And iirc not having the major armor debuff increases armor penetration's relative value...
If the mobs debuffed armor is < 7616 (trash mobs, lower level mobs, heroic throw applied to boss), the damage increase gained by a given amount of ArP is dimished the lower the debuffed armor is. So your statement is true in this case.
But in the (standard) case of the boss debuffed armor being > 7616 the dmg increase of ArP remains constant for a given ArP value.
Some milestone numbers to keep in mind for this standard case (seeing as FF + sunder/expose do not reduce the boss armor below 7616) are:
In burn phases having SS + FF + Heroic throw active as boss debuffs (reducing boss armor from 10643 to 6471) the Dmg increase of 100% ArP is reduced to 42,5%.
Obviously having more armor debuffs on the boss results always in higher absolute dmg done. So Shattering Throw et al are never detrimental to your damage, no matter how high your individual ArP might be.
Last edited by suicuique : 06/03/09 at 6:16 AM.
Reason: Clarification
I've gotta question regarding trinkets. (I haven;t been able to successfully compile the program yet so I don't know if this is possible or not)
I have a wrathstone currently and its actually pretty damn awesome. Though the way I currently use it is to tie it to berserk, so it actually only gets used every 3 minutes. My question is, is this better over say an x minute fight (4, 5, 10, etc)
Also, does this increase or decrease the value of the trinket compared to the average values that I believe are used currently.
@Lolaan: It depends on fight lenght. During berserk the value is about double, so it depends on how much do you lose. For istance in my simulation (the one to make trinket ranking) (5 min fights) I tied wrathstone to the first berserk (about time 00:05) then used it at time 02:05 and time 04:05 suddendly after a tiger's fury to start with 100 energy. Waiting for the second berserk in a 5 min fight means that I can use it only 2 times.
@Cal, I loaded up Rawr, threw on every piece of ArP gear I could find, (keeping t8 helm/gloves for the ArP they offer and the bonus though I do not know at what % ArP the actual stat outweighs the bonuses) and was only able to attain 918 ARP, fully gemmed with Blacksmithing and JC. Someone else please feel free to check this, but at the moment I don't think the 85% mark would be attainable currently.
My Rawr tops out at 999 ArP, including the food (40 ArP) and elixir (45 ArP).
Nightcrowler, I've revisited your computations since my raid roll might be changing to cat main.
I don't understand the large discrepancy between your simulation and Rawr's/Toskk's models. I understand there's some fundamental differences between simulations and models, but there shouldn't be that big a difference that they both compute about 8500 DPS for BiS while you're at 7800.
I know the models kinda "average out" CP waste, could that make that big a difference?
So, I thought to eliminate that from comparison. As any feral with raiding experience knows, even with CP waste(you crit at 4 CP), it is quite easy to maintain a rip and roar rotation(i.e. no bite). Even with that change, both models come in at just a bit above 8400 dps.
I could write this off to maybe Rawr multiplying agi by 25% or something like that, but two models within single digits of each other both being wrong? Seems quite unlikely. I also thought you might have simulated human response time and connection lag, but with the way energy works and cats not being GCD locked, do either of those make a big difference?
Anyway, I'm not trying to say you're wrong, just trying to figure out what's causing the noticeable difference. My gear atm is at a point where the two models say I should stick with AGI gemming and your simulations says I should go armor pen.
Anyway, I'm not trying to say you're wrong, just trying to figure out what's causing the noticeable difference. My gear atm is at a point where the two models say I should stick with AGI gemming and your simulations says I should go armor pen.
AFAIK, Rawr is based on Toskk's model. That explains why they produce similar results.
I believe you are being told to go AGI in Rawr because you never changed gemming templates, it uses agility by default unless I am mistaken. In addition, Rawrs potential DPS listed is pretty unrealistic, at least in my experience, I do not know what causes the discrepancy, I do know that Nightcrowlers simulation gives more realistic numbers.
I have gemming templates for str, ArP, and agi. JC and regular.
How unrealistic is it though? And I'm just speculating here. The only major differences I can find is the rip/rake uptime in FBN's simulations hover around 81%-83%, while the models have them at near 100%. Now I understand that near 100% uptime on both of those is impossible, but lower 80's seems low too. And even then, a max of 20% uptime on rip/rake doesn't seem like it would account for that much dps.
I have gemming templates for str, ArP, and agi. JC and regular.
How unrealistic is it though? And I'm just speculating here. The only major differences I can find is the rip/rake uptime in FBN's simulations hover around 81%-83%, while the models have them at near 100%. Now I understand that near 100% uptime on both of those is impossible, but lower 80's seems low too. And even then, a max of 20% uptime on rip/rake doesn't seem like it would account for that much dps.
