I'm actually wondering how wrong it is if you have slower OH weapon than MH? Even if its 0.1 difference?
Did some research on that but couldn't find any useful up-to-date info. These days rage generation shouldn't be an issue but still I'm curious.

So far I was using Shadow's Edge on MH and normal Cryptmaker on OH. But today I got heroic cryptmaker in my hands.
I just switched SE for offhand for now but I'm wondering is Shadow's Edge 3.7 speed appropriate for offhand?

Weapon speed isn't really the deciding factor on what you use as an offhand or main hand weapon. It's the damage range on the weapon. If a weapon is of same Ilvl the slower weapon will have the higher damage range and thats why we chose it. There's nothing wrong with having a slower offhand weapon the only thing it would have a negative effect on is your rage generation and well thats not an issue.

Your best bet is to spreadsheet it. But if you don't want to do that, you can determine which should go in mainhand just by putting each weapon in your main hand,and see what the total damage is. Whichever one is higher goes in the mainhand. (Be sure to check it with full raid buffs, because the speed difference can cause the slightly slower weapon to pull ahead with more attack power).


As a basic example, let's figure you have 8500 raid buffed attack power.

With Heroic Cryptmaker you have 3.6 speed, base damage of 938-1407. That comes out to 3123-3592, average hit of 3357. (before two handed weapon specialization and the titan grip penalty)

With Shadow's Edge, you have 3.7 speed, base damage of 872-1309. That comes out to 3118-3555, average hit of 3336.

The difference is ultimately pretty minimal, but in this scenario cryptmaker is ahead by ~20. It looks as though Shadow's Edge picks up by ~14 relative damage for every 2000 attack power, so if you have an average attack power of ~10.5k with average procs/raid buffs, shadow's edge breaks even, higher and shadow's edge pulls ahead.

My math may be off by a bit for not accounting for the titan's grip/two weapon spec, but given both are straight multipliers the ratio should remain the same.

Yes it is, but not because of the reasons you cited.
BT is not dependent on weapon speed.
Neither is WW (this one depends slightly on base damage range though, which favours heroic Cryptmaker still).
Auto Attack, Heroic Strike and Cleave damage over time is largely independent on weapon speed either (ignoring the fact that they favour faster weapons because of the static ability bonus, but this is a miniscule detail here).

The only fury ability which favours slower weapons is slam. This is where Shadow's Edge profits from the 0.1 sec slower swing speed. But to overtake heroic Cryptmaker you'd need WAY MORE buffed AP than 10.5 K. Though I'm too lazy now to make an educated guess. This obviously depends on your ratio of slam damage to total damage.

Heroic Cryptmaker goes in the MH. No questions asked.

Thanks for answers guys but I think I havent make it clear enough.
It's obvious that HC cryptmaker going to MH. Didn't even thinking about using it other way. My dillema was more about offhand choice like SE versus the same dps/ilvl weapon but different speed (e.g. heroic Ramaldni's). All I was confused about was offhand speed: 3.7 or 3.5 while having 3.6 speed mainhand. That would mean quicker HSing that offhand generation at some point.

But you answered me by the way that OH speed doesn't make difference at all so thank you

Um, BT doesn't care what your weapon damage is at all. BT doesn't care if you have a ilvl 185 weapon or Shadowmourne, it goes based on attack power. Bringing it up is irrelevant. Both Whirlwind and Slam go based off of your average damage per hit, which is what I was comparing. Slam and Whirlwind both favor slow weapons equally. Slow weapons are always better, assuming all else is equal.

In the case of heroic cryptmaker vs shadow's edge however, all else is not equal. You have a dps difference in favor of the cryptmaker. You always choose what your higher damage per hit weapon is for your main hand weapon to maximize the damage done by whirlwind/slam. I just relooked at the numbers, and you are right that 10,500 is not the correct cutoff point. But basic math says that at some level of attack power, the slower weapon will pull ahead, it's just a matter of how much attack power that is. I did double check the numbers, and 10,500 is too low. Around 11,500 though the weapons hit equally hard, though while they're equal Cryptmaker is still probably ahead due to increased heroic strike frequency. At some point (I dont feel like calculating exactly where) Shadow's Edge would pull ahead, but it's not realistic to attain even 11.5k AP, much less more than that, so you are right that in general you can just keep Cryptmaker mainhanded and not worry about it.

Since the expansion is effectively coming to a close and Cata is on its way, I figured I'd expose two fairly major bugs regarding warrior DPS that I haven't seen mentioned before.

