I believe this calls for focused testing. It would be good to test with and without using special attacks, as well as with and without an offhand. I am not a statistician, so I can't say what the margin of error is, but 10k attacks should probably be the bare minimum.

Not just one sample of alot of attacks, but rather as many as possible to be honest. Variance can more than account for the OPs figures. I would love to volunteer but have no shadowmourne

Getting the ball rolling a bit on a few of those tests:

Dual wielding, no abilities, level 70 target dummy:
World of Logs - Real Time Raid Analysis
1147 fragments
2558 hits_____
44.84% chance to proc a fragment


Mainhand only, no abilities, level 60 target dummy:
World of Logs - Real Time Raid Analysis
1080 fragments
1440 hits_____
75.00% chance to proc a fragment

Hopefully I will be able to find time to do more testing this weekend, but for anyone wishing to contribute to the testing please specify if you used an offhand or not, abilities or not, and what level the target dummy you used was, then your personal data (fragments gained/number of hits) then also show your data added to mine, to create a running total. To eliminate eliminate Chaos bane skewing the data, make a cancelaura macro for soul fragments and just make sure to use it before you hit 9 fragments.

Ran a couple statistics tests for this data (much more is needed for it to be accurate). Mostly a 2 proportion Z test, calculating the correlation between the two sets of data, (level 70 and level 60).
The data that I obtained stated that there is absolutely no correlation (approximately 0, Z value of 18.4) between the two values. My hypothesis was that proportion sub1 =/= proportion sub2. Here's the data:


2 Proportion Z Test
p1 =/= p2
zstat- 18.429
pval- 8.1209x10^-78 (0)
p1hat-.75
p2hat-.4484
phat(pooled)- .5570
n1-1440
n2-2558

2 Proportion Z Interval
C-Level- 95%
ME-.0295
p1hat- .75
n1- 1440
p2hat- .448397
n2- 2558
phat difference (pooled)- .301603

C-Interval-{.2721, .3311}

We can be 95% confident that the pooled data's proportion falls between {.2721, .3311}. Since it does not include 0 or both negatives and positives, it is highly statistically significant in the fact that there is a large difference. All this data points to the fact that there is no correlation whatsoever between both data sets, thus leading to the conclusion that there is a very large difference between the tests on a level 70 dummy and level 60 dummy, or using a main hand and an offhand. The data could be effected by either variable. If tested, to avoid data skewing and bias, run tests based on whether it is a mainhand/ offhand on the same level training dummy, then test the level difference for the procs.

In about 4 weeks I can provide data as well. Aye it's a long wait. /sigh. Need data for correcting values in the spreadsheet to summarize stat-scaling implications of the weapon.

To try and isolate the cause I eliminated a variable

Dual wielding, no abilities, level 60 target dummy:
World of Logs - Real Time Raid Analysis
1303 fragments
2941 hits_____
44.30% chance to get a fragment

Now obviously the sample sizes are still pretty small but it appears that having an offhand lowers the overall chance to get a fragment. Over a given time-line I wouldn't be surprised if the number of fragments gained with and without an offhand would end up being very similar, due to some sort of normalization mechanic.

Were you using a slow offhand by chance? For all of my dual-wielding tests I was using a 1.5 speed axe. If there is some sort of normalizing mechanic that determines fragment gain over time, then using a fast weapon like I was would effectively lower the chance per hit to get a fragment, which could explain the discrepancies between our two logs.

Yeah, I used a 3.6 speed offhand. Guess that makes sense, will test it later.

Level 60 target dummy, dual wield, no abilities, 3.5 speed offhand:
World of Logs - Real Time Raid Analysis
1977 fragments
2604 hits______
75.92% chance to get a fragment

Taking it a step further, if you divide the total number of fragments gained by active time, you get fragments per second:
1977 fragments
3696.8 seconds______
0.53478 fragments per second.

Now when I do the same thing to my previous parses using the fast offhand:
70 target dummy:
1147 fragments
2287.7 seconds______
0.50137 fragments per second.

60 target dummy:
1303 fragments
2450.3 seconds______
0.53177 fragments per second.

This again points to there being a normalization mechanic in place to control fragment gains over time. Working in terms of PPM, auto-attacking appears to cause fragments have ~30PPM.

edit:
Combined total:
4427 fragments
8434.7 seconds____
0.52485 fragments/second

To cut out RNG, I would suggest we keep testing until we have about 10k fragments gained total from just auto attacking. A second running total might be necessary for fragments/second while using abilities as they more than likely have a big effect on fragments/second. I wouldn't be surprised in the slightest if it more than doubled due to using abilities.

Hey, I've recently been jerking around alot in rawr/simcraft to try to find an optimal combination of gemming/enchanting my gear and found something rather odd.

When I go close to the hard cap (~1350arp) the normalized value of each point drops to a minimum (~0.4 according to simcraft) and even when I'm at around 1300arp the value of each point is less then that of strength.

Is there any existing math done on the matter, Is everyone totally obvlivious of it since most guides say "stack until hard cap or softcap with trinket" or am I doing something wrong?

