654
add these expected rage numbers together and you get 36.635 rage, which is what I observed on my rage bar.

If flurry reduced my rage per hit for those attacks, my expected rage would have been 31.56, which is quite a bit different than the 36 rage I observed.

I think this test in blasted lands was more simple, and enough all by itself to prove my statements.

Conclusion:
Haste effects do not reduce rage per hit.

Signu
04-17-2008, 05:54 AM
Awesome! Nice work.

Kazeyonoma
04-17-2008, 07:52 AM
So what you've concluded from your research is that haste effects don't change the weapon speed on your rage generation formula. Very Very interesting...

veneretio
04-17-2008, 09:15 AM
Nice work. Have you done any research towards whether constant haste gains offer different results than limited duration buffs?

bagelbite
04-17-2008, 10:29 AM
No, I have very little gear with passive haste rating on it.

Ciderhelm
04-17-2008, 10:42 AM
Great read, thanks again for being really informative bagel. Thought we lost you back at the end of last year when you stopped posting on the WoW forums for a while. :(

mattdeeze
04-17-2008, 10:43 AM
Very Impressive. I vote bump

Crimsonstorm
04-17-2008, 02:29 PM
Therefore, haste increases your overall rage per second generation,m kaing it slightly better than otherwise believed. Good to know.

Corbusier
04-17-2008, 02:32 PM
Mind if I link this in my guide and/or rip off your testing?

bagelbite
04-17-2008, 03:07 PM
Go for it.

Corbusier
04-17-2008, 06:12 PM
Thanks!

Rampart
04-18-2008, 03:42 AM
This pretty much changes everything. Based on this data, if it all holds true, haste should be weighted much more for both DPS and tanking. Theoretically, haste would give quite a large boost in generated rage. Now if only tanks could be rocking 30% crit rates...

Tizack
04-18-2008, 04:28 AM
Very nice post. I was looking for a forum to find out if haste did affect Rage gen. since the sword from kalecgos has haste on it. I still believe that haste can help threat generation though and i am glad to ahve most everything cleared up nice read and thanks much.

Tanksinatra
04-18-2008, 11:57 AM
my perception of reality has been shattered, great stuff bagel! :D

now someone get this man some passive haste gear so we can find out how that affects rage generation

tobarstep
04-18-2008, 03:09 PM
This seems more logical to me than the argument against it, to be honest. I mean, haste increases your attack speed. Haste is not found solely on weapons, so it shouldn't increase the attack speed of the weapon that you are using, if that makes any sense.

Corbusier
04-19-2008, 01:50 PM
Was there ever any data presented to the contrary?

I know a lot of people ran around saying "this is how it is, dammit!", but I don't recall ever seeing any evidence.
Only data from around when BC first came out, when our rage formula was being borked, fixed, and reborked.

Radhja
04-20-2008, 05:47 AM
Very interesting. I've wondered before if rage was always determined in whole numbers or fractions... this is very informative, and possibly spec-altering.

So, does this ultimately mean that stacking Haste in itemization is better than taking Rage-generation talents like Unbridled Wrath, Anger Management, etc.? (It certainly doesn't affect rage-cost reduction talents like Focused Rage still.)

Very good read. Time for some theorycrafting!

Jathine
04-28-2008, 10:08 PM
Reposting this from the warrior forums on Cider's request:
WoW Forums -> Haste-Rage Relationship (http://forums.worldofwarcraft.com/thread.html?topicId=6214202322&postId=62134635632&sid=1#5)

When Corb put up his thread about his Fury Guide and was taking suggestions for topics I offered help with determining how Haste effects Rage generation. I was pretty busy at work for the next week or so and was dragging my feet on getting it done.

I saw a thread on the Tankspot forums that covered this topic before I could finish my testing and do a write up. Here is the thread by Bagelbite: http://www.tankspot.com/forums/theory-articles/36614-rage-generation-haste.html He was using the same basic methodology that I had been using, but I thought that he was being a little too limited in his testing. Namely in his weapon choice. He only used one type of weapon and with a very narrow damage range. From reading his write up I came away thinking that he also didn't do very many repetitions. He also didn't do any tests to see if rage behaved differently when mulitple Haste effects were stacked.