Assuming a rip tick of 3k in a 300 second fight a 20% difference in up time is 300dps.
If the models are calculating in clearcasting procs from rip ticks that'll be missing shreds, which is also then missing combo points that would probably turn into FBs.
So yeah, 600-700dps due to a 20% difference in both bleed uptimes isn't that hard to imagine just napkin mathing.
Originally Posted by JamesVZ
Yes but you can't guarantee that every 10 man raid would have Party Grenades available right now, so giving this effect to Disco Priests I think is a worthwhile endeavor.
Does the algorithm need some work then? The situation I raid in, and I imagine quite a few of us raid in, is with an arms warrior. And I'm taking out FB to simplify the comparison since I'm removing it from the higher dps result. So you're left with:
Savage Roar
Rip
Rake
Shred
Insert tiger's fury when you're below 35 energy. That's not a terribly complicated priority queue. The only complications to rip and rake uptime that I can see is if they closely coincide with roar and rip/rake. Since that priority queue isn't using FB, how often is a 34s, 22s, and 9s abilities going directly interfere with each other such that they lower uptime by almost 20%?
@Lolaan: It depends on fight lenght. During berserk the value is about double, so it depends on how much do you lose. For istance in my simulation (the one to make trinket ranking) (5 min fights) I tied wrathstone to the first berserk (about time 00:05) then used it at time 02:05 and time 04:05 suddendly after a tiger's fury to start with 100 energy. Waiting for the second berserk in a 5 min fight means that I can use it only 2 times.
In that situation you should obviously save the second berserk for the third use of the trinket. The original question of whether to save the trinket for zerking forgot you can do it the other way around. Using tf -> trinket + zerking as soon as you can without getting agro, then the trinket with tiger's fury, then not burning zerking right away, but waiting until the trinket comes up again lets you stack much better. Probably using a haste potion with the second trinket use too.
AFAIK, Rawr is based on Toskk's model. That explains why they produce similar results.
No, it's not. Rawr's model is entirely separate. A version of it months ago, pre-WotLK, was, but that got completely replaced with WotLK.
EDIT: Rawr (and I believe Toskk's) do indeed expect a near-100% uptime bleeds (near- meaning whatever's the max with your hit/exp). That's idealistic, sure, but it is indeed what's attainable, and what you should be striving for. Besides, if you're not using that much energy on Rake/Rip, you're using it on something else like Shred or Bite, so that the DPS lost is less.
I do know of two issues with Rawr's Cat calculations currently:
First, it doesn't have the updated FB formula from 3.0.9. I'll be releasing Rawr v2.2.6 tonight to fix that. That will *increase* damage overall though, so FBN still will be way behind.
Second, it doesn't know yet how to spend only some 'extra' energy on Shreds, and some on Bite. If you're going to bite, it's going to dump all your extra CP on bites. This produces a couple odd situations where sources of additional energy actually slightly lower your DPS. This is a complex problem to solve, and I will be working on solving this soon, but the impact is small and rare.
@Deathwing: well actually It's possible that I need a small update (the new discovered weapon and legs from hard modes) I haven't updated my sims but they should push another 200 dps or so at least from gear comparing. FBN was always lower than Rawr. Actually putting my data into simcraft and toskk's still output a lower dps than rawr and something between FBN and rawr itself. Looking at "real world" data it seems that the "real" output is also lower than FBN. As for the model difference, the uptime of rip and rake in my model is wanted. When FBN need to use an ability it runs some quick theorycraft based on your condition to estabilish wich ability will made you lose less dps. For istance if you have 6 seconds on rip , 14 seconds on SR and 90 energy it will tell you to FB, Obviously you'll lose RIP uptime but will lose more dps if you wait for rip to expire. As for rake: using clearcasting for rake is a dps loss, using rake at 5 cps instead of a finisher is a dps loss, hit/expertise uncapped will lower your rake uptime, using rake instead of mangle without a manglebot is not an option, sum up all of them and you will see that rake downtime. You can adjust latency into the simulation, If I remember correctly the standard lag+reaction is 90 ms.
@Astrylian: If you want we can double check our programs, have you got a table or something with the different abilities formulas, base stats and so on? I think that the only real issue can be there.
EDIT:
Another source of dps difference is Arpen. Actually you can gem arpen only till the soft cap, rawr average out trinkets, so you can go over the soft cap without problems with rawr.