Note: I'm aware that the hard cap is ~1400, the value I was using is ~1350 though.

Did you have the food set to armor pen food? If that was the case, and you were sitting at say 1359, it would treat you as hardcapped, and calculate the SEP value accordingly.

Silly question, but is there a "stat increment" setting anywhere for Rawr/Simcraft? If they're calculating the value of 100 points of ArP (for example) instead of 1, you're going to run into boundary conditions.

I have 2 simple questions that may be considered dumb, but I have to ask them, just to make sure I don't make a dumb mistake.

1. I have seen rogues use saronite Bombs in their rotation, this is easily a (correct me if I'm wrong on the number) 200 dps increase? more or less...Is this viable for fury warriors, if so, I've never leveled Engineering so I don't know, is the bomb on the GCD or its on separate?



2. Question answered

Global thermal sapper charge on trash is fun (BIG NUMBERS!), 6 min cool. lets u do 5 saros (or 5 cobalts 3 sec stun, a good throw on blood beast is nice) in between, 1 min cool - bombs share cools ofc. bombs aren't on global gcd - saro and cobalt tho need aiming - do it wen gcd happening anyway.

the speed gloves are like an extra trinket, bind to DeathWish and hit second time they available as well. or use hand rockets for running attacks.

boots get crit + 5 sec run boost, with heroic fury = uber mobil.

raid utility: repair bots, jeeves, mailbox.

general utilty: wormhole handy for a daily.

personally, i think JC and Eng is a good fury set up prof-wise.

I have been a long time browser of these forums. I recently came back to the game I have a question that I have been unable to find any information about after extensive research. I recently started gearing my warrior for raiding. I have read Landsoul’s Fury Warrior Compendium as well. My warrior is Orc and there is a section in the guide about Orcs and the benefits they provide concerning their racials. I was looking for a bit of clarity on a point.

If I am using an Axe in my MH (+5 expertise) and a Mace in my OH (which would give me 26/21 expertise) would I need to gear to cap my OH expertise or would I just need to ensure my MH expertise is capped and then concentrate on dps stats? I ask this as since I am just coming back and new to warriors my best options at this point are either a Bryntroll or Shadow’s Edge for my MH (Shadowmourne is out of the question).

Investing in expertise just for OH capping is a waste of stats. Go for str/arp instead, once MH is expertise capped.

About all this Shadowmourn proc thing, I'm actually wondering if someone did some tests to know if canceling Chaos Bane buff was worth it for single target DPS. It's pretty obvious in multi target phases it's way more effective to cancel it, and that Rets for instance do it even in single target situation, but as a Warrior I'm quite unsure about it.

Anyone having some input about it ?

Would you please explain this part? Also, there was some discussion about this in the spreadsheet thread.

Using the numbers from Zodiac's testing a few posts up, we can take an assumption we gain 30 strength(before modifications) every 2 seconds. At 9 stacks, you have 270 strength(30x9 = 270 strength). At the 10th stack, we get chaos bane which gives us 270 strength. I don't have Shadowmourne yet so here is where my assumptions come into play.

1) Assuming the worst case scenario of the fragments only being generated every 2 seconds and as long as chaos bane isn't active, we get about 12 seconds with 270 strength. Now we have to rebuild the stacks.

2) Assuming the Blast does 2k Shadow Damage at all times and doesn't scale, that's 2k damage about every 30 seconds or a 67ish DPS increase.

Now by cancelling it, we have the first 20 seconds the same as usual, however we lose the third section where we have 270 strength for 10 seconds. So we get the debate 270 strength over 10 seconds vs. 50% more 2k blasts.

Finally able to do a little bit more testing on shadowmourne, time to start working on how special attacks affect fragment gains. My method for testing this was to spam hamstring and heroic strike.

World of Logs - Real Time Raid Analysis
1535 fragments
2038 hits ________
75.32% chance to proc
Looking at the time interaction:
1535 fragments
1508 seconds ________
1.0179 fragments per second

When I get some more time I will work on getting a bigger sample size, though it wouldn't surprise me if with abilities the fragment gain averaged 1 fragment/sec.

EDIT:
Did a short test to make sure windfury wouldn't change the overall outcome:
World of Logs - Real Time Raid Analysis
849 fragments
848 seconds_______
1.0011 fragments per second

Making the running total for using abilities:
2384 fragments
2356 seconds_______
1.0118 fragments/second

There is one lurking variable when it comes to Shadowmourne and Fury Warriors.

TGM, if you are able to actually run the regular fury rotation, which I'm sure you're working on at the moment, you would have to incorporate the 4PcT10 into the equation due to the GCD reduction. I'm sure the 4% chance to proc will not skew the data very much, but it will have an effect on a large sample size.

Doing the standard DPS rotation:
World of Logs - Real Time Raid Analysis
926 Fragments
946 seconds_____
0.97885 Fragments per second

The reduced GCD from t10 4pc slams, and the fact that WW is two hits per use seem to have no effect on fragments gained per second.

New running total:
3310 fragments
3302 seconds_____
1.0024 fragments per second