#1: Offhand swings that land while HS or Cleave is queued do not suffer the DW miss penalty. They're treated as if you were single wielding when the swing lands.
Example of 0 OH misses during simple target dummy experiment queuing HS before each OH swing lands: imgur: The Simple Image Sharer
Example Saurfang parse with 1 melee miss over the course of the fight(possible if I couldn't HS every swing or re-queued HS late, likely at the start of the fight and execute range): World of Logs - Real Time Raid Analysis

The bug has 2 important impacts on itemization. The first is that hit rating past the soft cap is far more useless than most TC imagines. I glanced through a lot of my logs and saw miss rates at about 25% of what they should be on average. Simulating the bug only occurring on about 75% of swings gives it an SEP of .1 SEP or less most of the time. The second impact is that the crit soft cap has a fairly negligible effect since it only comes in to play on those swings that the bug doesn't affect. The crit cap on swings when the bug is in play is an unreachable 1-.24+.048 = 80.8%.

How you deal with the bug depends on your weapon speeds. The ideal situation is your OH swing landing just prior to your MH swing, since this gives you the most time to re-queue HS each swing. The worst situation is your OH landing just after your MH, since you have almost no time to re-queue HS after your MH lands and before your OH lands. If your MH is slower than your OH, you will be in the ideal situation most of the time(3.7/3.6 shadowmourne/anything weapon setup is perfect).

With those weapon speeds as an example, your first swings when you engage a mob will be simultaneous(which usually makes the bug's occurrence random, happens about 50% of the time with synced swings). Then, your OH will land after 3.6s, with your MH .1s later at 3.7. The next round of swings, your OH will land .2s before your MH, then .3s, etc. It does a full cycle back to simultaneous swings when your OH swings 37 times to your MH's 36. If you bind HS/Cleave to mousewheel, you'll be able to get the queue done in time up until your OH is landing about .5s after your MH, or for about the first 29 swings of a cycle with typical raid buffed haste. That takes about 65s to occur. Any time you have to be off the target for long enough for your swing timers to hit 0 and be ready again, the cycle resets. So, in most fights you will always be in that optimal range without doing anything at all. In tank n spanks like festergut/saurfang, you can force the reset to occur early by spinning your character around just before your swings are going to land for a very short amount of time(letting them both hit 0 and be ready) or by using Heroic Throw(preferably just after your MH/OH swings have landed).

If you have same speed weapons, life is a bit harder. When you engage a mob(as in, right click it or /startattack-ing), your MH swing timer is immediately ready while your OH swing is set to half its duration(i.e. 1.8s with a 3.6 speed weapon). This means if you engage a mob from range and then approach it, your swings will be synced and landing simultaneously, which is bad since the bug occurs less frequently in that case(as mentioned above, like 50% of the time). If you don't engage a mob until you are within melee range of it though, your MH swing will be able to swing immediately, and your OH will be perfectly desynced from it, being 50% of the way through its swing time each time your MH swing lands, which gives you plenty of time to re-queue HS every swing. This is how I ran the target dummy experiment in the recount above. Unfortunately though, on fights where you're forced off target this setup will be broken and is difficult to reset to the nice desynced situation.

#2: You get free Deep Wounds damage ticks if the debuff is refreshed within about .2-.25s after a tick.

This one is a bit tricker to explain. If you've heard of the "Ignite Munching" bug, then you probably are already aware that DW munching occurs in the exact same fashion. Basically, if 2 crits land near simultaneously, when the 2nd crit updates the damage value for the DW buffer, it overwrites whatever the 1st crit wrote and ignores it.

I.e.:
0s: DW buffer is at 10k damage.
0.5s: MH WW crits, reads the 10k dmg value, recalculates the new dmg value to be 13k, and attempts to write that value back to the DW buffer.
0.5s: OH WW crits, reads the 10k dmg value, recalculates the new dmg value to be 11.5k, and attempts to write that value back to the DW buffer.
.7s: MH WW crit finishes the write, and DW buffer is set to 13k dmg.
.7s: OH WW crit finishes the write, and DW buffer is set to 11.5k dmg.

The obvious corollary is that good munching can occur when a crit lands immediately after a damage tick. In the same way as above, the new crit that lands recalculates the damage the buffer should have using the damage remaining in the buffer prior to the damage tick. I.e. DW buffer is at 12k. It ticks for 2k, and you land a crit immediately afterwards. The tick sets the new damage in the buffer to 10k, but the crit calculates using the 12k value, adds 3k, and sets the new value to 15k. Effectively, you get the 2k dmg from the tick for free since it is never removed from the buffer like it should've been.