To modify his methodology a little I decided to use 4 different weapons. I used the MH Hakkari Warblade (1.70), Talon of Azshara (2.70), Halberd of Smiting (3.50) and Stormherald (3.80). I did this to make sure that the Rage equation didn't behave unexpectedly at different values for weapon speed. Just from looking at the equation's graph I believed that it wouldn't, but I wanted to make sure. I have seen equations act a little differently than I expected before in various classes. A little extra work to make sure seemed to be worth it.

To further change his method of testing I did two additional tests. It dealt with my hang up about seeing how the equation behaved when multiple Haste effects are stacked. For this reason I did a test with Flurry and the Abacus of Violent Odds up at the same time. I did another test with Flurry, Abacus and my Troll racial Berserking up at the same time. I did these test with only the Talon and Halberd. Mostly as a quick check for my hypothesis.

This brought me up to five different experiments. I did a control group (no Haste effects) where I did two runs with each weapon. I then did the experiment with the Abacus two times with each weapon. Then came the two runs for each weapon with 5/5 Flurry. I did the stacked Haste experiments as described above.

To record my data I did a little digging to figure out the best way to do that. My first stop was Curse to see if there was an addon that showed how rage gains on a per hit in a visible way. I found one mod that was around a year or more old and decided it wouldn't do the trick since it was so outdated. If I remember correctly it was last updated around the time the Rage Normalization went through.

I started looking through the mods I already had loaded up. Under my SCT menus I found an option to show all power gains. I checked the box and went out to hit a few mobs to see if it would work how I wanted. I took a few swings at a mob in SMV and saw “+13 Rage” and “+7 Rage” scrolling on the left side of my SCT. I asked the author of SCT if there was a way to log this data (because it doesn't show up in any combat log data for some reason).

I never received a reply from him so I resorted to Fraps'ing each test and then using the good ol' pen and paper to record the data for each hit. This made things go pretty slow, but it allowed me to review exactly what was going on as many times as needed. By doing this I was able to take larger samples with each attempt. This allowed me to see if any oddities would occur over due to the rounding process for Rage.

Here is a sample of my data for the Talon of Azshara:


Speed Damage Rage

2.70 471 11

2.70 469 11

2.32 472 12

2.32 474 11

2.16 450 10

2.16 480 12

1.85 541 12

1.85 553 12

1.68 449 11

1.68 486 11

The first set is from the Control, the second set is from the Abacus, third from 5/5 Flurry, fourth from Flurry+Abacus and the last set is from Flurry+Abacus+Berserking. The Rage is the actual in game Rage gain that you can use.

Some of you might have noticed that in one case 450 damage generated 10 rage while in another case 449 damage generated 11 rage. This is because the game always rounds the fractional portion of Rage down and reports it for the Rage gained. Bagelbite saw the same thing while he was doing his test. He hypothesized that the game carries the fractional Rage (but still rounding down for usable rage) until there is enough to add 1 Rage to the swing. From looking over all my data I have to agree with him.


Conclusion:
Just from looking at the Rage generated from the different speeds anyone can tell that in no situation does Hastened weapon speed change how the Rage equation behaves. The only reasonable conclusion is that it always uses the base weapon speed no matter how many Haste effects are in play.

When I sit down and think about how the Rage equation works, I always come to the same conclusion: it has to always use the base speed. This is because there is really no way to efficiently to account for OH Flurry procs instantaneously increasing the MH's attack speed mid swing. The game would have to have some sort of "pro-rating" system in place to weight the length of time for each speed to determine the final rage generation. In the length of time that can happen when using non-matching weapon speeds it would be impossible for the servers to do the calculations with lag factored in. Base weapon speed is the only logical choice to use for the Rage equation.

Eh, it didn't copy the table over very well. It reads better on the warrior forums post. The link can be found at the top of the post if you are interested.

Kazeyonoma
04-29-2008, 08:55 AM
Edited your post a bit to make the table look nicer ^_^

Jathine
04-29-2008, 12:19 PM
Thanks! I was pretty drunk when I copied my warrior forum post over and didn't have the patience nor mental capacity to make that table look right.