enno

 

Mojofabulous

Thanks, read through it. I guess I'm not sure how that affects us. If this has always been the face as the thread seems to suggest then why should it affect our dps at all. Or is it only affecting our theoretical spreadsheet dps based on an incorrect assumption(given that we were using haste rating items on that spreadsheet)?

 

• Aldriana

It's not going to affect real DPS, as no mechanics change is actually being made. However, the spreadsheet models are going to need to be updated, which will change the relative value of certain stats and abilities.

Of note, the valuations in this spreadsheet will be affected in the following ways:

1)The power of all PPM effects will go down somewhat, including, notably: Dragonspine Trophy and Mongoose enchant (along with Madness of the Betrayer, BS maces, The Bladefist, Rod of the Sun King, etc.)

2) The value of haste rating will drop somewhat. Currently, the model is that haste rating increases both direct damage done, and also increases proc time which will increase damage further. With this change, haste rating will still increase the direct damage done but will *decrease* proc uptime, which will yield a smaller damage return.

In terms of actual in-game effects: I think the only difference is that if you're not using any flat +AP activated trinkets, it is now understood to be optimal to use AR separately from activated haste buffs (Blade Flurry, Heroism, Haste Pots, etc.) rather than at the same time. It's not going to make a *ton* of difference, but it'll buy you a little extra proc uptime.

 

• Vulajin

I don't think you can say that haste rating will decrease proc uptime. I've been following that thread pretty closely myself, and I haven't seen any solid testing done on special attacks yet to indicate that they might suffer the same proc rate adjustment that white attacks do during haste effects. Hell, I'm not even 100% convinced yet that proc rates adjust with haste effects. There are still conflicting results and it's a bit early to conclude anything.

As far as how much it would affect our models if this turned out to be true -- well, in the case of DST, not actually very much. Using a model I cooked up in a spreadsheet of my own with my gear as the basis, DST uptime comes out to 27.85% if hastes increase PPM, and 23.43% if hastes don't increase PPM. Note that this assumes the 20-second internal cooldown from the upcoming patch.

The first number is calculated using the same method as your spreadsheet, by figuring a number of procs per second from the total number of attacks per second, then taking (Duration*PPS)/(Cooldown*PPS + 1). The second number was calculated by modifying the procs-per-second formula:

((1 - MHMiss - MHDodge) + (1 - OHMiss - OHDodge) + (1 - SPMiss - SPDodge) * MHSpeed / SPSpeed) / 60

This assumes DST is 1 PPM, regardless of the speed of your attacks. Thus, you get 1 PPM from your main hand times the fraction of MH attacks that actually land, plus the same from your offhand. Finally, from special attacks that land, you get 1 PPM times the ratio of main hand [base] speed to special attack speed (for this you need to know approximately how often, in seconds, you perform specials). Then divide by 60 to get procs-per-second.

---

If it weren't for the internal cooldown, the effect would be larger, I suppose. I repeated the same calculation for Mongoose (in the middle of typing up this post) and the main hand's Mongoose uptime was reduced from 46% to 25%. The loss of agility is approximately 25.

(edit) Oops, screwed up that Mongoose calc. It actually went down to 35%, with a drop of about 13 agility. Still rough when combined with the effect on the OH. Still, if that thread is to be believed, it's not like these are nerfs going in...we've just been modeling it wrong all along.

(edit 2) Perhaps I've royally screwed something up, but it really doesn't look like the effect is that big. After completely redoing the calculations in my spreadsheet, my DPS decreased from 1415.02 to 1397.02 (a difference of 18).

 

• Aldriana

It is too early to say for certain that it will decrease the proc rate, as it's uncertain what it will do to instant attacks. However, I would be pretty darn surorised if we don't ultimately find that haste does, in fact, reduce the chance of your instant attacks to proc, given the rest of the data in that thread. I could be wrong, but that would be my expectation. Guess we need someone to do some no-autoattack, instant-attack-only testing with and without haste rating.

As for the rest: sure, maybe Dragonspine only drops from 28% to 24%. That's still a 15% reduction in the power of the proc, which, coupled with the haste nerf and the internal cooldown, will bring it much closer to in line with existing trinkets. Which, I agree, is not a bad thing, necessarily - it's silly when you have BT rogues still needing to farm Gruul to gear up. But it is worth knowing that the power of these procs is lower than the current model predicts, since that is going to have effects on both the value of haste rating and hit rating.

 

dmw

Unfurtunately I have no WWS log. I just use Recount for my own, no combatlog which I can uplaod.

With non-dagger spec the damage is ok, I hit there #5, even w/o WF and BS. I was just a bit confused about my low damage I do with daggers... if I'm dagger again sometimes I will post again if my damage sucks again then, but for sure I'm mace/sword the next time. =)

 

baldycenarius

The only thing that conflicts in the haste and PPM thread is my fiery test, it could simply mean fiery is based on a set %procrate rather then PPM. I dont have the time to get on the PTR and spend 2 hours beating on mobs again to see if thats changed, as it seems plenty of other tests would suggest other results anyway.

