Ounce

I have read this thread and the others for a while now, and I have a quick question for Aldriana that I have never seen come up in the thread:

Why do you have on-use effects (BF, Bloodlust Brooch, etc.) modeled in terms of their average impact? When raiding, I always stacked together my 2-minute cooldowns to get full benefit. Pretty invariably, BF, haste potion, drums, and an on-use trinket (when I used one) were always up together. It seemed that the "average" approach undervalues each of these things--most importantly it would undervalue the on-use trinkets, since the primary purpose for this sheet for me was when I needed to choose between two items.

I apologize if I misunderstood how you have these modeled.

 

◊ Vulajin

Conveniently, I have a personal spreadsheet which models on-use trinkets in exactly such a fashion, and I found that their effect was not greatly increased relative to simply taking the time-average of the on-use effect. Sadly, the spreadsheet is not currently functional, or I would replicate some of the comparisons for you; however, if you're willing to take my word for it, then the answer is simply that it doesn't make a very big deal, therefore it saves a lot of computation time to model them as average effects.

 

• Aldriana

A couple of reasons:

1) It doesn't help as much as you might think. I mean, don't get me wrong, it will get more damage out the trinket, but we're talking at the level of a couple of percent. It's not going to, for instance, double the value of the trinket.

2) It's extremely hard to calculate in the general sense, particularly in spreadsheet format. I mean, Bloodlust Brooch isn't that bad, since it's cooldown aligns with Blade Flurry and it's pretty easy to see how to handle that interaction... but what about trinkets with cooldowns of 3 minutes? Or 1.5 minutes? How do you handle the notion of waiting till a proc is up to use them? There's just a lot of variables that apply that make it hard to model a general activated ability.

So, basically: it would be a lot of work, and I doubt it would change the ordering of trinkets much anyway. And, frankly, it's not even the biggest inaccuracy in the sheet, so I'm just not inclined to worry about it. If I were going to do major refactoring to more accurately reflect reality I'd start by giving fights a finite duration long before I worried about activated trinkets.

...which is not to say I won't end up doing it eventually anyway. It'll just be a while, and I don't think it's a major concern in the meantime.

 

Ounce

Everything that both of you said makes sense. I guess in my quick estimation I far over-valued the benefit of stringing these effects together. I think many times (such as when we were first attempting Morogrim) I instructed our rogues to hold off using effects until it was a good time for blade flurry--no watery grave, no murlocs. We probably would have been better off using all of the effects except for BF, then using BF on its own when it made sense--and achieved a lot greater overall up-time.

I guess that's always the most important part -- taking these "theoretical best" rotations, gear choices, etc. and adapting them to real-time encounters. It helps to know the value of each component when you are making such decisions. The information was always available to me, I should have looked at the math more closely myself. In terms of the on-use trinkets, heroism, haste potions, etc. I guess I will just assume going ahead that they are slightly stronger than they are listed in the spreadsheet for the fights where you can easily time the uses together.

 

• Aldriana

I think the general rule of thumb for combining activated abilities is that it's a good idea insofar as it doesn't reduce the number of uses you get during a fight.

That is to say: if you have a 2-minute cooldown and a 5 minute cooldown that synergize each other, if the fight lasts 6 minutes you can get a maximum of 3 of the first and 2 of the second; hence, you can safely do the 5-minute at .5 minutes in and 5.5 minutes in, and the the two minute at the same times, plus once as is convenient between the 2.5 and 3.5 minute marks. But if the fight were, say, 7 minutes, waiting in this form would cost you the 4th usage of your 2-min cooldown, and, as such, you should just use it at .5, 2.5, 4.5, and 6.5.

That is: the benefit of using a cooldown n+1 times is generally larger than doing it n times, timed optimally against other effects. The notable exception to this is Blade Flurry, if and only if you're waiting so as to be able to use it on multiple targets.

Thus, particularly if you're unsure of the fight duration (i.e., if it can take you guild anywhere between 10 and 12 minutes), it's generally safer to employ the greedy algorithm of cooldown usage.

Note, however, that this analysis has been laced with words like "usually" and "generally". This is one of those fine points of DPS that it's somewhat hard to come up with a hard-and-fast rule for. Some general guidelines (such as the ones in this post) can be posited, but in the end there's no substitute for understanding the tradeoffs and making decisions as you go.

 

TrickBlades

Ald, do you have an ETA on the updated spread sheet yet? I know you were still working on getting a few kinks worked out. Just curious.