EDIT2:
I've putted into rawr the same data/profession/equip of my BiS gear simulation (as I said I didn't know some of the gear when I did my sims) here the results (far more near to my simulation than expected):
This may be a dumb question but I've read all through the thread and haven't seen an answer. Is the 566 armor pen soft cap (with mjolnir runestone) including the armor pen food buff, or without?
ArPen Cap is at 1232 (actually 1231.6 or something).
Mjolnir Runeston proccs with 665 ArPen.
1232-665 = 567.
So 566 (actually the cap is at 566.6, so u may go for 567) ArPen after everything except Trinket Procc is the cap.
I'm still shocked that mongoose is the best weapon enchant for kitties now. My Rawr rates weapon enchants different:
executioner - 91.64 points
berserking - 90.93 points
massacre - 75 points
mongoose - 72.14 points.
How come the big difference? Agi is rated quite similar in FBN and RAWR - relatively close behind ArPen. But in RAWR, Haste is terrible. Could that be the reason?
Actually Murna is correct the cap is more likely to be 567 than 566 (I just don't want to waste anything).
As for Mongoose vs. Berserking with my gear choce I have (rawr 2.2.6) (in brakets dps value by FBN):
Berserking: 99 dps (94 dps)
Massacre: 82.5 dps (82 dps)
Mongoose: 80.6 dps (100 dps)
Executioner: 77.9 dps (57 dps) (but remember that rawr doesn't take arpen soft cap into account).
So there is a 18% difference between bersk and mongoose in rawr while in my sims there is 6% difference in favour of mongoose. Actually I think it's not really an "haste" problem while str/ap are valued less in my sims than in rawr. And it's abvious because rawr has a rip/rake 100% uptime and for those spells ap is more important than haste/crit.
Rawr actually does account for soft cap. Once you get within a gem's worth of 566 rating(and you edit the proc on runestone), the item selector stops suggesting ArP over agility.
I wonder if it would be worth running a simulation that put emphasis on bleed uptime instead of using extra CP for FB. I'll have to find a copy of visual studio somewhere.
EDIT: Night, I think you quoted the wrong DPS row when you compared with Rawr. I put your U25 BIS into Rawr as well:
@Deathwing: Have you got the last version of Rawr (2.2.6)? Actually for me 8195 is the optimal rotation (equivalent to: rip+rake+shred+FB with 4point SR), if I try to make a custom rotation without FB the custom rotation dps becames 7885.
I'm also using a 675 duration to take care of Heroism uptime. I'll send you a screen shot when I have time.
I know for a fact that Toskk's doesn't take into account fight start-up sequences (i.e., you can't possibly apply rip until 6 seconds into the fight, and as late as 8 seconds or worse if you're saving berserk) and I would be surprised if Rawr has implemented that. The simulators do this "naturally" and properly model damage/buff downtime at the start of a fight.
I know for a fact that Toskk's doesn't subtract Rip/Rake uptime when he assumes that all OoC procs are used for shred, and would not be surprised if Rawr did the same. If you get an OoC proc when you're supposed to refresh Rip or Rake, you either get one or the other, not both.
I'm pretty sure that Rawr takes all extra CPs and uses them to calculate the number of FBs, but I know that they get wasted far more often in simulation/practice.
I'm pretty sure these factors alone will close the gap-- unlike in Classic/TBC, the "ideal" is a lot further away from reasonably reachable than it used to be, which is why the sims are closer to what people are achieving than the simulators.
Nightcrowler: Yes, I'm using 2.2.6. Why are you boosting the fight length so far out for bloodlust?
Downing 65 energy on a FB while you have 6s rip, 14 sr seems a bit suspect. That bite probably doesn't have that much better DPE than a regular shred. Plus you're definitely going to lose some damage from rip.
Allev: Those are some good points, thanks. Has anyone considered using OOC on rake IF it's going to affect rake uptime? 7 energy saved vs. 1s minimum of rake dot.
I'm working on a simple algorithm that puts the extra CP towards refreshing SR early to avoid times where rip and rake are down due to more important abilities needing a refresh.
One thing that might help is to run a simulation and calculate the average DPE(mostly for bite) and the DPCP(damage per combo point).
Just started tracking the ICD on my Grim Toll and Greatness deck. When a thought occurred to me. How long should I wait for a Greatness deck proc before re applying rip to get a dps boost. or will waiting at all be a dps loss? Any thoughts. Also are there any trinkets that you should wait for the proc, like the blood of the old gods ap buff?
And on a similar question. Given the 2 piece tier 8 set bonus, is it better to rake or shred during a grim toll proc. Dont remember reading about this anywhere in this thread or 'cat dps rotation' threads.