 

• Vulajin

I, on the other hand, feel like it would make more sense that the proc rate on instant attacks weren't modified by haste procs. However, that's what experimentation is for.

In the meantime, what I can do is to compare the two models and see what the difference really is. So I've been heavily refining my spreadsheet over the course of the day to bring it in line with the theory that haste does not change the effective PPM of a PPM-based proc. After all my modifications, I equip a Dragonspine Trophy in place of my Bloodlust Brooch (in reality, I don't have a DST, but this is just for the sake of argument). My DPS is given as 1362.45.

Now, I modify the spreadsheet to assume that special attacks suffer a penalty to their proc rate on PPM enchants while under the influence of haste effects. I check my DPS again. It is down to 1357.14.

For the sake of argument, I now equip a Shadow-Walker's Cord in place of Socrethar's Girdle. My DPS now increases to 1375.64 (if special attacks aren't penalized, it's 1381.06). Then I swap in a Belt of One-Hundred Deaths and it increases to 1376.83 (1381.64). Belt of Deep Shadow makes it 1376.42 (1381.23).

In the end, I don't think there's actually that much of a difference whether you model it with special attacks getting penalized or not. Obviously we want to be as accurate as possible with our models, so it will be nice to know for sure, but all in all the total error is not that major.

 

• Aldriana

I agree that it doesn't make a whole heck of a lot of difference in terms of DPS; however, even small net changes in DPS can have significant effects on stat weighting. Consider, for example, the discussion I give of inaccuracies of the original rogue spreadsheet in post 619. The basis of my argument was that in practice one had 210 more available energy than the sheet was modeling for over the course of a 7 minute fight.

Now, in terms of actual DPS, 210 energy is 3.5 backstabs, so maybe 4k total damage. Which, over a 7 minute fight, is under 10 DPS - barely over 1%, and not a big deal. So in terms of estimating your actual damage output, it was fine.

However, consider the effect on the value of, say, a talent point in lethality under this change. Since Lethality only affects your yellow damage, the amount of damage you get from a point in Lethality increases by 5% when you move from the spreadsheet model to the real scenario. Meanwhile, a talent point in, say, precision, still yields exactly the same amount of damage either way, while the value of a talent point in, say, murder, is going to increase by about 2%. So even though ones overall damage changed by barely 1%, the relative value of stats can easily change by up to 5% in this example. Now, admittedly, 5% still isn't a *lot*, but when you consider that, for instance, the value of agi and hit will be within 10% of each other after the Dragonspine nerf, suddenly changes on the order of 5% look like they could really change what your optimal gear is, even if they don't directly affect your damage that much.

I admit this is to some extent splitting hairs, but I do think it's important to note that even reasonably small differences in damage can signify larger changes in equipment selection.

 

Mojofabulous

Doesn't it matter very little? Even if instants aren't affected by the normalization...doesn't that only significantly change items like say.....warp-spring coil(even then it shouldn't)? If an item has an internal cooldown and a PPM the few extra instant attacks in the midst of all the auto attacks shouldn't hold that large a part of whether a PPM procs faster or slower. I guess maybe something like a heroic strike could be affected...but a sinister strike has nothing to do with haste rating other than combat potency procs.

If I'm flat wrong here please do tell!

 

• Aldriana

Well, it's not a huge difference, but again, it matters some. Lets take a look at MH Mongoose as an example.

Lets assume we're using Talon of Azshara, and are gaining enough energy for a Sinister Strike every 3 seconds (in practice, it takes a bit longer than that, but, as a ballpark figure, it'll work). Based on the way the normalization works, regardless of haste level, the autoattacks will generate 1 proc per minute. Now, if instant attacks are *not* affected by Haste, we will have 20 SS per minute, each of which has a 2.7/60 = 4.5% chance to proc, so we gain an extra .9 procs per minute from instant attacks, for a total of 1.9 PPM.

If, on the other hand, we have, say, 35% haste (which is trivial to get to, particularly with 2/5 T6), our Talon is now effectively speed 2.0, so the proc chance goes down to 2.0/6.0 = 3.33% chance; hence the 20 sinister strikes per minute only generate an additional .667 procs, for a total of 1.667 main hand mongoose procs.

So, it's a difference of about 14% in the number of MH procs, which equates to an uptime difference of about 11%. Hence, there's an 11% difference (or so) in the damage contribution of Mongoose - that strikes me as relatively significant.

