The Change to Windfury and You! (does not apply to warriors) - Page 11

Hellcry · 05/15/07, 12:03 AM

Dling latest PTR patch atm. Anyone done any tests?

Sure it feels abit "Onyxia is doing more deep breaths" over this one, but we can always hope...

http://forums.worldofwarcraft.com/th...02907913&sid=1

If someone here did some testing using the latest ptr build Id be very happy if you could post your results. A "fix" I would try to do myself if I was blizz would be to implement some sort of alghoritm that increases the procrate each time a non proccing hit lands (kinda like the new Darkmoon Card: Wrath).

◊ Rob · 05/15/07, 12:49 PM

Originally Posted by Hellcry

Dling latest PTR patch atm. Anyone done any tests?

Sure it feels abit "Onyxia is doing more deep breaths" over this one, but we can always hope...

http://forums.worldofwarcraft.com/th...02907913&sid=1

Reading the thread, there are quite a few posts from people confirming that things are as they always have been on the PTR (and WF5/WF5 is the same on PTR as it is on live).

Lujaar · 05/18/07, 4:45 PM

I'm still not sold on slow offhands being significantly better than fast ones.

We know that the amount of test data required to draw conclusions about windfury is prohibitively large. There's no enhancement DPS spreadsheet for a reason - unless a tester is willing to spend days on end AFK in the Blasted Lands, the results of any given test are, essentially, random. I could claim that windfury DPS is based on the alignment of the planets or what time I brushed my teeth that morning, and someone would have test data backing it up. I'm sure that for every test out there showing that slow weapons are better than fast weapons, there's a test showing that higher DPS weapons are always better, but we don't see them, because higher DPS weapons being better isn't news.

Maybe slow offhands are better, maybe they're not. I don't know, but I can't see why they would be, at least not to the degree that many people are claiming. We know that offhand procs sometimes proc mainhand windfury and vis versa, and we have no way of knowing based on combat logs whether any given proc came from a mainhand or offhand swing. If we can't trust test data, how do we even know it matters whether the windfury procs come from your mainhand or from your offhand?

drats · 05/19/07, 3:59 AM

Lujaar, Disquette has been making a shaman dps simulator script:

[Shaman] Itemizing Enhancement at 70

The quick answer is that slow offhands are better than fast ones. I didn't really believe it either until this week. Pick up a fast dagger and slow cleaver and try it our yourself.

When my MH was 2.2 and my offhand was 1.8, I found that my offhand was using a lot of my WF procs. I actually think the bug isn't as simple as WF5/4 working around the lockout, but that's a different matter. What's important is that I switched to a 2.6 offhand (a green that is only 59dps) and saw a significant increase in dps. Since I probably won't get a chance to clear Gruul again before the patch, I'll have to run some instances to get more specific numbers.

If there is a 3s cooldown on WF, you want your mainhand to attack as close to 3s as possible. The best way to achieve this is get a weapon that has a flurried speed of 1.5 or slightly above. In the WF5/5 world, this will basically ensure that your offhand won't be eating your mainhand windfury procs very often.

Lujaar · 05/19/07, 4:44 PM

Very nice. At least this script puts to rest some of the more, er, extreme claims about the windfury cooldown. Modelling the DPS of a slow lv10 white offhand over 10000 seconds showed really terrible damage, for example.

However, looking over the script, Disquette's model doesn't take into account the possibility of OH procs proccing MH windfury or the other way around. To me that's the sticking point. We know that it happens. We've all had our jaws drop at 4 simultaneous procs so big they could only be mainhand, and I've seen 2 simultaneous mainhand procs using 5/5 as well. Unless someone knows what circumstances cause one hand to proc windfury on the other, we're not going to get reliable data from scripted models.

Originally Posted by drats

Since I probably won't get a chance to clear Gruul again before the patch, I'll have to run some instances to get more specific numbers.

If you're testing during actual gameplay and getting any significant amount of data (several hours of attack time, at least) you're going to be either farming or killing trash. Frequently starting and stopping DPS like that will skew your calculations toward slower being better anyway. That's not to mention that if you're running instances, your DPS is going to have more to do with the tank's threat generation than with your weapons.

• Disquette · 05/20/07, 8:16 AM

the script assumes that the bugfix in patch 2.1 will allow OH to proc only OH windfuries, and MH to proc only MH windfuries. The script doesn't tell you anything about how dps should be in 2.0, only in 2.1.

witchfire · 05/20/07, 4:24 PM

If you're testing during actual gameplay and getting any significant amount of data (several hours of attack time, at least) you're going to be either farming or killing trash. Frequently starting and stopping DPS like that will skew your calculations toward slower being better anyway. That's not to mention that if you're running instances, your DPS is going to have more to do with the tank's threat generation than with your weapons.