If your looking to track the ICD on your trinkets. I got a mod from cursed called "InternalCooldowns" That allows DoTimers, or similar mod, to track the ICD.
@Deathwing: After the patch that fixed 30 to the maximum wasted energy and without berserking FB is basically always better than Shred (you should take into account wasted CPs or better, not generated one).
For the sake of true I've runned between many others the following test (and obviosly the actual FBN algoritm works better):
- using rake when OOC proc
- using rip when ooc proc
- using FB when ooc proc
- put an energy threasold to FB
- don't use FB at my algoritm level but at any other energy/rip debuff timer/SR buff timer level, the algoritm that's actually in FBN gives the better result for a large variety of stats level (basically from t7.10 to t8.25)
- don't use FB at all
- trying to maximize rip/rake uptime
- trying to clip rip
- trying to don't clip SR
As for DPE/DPCP calculation you already have the average bite damage and average energy spent on bite so you can easly compute the DPE from the first page reports. DPCP I don't understand what you mean, If you think about biting at 3/4 CPs. I've tried it. It's not good.
The algoritm I'm using already refresh SR early to avoid rip downtime, otherwise they will be far less. The main problem (that's not a problem) is that I don't want 100% rip uptime, because it's worst, what you want is to maximize your dps, we don't decide how to do so but it's the game mechanic.
I'll try to explain briefly how the finishers work in FBN (you can easly read both C++ or LUA code for more insight).
At every time instant it's computed the damage and the energy cost of the different abilities.
In the mean time we are keeping track of the following things:
- CPs
- Average time to generate x combopoints (it may vary depending on haste/ooc interaction, crit chance and so on)
- SR timer
- RIP timer
- Tiger's Fury Timer
- You must fix a wanted cps threasold for SR into the sim (standard is 2, for low level gear the best is 3 and 1 for 4T8)
- If SR ends -> refresh it (it's always better)
- If you reach 5 cps it's time to compute what you want to use:
-> If RIP is still up I compute the average time to reach 5 CP + have enough energy to RIP
-> the average time to reach XX CP for SR + have enough energy to SR
-> the cumulative time
All timer takes into consideration Tiger's Fury CDs so I'm actually computing 6 different times.
Then based on FB average damage (with instant condition) and RIP average damage I compute how many RIP ticks I can lose to let FB+RIP down time being better than nothing+RIP and I add this time to the RIP timer.
At this point we have different choce based on the previous computations:
- wait rip to expire without attacking (during this period I refresh rake/mangle if needed) and then rip
- clip SR
- Use Bite
All 3 (and only those 3) are possible. With this method I have an high SR uptime (>98%) because SR must be maximized, I could also have an high RIP uptime (removing the extra time for rip downtime) but what I've seen is that it lower the dps.
Usual value (the algoritm is computed on the fly so they change depending on gear/buffs etc..) are 3-4 seconds clip on SR and about 6 seconds of RIP downtime each rip. So as a general things "on average" it uses bite when SR>4, RIP>8, SR+RIP>14, but the value can chamge a lot.
As for bloodlust bah I had a stupid idea probably, It's a better idea to run my sim without it and uncheck it on rawr to see the difference it's also probably better to run a very long (1 hour?) simulation so I can cut side effects like starting time, I'll do that and post the results, now that I think about it could be cool to plot dps vs. fight time, I'll do it when I have some time.
Thanks for the in-depth explanation, cleared up a lot of the questions I had. And I know it takes a lot to articulate a post like that, so I appreciate it.
The reason I was asking about DPCP: isn't there a 4th option? Wait for rip to expire while attacking(only to prevent energy waste), and then rip? My idea was to see if there was a value you could attach to a CP you know would be wasted.
Either way, that first list contained all the different simulations I wanted to run, so I guess that puts this idea to bed.
I wonder if there might be some way to put FBN's simulation code into Rawr?
Will there be an update of the AddOn soon? Or better: how do I import your code into the addon?
Every druid that asks me "how to dps" gets the url to this addon ^^
What is everyone who uses FBN using for SR minimum refresh CP? I'm trying to experiment with different values right now to see which one gives me better dps, but have to find any conclusive anecdotal evidence that favors any particular value right now. The big this is that I'm still using 2T7/2T8 so my assumption is that a value of 4 would be best, but once I get 4T8 set, 1 or 2 would probably be better.
Will the new FBN automatically pick a value for SR minimum refresh CP instead of it being user set?
EDIT: I've also noticed an option to check if you're using the Relentless metagem. Is there any particular reason for this? I personally don't see how it impacts the rotation since it should affect the majority of your damage roughly the same.
06/03/09, 6:06 AM