This bug wasn't a serious concern until crit rates got as ridiculously high as they are now, since the bug's impact scales exponentially with crit chance. The DW buffer should reach an equilibrium point where the damage going in from crits matches the damage going out each second. So, more crits per second equals more damage going into the buffer, which means the buffer will be balanced at a higher level(meaning higher damage ticks each time it ticks). On top of that though, more crits also causes the debuff duration to reset more frequently, increasing the average time between ticks, which means the average tick will be higher yet. Note that this does not in any way make the damage output of DW any higher yet, since the damage going in is still just equal to the damage going out.

However, the good version of DW munching benefits from higher DW ticks. If you munch a 15k tick instead of a 2k tick, you're obviously getting a lot more damage out of each munch. So with higher crit rates, the DW ticks being munched are naturally higher, and the number of collisions between ticks and crits(causing the munches to occur) is also higher.

Examples can be easily found in anyone's parses. Here is that Saurfang log from above, with the DW ticks and refreshes parsed out:
World of Logs - Real Time Raid Analysis

It happens dozens of times throughout that log, but here is an especially clear excerpt to show just how frequently it can occur:
[20:57:35.961] Rallik Deep Wounds Deathbringer Saurfang 7835
[20:57:36.149] Deathbringer Saurfang's Deep Wounds is refreshed by Rallik
[20:57:37.243] Rallik Deep Wounds Deathbringer Saurfang 8344
[20:57:37.383] Deathbringer Saurfang's Deep Wounds is refreshed by Rallik
[20:57:38.430] Rallik Deep Wounds Deathbringer Saurfang 8852
[20:57:38.586] Deathbringer Saurfang's Deep Wounds is refreshed by Rallik
[20:57:39.602] Rallik Deep Wounds Deathbringer Saurfang 9170
[20:57:39.743] Deathbringer Saurfang's Deep Wounds is refreshed by Rallik
[20:57:40.789] Rallik Deep Wounds Deathbringer Saurfang 9678

You can see each refresh(a crit landing) occurring within about .2s of the preceding tick, and all 4 of these in a row munched the tick, not only giving me the damage of that tick for free, but also making the subsequent munches more valuable since they munch a higher damage tick. In case you were wondering, you can prove the refresh alone isn't enough to account for the damage values you see there with some simple math.

Taking, the last munch there for an example since it's most obvious:
The tick before was 9170 damage and was the first tick after being refreshed, meaning the damage in the buffer before the tick had to be 9170*6 = 55,020, and the buffer should be set to 55020-9170 = 45,850. The next damage tick was 9,678 after a refresh, meaning the damage in the buffer had to be 9678*6 = 58,068. If no bug occurred, it means that single refresh increased the damage in the buffer by 58068-45850 = 12,218. Since DW is 62.4%(including mangle debuff) of the average swing range of the hand that procced it, I'd need an average swing of about 20,000 for that to be possible. If you assume the tick was munched instead, the damage gained is only 58068-55020 = 3048, which is reasonable.

Tick munching is pretty much entirely RNG and uncontrollable, but it does increase the value of crit rating fairly significantly(as explained above, more crits = higher average tick sizes + more tick/crit collisions) and can have a huge impact on things like DPS records for a fight since random streaks of munches can add a ridiculous amount of DW damage.

In Cata, the OH miss bug will be gone when on next swing attacks are eliminated. It can't occur without them. The DW bug will still occur since it's caused by whatever it is that's causing ignite munching, which has been broken for years and will likely never see a fix. It will, however, be much less prevalent if crit rates are significantly cut and not allowed to reach the same ridiculous levels that the end of this expansion is seeing.

TL;DR: Queuing HS a lot makes your OH not miss. Lots of crit makes DW do broken things. Hit rating past the soft cap is beyond terrible, the crit cap is something you should stop worrying about, and crit rating is better than you thought. Don't gem haste.

re: OH miss rate

Ouch, this is a pretty serious bug. Just glancing through my ICC logs I can confirm to have only 1 or 0 misses for encounters. This is with 8 or 9% hit (either with or without heroic presence). Without exactly calculating the magnitude of the bug, I'd reckon its implications to be really big. Even moreso as this has a loop effect (no OH misses -> enough rage -> HS/Cleave queuing -> repeat).

Since when is this bug active? This is a big one.

I can only say for certain that it's been that way the entire expansion, but I'm fairly sure it's been around since release. I remember noticing my misses being somewhat lower than expected in raids back in BC, but auto attacking target dummy tests always showed expected miss rates, so I just chalked it up to RNG luck. It wasn't until Ulduar that I really dug through logs because miss rates were too low to possibly be just random chance. I noticed a major discrepancy between my Vezax misses and other fights, and realized the only real difference in the fight was the melee slow aura making rage too choppy to HS consistently. Connecting the dots, it makes a weird sort of sense that turning your MH attack into a special means your OH is the only thing auto-attacking.