Unless you test on one of the servants in blasted lands that only die if you kill their pylon.

If there is a 3s cooldown on WF, you want your mainhand to attack as close to 3s as possible. The best way to achieve this is get a weapon that has a flurried speed of 1.5 or slightly above. In the WF5/5 world, this will basically ensure that your offhand won't be eating your mainhand windfury procs very often.

Then why enchant dual windfury in the first place? Either the bug fix will put a shared 3 sec cooldown on WF and there will be no reason to enchant the off hand with it or the cooldown will be seperate on the 2 weapons and the OH won't steal Main Hand procs.

• Sebudai · 05/20/07, 5:09 PM

Because Windfury is that good. Off hand Windfuries for half damage are still going to be better than any of our other weapon buffs, unless your weapons/gear isn't very good. You're going to be sitting outside of the 3 second cooldown quite a bit with Windfury only on your Main hand weapon.

Droo · 05/20/07, 7:26 PM

How much raid damage do our buffing abilities actually account for? I've seen people suggesting it's around 10%~ of rogue/warrior damage, but it'd be nice to have a well researched figure if I end up doing less damage than the tank in 2.1.

Lujaar · 05/20/07, 7:38 PM

Another oddity: Disquette's damage modeller shows that for me, WF/FT is superior to WF5/WF5. I'm offhanding a Malchazeen, and the model showed about a 50 DPS increase from Flametongue. I tested with Runic Hammer as well and got similar results. With the Gladiator's Cleaver the damage from WF5/WF5 and WF/Flametongue were about the same. My gear's not bad by any means, so it's not a scaling issue.

I haven't played around much with WF/FT and last time I did I remember struggling to keep up on damage meters in a kara group where I was normally fighting our warlock for top. The disparity could be a flaw in the way the script models flametongue (no offense Disquette) or the offhand-proccing-mainhand thing, or maybe just a bad night the night I was testing WF/FT. Regardless, I should probably play around with WF/FT more.

Originally Posted by witchfire

Unless you test on one of the servants in blasted lands that only die if you kill their pylon.

Yeah, I know about that and all. I probably should have been more clear on what I meant by "actual gameplay" - logging your DPS while going about your normal business.

Originally Posted by Disquette

the script assumes that the bugfix in patch 2.1 will allow OH to proc only OH windfuries, and MH to proc only MH windfuries. The script doesn't tell you anything about how dps should be in 2.0, only in 2.1.

I'd have to jump on the PTR to see if that's the case, but I know that using 5/5 on live you will occasionally see one weapon windfurying off a proc from the other. It's hard to tell how often though, because the only time I can tell for sure is if I see a double proc, and both procs are either too large to be offhand or too small to be mainhand.

• Nite_Moogle · 05/20/07, 10:53 PM

Unless you test on one of the servants in blasted lands that only die if you kill their pylon.

Which will horribly skew your results due to level difference. They aren't useful for testing anymore, log parses are pretty much the only way to go these days.

• Disquette · 05/20/07, 11:20 PM

Hey Lujaar, for our reference in figuring out why FT is doing so well for you, would you mind posting a pic of the results pop up on top of the configuration page, much like this following pic:

http://www.discofiend.com/wow/Shaman...ucidin_dps.jpg

this way we can see both your parameters and the output. As an aside, i've always said that my calculations for FT aren't nearly as good as my WF ones, because I put more effort into the WF part.

Oh, and one thing to check because it's hung up another poster before - armor mitigation paramater is: 1 = all damage goes through. 0 = no physical damage goes through. So, a mob has 20% armor mitigation, you'd put in .8 for that stat (i named the variable very counter-intuitively)

mek · 05/20/07, 11:37 PM

Originally Posted by Nite_Moogle

Which will horribly skew your results due to level difference. They aren't useful for testing anymore, log parses are pretty much the only way to go these days.

The results won't be a proper representation of dps vs a lvl 73 mob, but it won't be skewed favouring any particular weapon or enchant configuration, so for generating a comparison between different weapon layouts, it's a perfectly fine testing method.

slant · 05/21/07, 12:40 AM

Well, no, not really. Your melee missrate against a clvl+3 mob is +0.6% while your spell miss ("resist") rate is +17%. So tests against lower level mobs will make flametongue/frostbrand look significantly better than they really are.

Lujaar · 05/21/07, 2:15 AM

Disquette, here are my test results.

WF5/Flametongue: http://www.fusionpics.com/show.php/38906_wf5ft.jpg

WF5/WF5: http://www.fusionpics.com/show.php/38907_wf5wf5.jpg

These were fairly typical results. Values are for Black Planar Edge MH/Malchazeen OH, using something close to my AP/crit/hit pre-2.1 gear buffs.