 

Mojofabulous

Ok, I was thinking about it terms of how many extra sinister strikes you would get via haste rating and whether those could possibly proc mongoose. Clearly improper logic! You math people complete me, lets get hitched :-)

 

• Vulajin

Not that I disagree with you at all, since splitting hairs is by and large what we do as theorycrafters. My point is mostly that, as long as we don't know whether special attacks get penalized or not, we're still not going to be too far off regardless of which way we model it.

 

• Aldriana

As a quick update: I hacked up some estimates of assuming

a) Haste Nerf
b) Dragonspine Nerf
c) New findings on Haste/PPM.

Findings are pretty interesting; for one, agility does indeed pass hit in terms of total utility per point (at least, by weighted total if not by actual damage output - hit still has the damage edge).

Additionally, the ranking of the endgame trinkets is as follows:

1) Dragonspine - 209
2) Ashtongue - 173
3) WSC - 172
4) Madness - 167
5) Tsunami - 157

So Dragonspine is still the best. It's just not quite as mandatory as it once was.

 

Mojofabulous

alright I finally got my character copied over. This is about 55 mins worth of combatlog. Used bt trash mace 1.5 speed and arena sword offhand 1.5speed. No instants or finishers. I used blind I think 3 times, twice near the end because I ran out of potions. Mongoose on both weapons

Free file hosting by Savefile.com

this is of course to test out internal cooldown and PPM of tsunami talisman on the PTR

 

deluXE

If only I could get one...

 

antonmb

Will you have this updated on the latest spredsheet?

 

• Aldriana

Yes. I'm not going to bother releasing a new version until the patch gets off test though; while they're still changing things I don't feel like trying to keep stuff updated.

 

• Aldriana

Analyis of Tsunami Talisman data:

Shortest time between procs: 45.187 seconds, occurring twice.
Number of crits that occur at least 45 seconds after last proc: 443.
Total procs: 45
Proc rate thus lies between 7.3 and 13.0 percent.

Now, what I have in the spreadsheet is 10% proc rate, 40 second internal cooldown. From this data, it looks like more than a 45 sec internal cooldown. So either this data set is flukishly devoid of procs between 40 and 45 seconds (as in, there were 92 crits between 40 and 45 seconds after last proc, and no procs), or it's been nerfed slightly, or I have the cooldown a bit wrong on live. I'd say the third is most likely, but if people want to do additional testing I'll certainly crunch the data.

 

Mojofabulous

I'll do another long test tomorrow or next day when I have time. The idea of tsunami talisman being WORSE is pretty damn disappointing.

 

Kytrarewn

I know you're not thrilled with the modelling of +skill in the spreadsheet, but how certain are you about the DPS value of [Gloves of the Searing Grip]? Given that we're on the cusp of an Al'ar kill, and we haven't yet killed Leo with huge demand for T5 gloves, and the uselessness of the T5 2pc bonus, I'm wondering whether it would be worth it for a rogue who has weapon expertise anyway (we have no mutilate rogues as active raiders).

ΜΟΛΩΝ ΛΑΒΕ


Jesus don't want me in a sunbeam
Sunbeams are always made on me
Don't expect me to cry, for all the reasons I'm gonna die
Don't ever ask your kick of me.

 

Mojofabulous

from what I know weapon skill works just fine in the spreadsheet as long as you have at least 355 weapon skill. So unless you dont have weapon expertise, it should be giving you a pretty close assessment of those glove's value.

 

Kytrarewn

Mojo: There was some discussion on the weapon skill adjustment thing about the actual value of weapon skill, whether or not it scaled linearly, etc. Should I just take it at face value and say "It will be at least the DPS noted in the sheet" and deal with it?

ΜΟΛΩΝ ΛΑΒΕ


Jesus don't want me in a sunbeam
Sunbeams are always made on me
Don't expect me to cry, for all the reasons I'm gonna die
Don't ever ask your kick of me.

 

Karmon

The effekt of +skill on hit/miss is nailed down.
+skill lowers the basemisschance for mobs. In a sense you can say it will provde hit because those are complementary.
If you figet a boss (or lvl73 mob):
- It is linear for skills from 350 to 354 and from 355 on, but with different slopes
- There is a pronounced jump when going from 354 skill to 355 skill.
(The change of slope happens if the difference of Mobdefense and your weapon skill is >10.
So +skill is not a big deal if you only fight mobs between your level and up to two levels higher.
But as soon as you hit bosses you get a 3% decrease in miss if you have +5skill or more)

And of the converted skillrating only whole numbers are used:
INT( (rating1+rating2+rating3+....) * 52/203 )
Eg. +15 skillrating nets you ony +3 skill instead of the converted 3.8.

For the details look at http://elitistjerks.com/472228-post533.html and the whole thread there.

For total weapon skill of 355 or more the spreadsheet does provide the correct answer,
but expressd as gaining +hit instead of lowering basemiss. Does not matter in the end.

 

antonmb

Think we can have a feature on the next version that can show AP/crit as well, aside from just hit rating?