A few tests later, I had 100 straight OH swings without a miss by desyncing the swing timers as mentioned above and re-queuing HS before each OH landed to confirm it. I also tried queuing HS after each OH swing to determine if it was just the last MH swing being an HS that caused it as well as queuing HS when swing times were synced. The former did not reduce misses at all, and the latter had sporadic results that randomly shifted between normal miss levels, missless streaks, and reduced miss levels, but usually just averaged out to about a 50% reduction.

Regarding the post on OH miss reduction due to HS queuing,

Current gear 10.84% hit - 14387dps (2x Heroic Cryptmaker)
Alt Set 8.12% hit - 14302dps (Heroic Cryptmaker + 271 Glorenzelg)

Assuming the spreadsheet is overvaluing hit past soft cap would this de-value using heroic Cryptmaker as offhand enough to Allow Glorenzelg to surpass it. I am 93 over hit cap with my Duel Cryptmaker set if I remove 50 hit rating assuming that the extra hit is worth even 50% of what the spreadsheet values it at Glorenzelg surpasses it, hoping Landsoul can implement this into his next update so we can properly see how it will affect gearing.

You know I have always noticed that I don't miss nearly as much as I should, but looking at recent logs its pretty astounding how low the miss rate has become

Melee strike breakdown our last ICC clear;

Crit 45.8%
Hit 24.2%
Glancing 21.7%
Miss 4.7%
Block 1.1%
Absorb 1.8%
Parry 0.4%

Those numbers are without H cryptmaker when I was sitting just above 8% hit. Obviously no where near the miss rate one would expect.

Wow.

The second one is interesting, although I'd actually known about it for a while; I hadn't considered the implications it had for crit gemming past the softcap, though. The first one is huge, and it goes a long way to explaining why the damage has gone up so much ever since Heroic Strike has pretty much been held down for the entire fight. I'm going to switch my gemming over from STR/Haste to STR/Crit and see how it goes, and I'm considering swapping Mongoose in as well.

@Nomepunter

I believe Aldriana recently found a stealth nerf to Mongoose. Read up on the Rogue forums for more info. Just a heads up.

This is interesting. How long has this bug been around? I can put an option into the next SS build for "use HS + 0 off hand miss bug."

Unless something has changed/I am mistaken. Whirlwind is normalised to weapon speed, where as slam is not. meaning slam would gain a lot more from a slow weapon than whirlwind. Back in naxx, The Jawbone was doing more slam damage than Betrayer of Humanity, being an 3.6speed ilvl 213 weapon, compared to the 3.4speed ilvl 226 weapon, making Jawbone the main hand choice for some.

Forgive me if this has changed, but the formula used to be "base_weapon_damage + weapon_speed * Attack Power / 14"

With normalised weapon speed being 3.30 for 2handers, this counts for whirlwind, where as for slam you would use the actual weapon speed in the equation. In the case of Shadowmourne/Shadow's Edge, that would make a big difference of benefit from weapon speed, between whirlwind and slam.

I noticed the HS/Cleave bug with the OH miss recently as well (I found it very strange that I had 3 misses on saurfang while a fellow warrior had 33). He was rage starve bugged some how, thus he wasn't able to queue HS the entire fight.

How much DPS does this account for? The white swings itself seems about 2% damage, but I don't know the other ramifications (DW, etc).

But yeah until it's fixed, any hit past 5% is terrible.

These Deep Wounds bugs certainly have some implications on the value of crit. At the very least, it will once again put crit rating over haste rating at high gear levels where warriors were starting to gem 10str/10haste over 10str/10crit. It will also undoubtedly put Heroic Deathwhisper Fanged Skull over Heroic Death's Verdict/Choice in the BiS setup.

Even more interesting, though, is the implications on the value of Recklessness. Currently, it is a generally marginal cooldown as our raid buffed crit chances approach 60%. With this bug in play, though, it could GREATLY increase the value as a warrior could coordinate a critical Heroic Strike with a critical GCD ability.

I am not well-versed in the math backing up this bug, but it seems that an increase in crit chance from crit rating would exponentially increase the chance of munching. E.g., a Heroic Strike and a Bloodthirst land at the same time, and each has a 45% crit chance, so the chance of a simultaneous crit is .45*.45 = .2025. When the crit chance increases to 50%, the chance for a double crit is .5*.5 = .25. Am I looking at this correctly? If it is true, the implications could be very great.