 

Crimsonsky

I have a curious question for anyone who might be able to make an educated guess. I picked up the pattern for the Carapiece of Sun and Shadow today and when I plug it into the spreadsheet my recommended cycle goes from 3.1s/5r to 5s/5r, is there a obvious reason for this that I'm just not seeing? My hit stays the same, since I'm at the 316 cap(we have a moonkin with imp FF) so all I gain is ap and haste, but it doesn't make much sense that gaining the 30 haste or so would switch my cycle so dramatically. Anyone know why this would be? For what it's worth the Rogue DPS spreadsheet still recommends 4s/5r snd cut when changing from Slayer's to Carapiece.

 

Arindelest

As has been discussed very recently as regards all these cycles, 4s/5r for any high-end gear will always be within a couple DPS of 5s/5r, because more Sinister Strikes become a good tradeoff for less Rupture uptime as your gear improves. Rupture does not scale as well as SS with AP, so in all likelihood you were very close to the point where it becomes better to spend all your energy on SS and none on finishers, rather than spending a little bit extra energy to increase your rupture uptime. When you got the extra AP from the chest, it pushed you over the point.

 

nelalas

I think Aldriana is currently waiting on updates to the sheet from me before releasing the next version; I should be finished tonight or tomorrow.

 

Crimsonsky

Makes sense, thanks.

 

Nock

Would it make sense, as a way to (hopefully) cut down on the number of "Why has my cycle changed from Xs/5r to 5s/5r", to just make a grid on the Talents/Equipment page that shows the 6 different cycles on the DamageCalcs Sheet? I imagine just having a grid just showing the titles from A14-A19 and the data from G14-G19 and B14-B19 would do wonders to showing people that the difference between 3s/5r and 5s/5r is quite small, and that's why a gear change can bounce it one way or the other.

 

nelalas

There are certainly changes to gear that will nudge cycles in either shorter or longer directions, especially when coupled with changing buffs, debuffs, and boss characteristics (armor amount, interrupted combat time, etc.). A lot of these things are considered in the spreadsheet but I think that it is important to remember that energy use cycles contain a bit of stochasticity. The new version of the spreadsheet that introduces cycle slackness begins to touch upon this issue and, to me, the important question isn't "Zomg what super-tight cycle will give me the most 1337 DPS that I can only employ 10% of the time?!!?!" but rather "Which is the highest-DPS cycle length that is reasonably sustainable for the fight at hand?" Perhaps in other words, the theoretical cycle that is suggested on the spreadsheet may or may not reflect what is actually best in a given situation.

 

• Aldriana

So, thanks to some extensive testing by Capek, we have some updated information regarding the proc rate and damage of the SSO Neck. The major test, which I'll be drawing most of my conclusions from, can be seen in WWS here and the raw log downloaded here.

Observation 1: This log contains hits of every size from 333 to 367, and crits from 666 to 734. As such, I feel reasonably comfortable saying that the damage range is 333 to 367, and crits (barring RED and the like) do double damage.

Note 1a: There is additional testing performed at a range of different APs that confirm that the damage doesn't depend on AP, as some have conjectured. I think the oversized crits that people have been seeing in raids are due to the combination of one or more of RED, Murder, Misery, Curse of Shadows, Imp Sanctity Aura, and Ferocious Inspiration.

Observation 2: Procs are critting about the same rate as melee attacks. I didn't do any real statistics here, but it's clear that they're not critting at the spell crit rate (which would be 0), and it's equally clear that it's about the melee crit rate.

Note 2a: There's additional testing that shows that bumping melee crit rate up to 20% increases the crit rate of procs to ~20% as well.

Based on this, I'm fine with saying that the conjectured behavior - it crits like a melee attack - is what's actually going on. All this leaves to determine is the proc rate. And on that topic:

Out of 22477 total attacks, 3822 appeared 45 seconds or more after the last proc - this would be the total count of eligible attacks. These 3822 attacks procced a total of 666 times; thus, the average proc rate works out to be 17.4%. Grinding a few statistics, we find that the proc rate is between 16.2% and 18.6%.

Now, the thing to note is that there is some random variation in this - some procs occurred as little as 44.6 seconds apart. This general noise in combat logs is pretty common, so, realistically, there's probably a few other attacks eligible to proc it. So, one could argue that any attack more than, say, 44 seconds after the last proc is eligible to proc. Counting these, we get a proc rate between 15.7% and 18.1%.

Based on this, we can say with a reasonably degree of confidence that the proc rate is *not* 15%. My guess would be that the proc rate is 1/6 = 16.7%, but I am just guessing at that.