• Disquette · 05/21/07, 9:21 AM

Originally Posted by Lujaar

Disquette, here are my test results.

WF5/Flametongue: http://www.fusionpics.com/show.php/38906_wf5ft.jpg

WF5/WF5: http://www.fusionpics.com/show.php/38907_wf5wf5.jpg

These were fairly typical results. Values are for Black Planar Edge MH/Malchazeen OH, using something close to my AP/crit/hit pre-2.1 gear buffs.

thanks very much, it does indeed help me figure out where the craziness is.

In this case I'm wondering if this was an early version of the script where I didn't say "use 36% chance of WF proc'ing *only* if I have it on both weapons, otherwise use 20% chance of proc".

As an example of why i think this, I just ran the test with 10x the time limit you did....

WF/FT
lujaar
FTDPS    WhiteDPS    WFDPS
 53.5         321.9        208.8

disquette
FTDPS    WhiteDPS    WFDPS
 53.2         323.5        143.3


WF/WF
lujaar
FTDPS    WhiteDPS    WFDPS
   0            331.4       195.0

disquette
FTDPS    WhiteDPS    WFDPS
   0            331.0       203.1

So, the *only* number that's out of whack there is WF on your FT test. I haven't updated my code for distribution, in large part, because Pater has been making a java version of it, much like Friedrich made a C version of it.

I'm really sorry that I didn't announce it earlier, but to tell the truth, based on how few people were downloading and using it, I figured I'd just run the sims with my updated/fixed code when people wanted some info.

So, if you want to play with some messier code, I'm posting it here:
http://www.discofiend.com/wow/Shaman...2007-05-21.vbs

The differences are many, including the reporting (you dont get a nice pop-up, it exports a text file), and you *have* to specify end values for the same variables you do input for. Here's a sample of the code, so that people see how I do the looping for all the important parameters by only inputing my test parameters once. If you want a single event test, you simply make the start and ending values for all the parameters to be the same (shown below):

Do While MHWSpeed < FinishMHspd + 1
	Do While OHWspeed < FinishOHspd + 1
		Do While Crit < FinishCrit +.001
			Do While HitBonus < FinishHit +.001
				Do While AP < FinishAP + 1
					AttackSimulation
					AP = AP + 100
				Loop
			AP = StartAP
			HitBonus = HitBonus + 0.0316
			Loop
			HitBonus = StartHit
			Crit = Crit + 0.0225
		Loop
		Crit = StartCrit
		ohspdpercent = 200 / OHWspeed
		OHWspeed = OHWSpeed + 200
		OHAvg = round(OHAvg * (1+ohspdpercent),1)
	Loop
	OHWspeed = OHStartSpeed
	OHAvg = OHStartAvg
	mhspdpercent = 200 / MHWspeed
	MHWspeed = MHWSpeed + 200
	MHAvg = round(MHAvg * (1+mhspdpercent),1)
Loop

So, in the input section, you now see:

HitBonus = 0.176 ' Your hit bonus:  .251 = 25.1% bonus to hit chance
Crit = 0.229 ' Your crit rate:  .251 = 25.1% chance to crit
Glance = 0.0 ' chance to have a glancing blow.  .25 = 25% of white attacks will glance
AP = 1352 ' unbuffed AP (or put in raid buffed AP, w/e, but dont include Unleashed Rage)
DodgeBlockParry = 0.00 ' chance to have an attack avoided.  .25 = 25% of white attacks will glance
MHWSpeed = 2700 ' Main Hand Weaponspeed 1500 = 1.5 speed, 2600 = 2.6 speed, etc
OHWSpeed = 1800 ' Off Hand Weaponspeed 1500 = 1.5 speed, 2600 = 2.6 speed, etc
MHHigh = 320 ' top end damage of Main Hand Weapon
MHLow = 172 ' low end damage of Main Hand Weapon
OHHigh = 199 ' top end damage of off Hand Weapon
OHLow = 132 ' low end damage of Off Hand Weapon
StartHit = HitBonus
StartCrit = Crit
StartAP = AP
.....

.....
FinishAP = 1352
FinishHit = .176
FinishCrit = .229
FinishMHspd = 2700
FinishOHspd = 1800

Friedrich · 05/21/07, 12:41 PM

Originally Posted by Disquette

t
So, the *only* number that's out of whack there is WF on your FT test. I haven't updated my code for distribution, in large part, because Pater has been making a java version of it, much like Friedrich made a C version of it.

Perl not C =) . C would require much more work, because string manipulation in C is very cumbersome.

I think the most telling thing about Lujaar's results is that his total Windfury DPS actually went *up* when comparing his WF5/5 test to his WF/FT test (from 195 to 209). There's no way that can be correct unless you fuck yourself over to a far greater extent than I had thought possible by using a very fast offhand with linked-cooldown Windfury.