I am interested to see this put into effect in the spreadsheet to see the effect on crit rating.

The fuss is because fixing this bug would have a compounding impact on Fury DPS. Most Fury Warriors are riding the 8% hit line, and as such are expected to miss offhand swings ~19% of the time. Average actual miss rates seem to be 3-6% because of the bug. As such, if Blizzard fixes the bug an additional 13-16% of our OH swings will miss, lowering DPS directly. There will also be a corresponding indirect decrease from lost rage generation. Roughly speaking if 15% of your damage comes from OH melee hits, you can probably expect a 2.5-5% DPS decrease if the bug is fixed.

Blizzard may not bother due to how late in the expansion this is, and the impending removal of next swing mechanics. It's not impossible though, and a nerf to Fury could potentially make Arms more viable or even clearly superior.

I think you're confusing crit munching(simultaneous crits) with tick munching(a crit immediately following a DW tick). Simultaneous crits are a DPS loss, since you lose the additional damage added to the DW buffer from the first crit when the 2nd crit overwrites it. Tick munching is a DPS gain since you gain the damage of the DW tick while the crit following it prevents the tick's damage from being lost from the buffer, effectively making the tick free damage.

An interesting consideration about this bug, which I have not yet seen anyone mention, is the apparent ambiguity of its net effect on our DPS. As is pointed out, this bug presents itself in two ways; which for ease of differentiation I will call munching and growing. Deep Wounds munching represents the loss of Deep Wounds damage when two critical strikes occur in close proximity to each other, while Deep Wounds growing representing the damage gain that takes place when a critical strike lands immediately after a Deep Wounds tick.

The frequencies of occurrence for both munching and growing have a positive correlation with critical strike chance; however, their rates of change are not likely to be the same relative to each other. Instead, their relative rates of change will be dependent upon the frequency of opportunity for each to occur. If we call 'T' the time frame within which the two events must occur in order to cause munching or growing, then the two numbers we would need to find and compare would be:
M = Number of times per minute in which two critical strikes land within T of each other
and
G = Number of times per minute in which a critical strike lands no more than T after a Deep Wounds tick

The ratio of these results would then have to be further modified by the ratio of their average impact on the Deep Wounds pool. Using simple numbers as an example, if the average instance of munching was a loss of 1000 damage from the deep wounds pool, and the average instance of growing was an addition of 2000 damage to the pool, then our break-even point, at which the entire bug has no net impact on our DPS, would occur when M = (1/2)G. In this scenario, M < (1/2)G represents a net DPS loss from the bug, and M > (1/2)G a net gain.

Obviously, the actual numbers are not going to work out quite as cleanly, and may be influenced by a number of other factors, such as gearing and buffs. For example, if you are syncing two weapons of the same speed it should increase the rate at which munching scales relative to growing, by increasing the frequency of opportunities for munching to occur. An absolute solution to the system is probably an unrealistic goal, but through testing we should at least be able to get a feel for which direction it tends to go at current gear levels.

Rallik, I am curious if while you were testing the growing bug you took down data regarding the impact of munching. It may very well be the case that you already have the answer I am looking for. If so I would like to know what you found out, and if not then I may do some testing to try and get some basic numbers.

The only thing I hadn't tested was AoE fragment generation, so I fixed that. I used the Theramore target dummies to have the max possible number of hits With cleave/whirlwind.
World of Logs - Real Time Raid Analysis
1946 Fragments
951.0 seconds_______
2.0462 Fragments per second

My WW usage efficiency was only ~80% and I missed a few cleaves so I suppose that could skew the data some, though considering how fragment generation has been working on a single target I would think that the above data is pretty accurate. Given more testing I would expect to see the fragments/second to average out to exactly 2. Also when I use the above data in some old math I did to determine whether or not a cancel aura macro would be viable or not, no cancel aura macro still pulls ahead by about 40 DPS.

Offhand swings that land while HS or Cleave is queued do not suffer the DW miss penalty

So does it mean using dual H Cryptmaker becomes even worse than using H Cryptmaker in MH and Glorenzleg (or even Shadow's Edge in OH)? Because dual H C + rep ring gives you 99 x 2 + 59 = 257 hit rating (way over soft hit cap then the redundant hit rates may be worthless), weapon dam from dual H cryptmaker can be compensated by stats (crit rating, expertise) which are given by others weapon mentioned above.

I am confused about using dual H Cryptmaker after this bug was noticed, Landsoul's SS hasn't recorded this bug as dps effect as well.