So, based on this information, I'm going to bump the proc rate up to 16.7% in the next spreadsheet release. And while I can't be sure that's exactly correct, I'm comfortable that it's close enough for our purposes - the difference between a 16% proc rate and a 17% proc rate on a proc with a 45 sec cooldown will make very little difference to the actual proc uptime.

 

◊ Vulajin

Did you happen to do testing with different weapon speeds to be certain that it's %-based and not PPM?

(edit) to avoid making a second single-line post in a row. Sounds good to me, thanks for grinding the numbers.

 

• Aldriana

Not in this latest round of testing, but the PTR testing that koaschten did struck me as sufficiently conclusive that I'm comfortable with saying that it's percent based. I'd have to dig up the posts to get exact numbers, but it was something to the effect that two weapons 50% different in speed had proc rates less than 10% different, which basically kills the idea of PPM.

 

• Aldriana

Version 0.10.1 is up and running.

Chanegs:
* Added option for twisted GoA.
* Added buff scrolls + thistle tea.
* Added option for cycles with adjustable slack.
* Fixed bug with SSO Neck on 5sXr3s sheet.
* Updated SSO Neck proc rate.
* Assorted other bug fixes throughout the sheet (with a nod of thanks to Latito and Havenwood for pointing some of them out).


Note: If you download this sheet, and something isn't working the way you expect, please read the FAQ section in the first post before posting a question about it. Many of the questions that get asked in this thread are already answered below.

Version 0.10.1, 4/28/08: Download

 

drumbum

Since you're on the topic, I thought I'd point out something that is listed in this thread: WoW Forums -> 2.4 Known Issues list.

Keep in mind this is in the middle of a list of known bugs, so this implies that it will no longer be dodgeable in a future patch. I'm not sure if you're treating it as dodgeable now, but I suppose it's not going to skew the results too significantly either way, especially if you have some expertise.

The patch notes for 2.4.2 do not mention this however so I have no idea if it's actually fixed in the upcoming patch or not.

 

nelalas

Just a quick note with regard to Twisted Grace of Air in the spreadsheet. You'll have to first enable the base Grace of Air buff and then, additionally, enable the Twisted option. Checking Twisted Grace of Air only will not enable the buff (similar to Solarian's Sapphire under Battle Shout).

 

• Aldriana

And as an additional quick note: what the description says notwithstanding, scrolls do stack with each other. It's just a text bug - the sheet is computing everything correctly - the description just got entered incorrectly.

 

TrickBlades

I am seriously debating this. As of right now I have [Shoulderpads of the Stranger], [Belt of One-Hundred Deaths], and [Shard of Contempt]. Putting me way over the expertise cap (which by the way I have not been able to find by googling, everyone is still very unsure about it from what I've seen. So if anyone has a hard number I'd love to know it.) at 30 expertise with 2/2 Weapon Expertise says the spreadsheet. Perhaps putting some extra points somewhere else would be helpful.

Any suggestions?

 

drumbum

It's been determined independently by multiple class threads that the expertise cap is 26 (6.5%) for dodge. Both spreadsheets take this into account so you can use the spreadsheets to find out if going barely over the expertise cap is worth it over an alternative piece without expertise.

 

• Aldriana

Speccing out of WEx is usually a bad idea, purely because there's really no alternative DPS talents to take. You wind up getting random filler with that point instead, which doesn't really gain you much; as such, if you have any reasonable alternative for any of your expertise pieces, it's generally better to replace them than it is to respec.

 

Caelius

First time poster, long time reader.


Just getting into the rogue 70 world and new to the whole spread sheet. I'm looking at this in excel and i can't figure out how to get the lower classed epics to come up even blues and such. What am i missing?

Thanks.

 

saedo

Those are probably items that were deemed not worth getting cause there are better alternatives, which are on the list.

 

• Aldriana

So, the thing to keep in mind is this: spreadsheets such as this are designed to model endless combat. In practice, of course, no fight is endless; but to get to fights where this is a reasonable approximation, they needs to be at least a minute or two long... which one doesn't tend to see before Karazhan. As such, my assumption when compiling the gear list is that people are, at the least, starting Karazhan, so the gear list in the sheet starts with the better level 70 blues and works up from there. Thus, if you're just starting out, a lot of your gear probably isn't in the sheet - and your first priority should probably be to get stuff that is, as it's almost certainly better than what you're using. There's usually at least one item for each slot that can be gotten pre-Karazhan.