EDIT: Oh and Lujaar you're like my evil horde twin!

Lujaar · 05/21/07, 4:07 PM

Thanks Disquette. That would account for the difference.

I'd be curious to see someone's percentage of mainhand procs vs percentage of offhand procs in a live test with a slow MH/fast OH. Testing number of procs should be a bit less prone to unknown elements modifying the results than a raw DPS test.

My mainhand's swing speed is 2.7, my offhand's is 1.8. With no shared cooldown and each weapon always proccing windfury off itself, I'd see something like 60% offhand, 40% mainhand. With the shared cooldown, I should see even more offhand, even less mainhand.

However, there's a possibility that either hand has an equal chance to proc either weapon, and that's what I want to test for. If that's the case I should see about 50% mainhand/50% offhand with any combination of weapons.

Anybody know of a mod that would track that? I'm gonna see what I can put together myself in the meantime.

• Disquette · 05/21/07, 5:05 PM

Oh, sorry Friedrich, Perl indeed!

As for your test, Lujaar, what I did was equip all my resto gear, then head to a blasted lands mob, drop a healing totem (hence the resto gear and very large healing stream totem ticks). I used a 1.2 dps weapon in one hand, and a high dps weapon in the other.

Because you're in healing gear without much attack power, the main difference in the damage per swing is the weapon itself. As a result, it's really easy to put the /combatlog into excel and classify each hit as MH or OH.

I haven't done this for a long time, however, and if you are willing to take the time to do this, I think we'd all be very interested in verifying that MH and OH proc in inverse proportion to their speed, with a slight nod to MH procs because (I'm guessing) that simultaneous hits check for MH proc first.

Originally Posted by Friedrich

Perl not C =) . C would require much more work, because string manipulation in C is very cumbersome.

I think the most telling thing about Lujaar's results is that his total Windfury DPS actually went *up* when comparing his WF5/5 test to his WF/FT test (from 195 to 209). There's no way that can be correct unless you fuck yourself over to a far greater extent than I had thought possible by using a very fast offhand with linked-cooldown Windfury.

EDIT: Oh and Lujaar you're like my evil horde twin!

And yes, if your MH WF % goes up to 36% per illegible hit (instead of 20% if you're single-handing WF), then you do indeed get a very nice dps boost.

In fact, that might be a really nice way to fix windfury: Windfury on the offhand has the affect of increasing the WF on mainhand proc rate by 20% (additive, not multiplicative)

That would fix pretty much everything, I think. No more fixating on weapon speeds - they'd still matter, but not nearly as much as they do now.

Lujaar · 05/21/07, 7:02 PM

OK, I've managed to modify the ShamanFriend mod to keep a running tally of mainhand and offhand procs.

EDIT:

Did my first test in Blasted Lands tonight. One hour, virtually uninterrupted (stopped swinging to throw 5 or so heals, but that's it).

Test conditions: WF5/5, full enhancement gear, no AP buffs. Stoneskin, GoA, and Healing Stream Totem. My assumption is that haste effects - flurry and Dragonspine Trophy - don't matter for this test. I don't care how much DPS I do, or what my overall WF proc rate is. All I want to know is what percentage of my procs are mainhand, what percentage are offhand.

Mainhand: 2.7 speed
Offhand 1.8 speed
(40% of my white swings are mainhand, 60% are offhand)

I'm using the mod Shamanfriend, modified to track WF procs as individual swings rather than adding multiple swings together. I have it set to keep a running tally of mainhand and offhand procs. Anything over 300 damage is considered a mainhand proc, anything below is an offhand proc. My offhand procs range from 200-250, my mainhand are about 500+, so I'm confident that I'm not misattributing swings from one hand to the other. Crits are divided in half before they're tallied as mainhand or offhand.

Results:

Mainhand procs: 594 (43.36% of total)
Offhand procs: 776 (56.64% of total)

Based on these numbers, having windfury on my (fast) offhand isn't lowering my percentage of mainhand procs. 40% of my white swings are mainhand, and over 40% of my windfuries are mainhand.

Sinyen · 06/05/07, 4:37 AM

I recently picked up a FuBar Addon that tracks WF Procs (WindfuryFU), and I noticed a really strange result. WF Proc Rate was 15% (using WF5/WF5) with a large sample (and it was consistantly 15%). So, Im unsure if the addon isnt reporting things right, WF was nerfed, or the more likely scenario - I'm misinterprating the addon info.

I would think the Proc Rate would be at least 20% if not more since Im using WF for the OH also. Whats going on?

Here is a screenshot of the addons detailed breakdown.

I appologize in advance if this has been discussed before, I couldn't find any threads discusses this.