SimulationCraft for Warlocks (4.0.3a numbers at 80 and 85) - Page 201

VoidStar · 11/23/10, 3:46 AM

Immolation Aura's DPET (and thus its position in an ideal priority list) changes linearly (although in a stepwise fashion because of its ticking nature) with remaining Metamorphosis time below 15 seconds.

To produce a "perfect" priority list there would have to be a conditional Immolation Aura casts added in several places in the list below its full duration DPET, something that would be quite difficult for a real person to emulate in real-time.

The absolute maximum you could squeeze out of an extra cast is 6 seconds worth of the second Immolation Aura (which is probably impossible to actually achieve) which would be 6/15ths of the DPET: 33025 for the profile given in the OP, still a respectable number but lower than BoD, Corruption and arguably HoGd refreshing Immolate. 5/15ths is better than Shadowflame but a challenge to achieve in reality. 3/15ths is still enough for it to be better than the 'filler' spells Incinerate, Shadow Bolt and Soul Fire. 2/15ths is very slightly better than Shadow Bolt.

TL;DR version: it does appear worth casting Immolation Aura twice during Metamorphosis under certain circumstances, but the conditions change very quickly and the benefit is relatively small.

Berthold · 11/23/10, 7:48 AM

Hi all,

as it seems now, the only interesting decisions we are going to make in cataclysm are how much crit/haste/mastery we want on our gear (apart from trinkets). Intellect is only a function of the itemlevel and hit *needs* to be at 17% hit, so we do not have that many otions.
Taking a look at the t11 BiS list we see that we have ~4000 rating points on crit+haste+mastery and depending on the actual gear and reforging it is divided differently.
I took this gear setup and simulated different combinations of crit/haste/mastery and made some graphics..
DESTRO:

DEMO:

AFFDRAIN:

Explanation: the back left corner has 0 haste/0mastery/4000 crit. Moving to the right side converts (!) crit into haste and moving to the front converts crit into mastery. This yields a barycentric triangle [1] with a heightfield(the dps). The front/right corner is just filled for the sake of the graphic - only take a look at the area left of the front-left to back-right diagonal.

Conclusions from the graphs:
We see that Destro and Demo have two haste bumps (that should be the haste values for additional immo ticks), and we see that afflidrain does not have them.
We also see that moving to the right, i.e., converting crit to haste, is always a good idea. For the affdrain simulation we see that crit to mastery also makes sense and sometimes is even better than crit to haste (depending on the current position on the surface).
In the case of destro it seems as if crit and mastery are round about the same (no change when converting one to the other).
Finally, demo should not be using mastery right now as the dps goes down when going from crit to mastery.

Overall conclusion and suggestions:
All of that is not really new, but (!) it took us quite long to figure that out for different gear sets, whereas anlyzing the 3D surfaces is quite easy. My main point of this post is to show you what can be interpreted by using some metatools using simcraft in its core and I hope you see why that is benificial. The question now is, whether we would like to see a 3D surface plotting feature as part of simcraft, and whether the developers would do the coding, because right now it takes some time to generate the graphs manually. This is even more important if you would want them to have more than 40 entries per surface.

Your thoughts?

1: http://en.wikipedia.org/wiki/Barycen...m_(mathematics)

p.s. Other classes might be a little bit more problematic because they juggle more then 3 variables

Dastey · 11/23/10, 1:11 PM

How do we calculate how much haste we will need to reach the 50% cap with 15% from ISF, 3% from Dark Intent and 5% from raid buffs? Do we simply say 50 - 23% = 27% and then we need that? Or are theese buffs multiplicative so 1,15 * 1,03 * 1,05?

radiskull · 11/23/10, 6:29 PM

I may be reading this wrong, but the level 80 Affliction profile doesn't seem to be doing anything with its Soulburn - it's not casting Soul Fire or Drain Life. The AffDrain profile will eventually get around to casting a Drain Life to use it up, but is this intended, or am I reading the profile incorrectly?

dedmonwakeen · 11/23/10, 11:11 PM

Very pretty pictures!

At the moment, SimC charting restricts itself to whatever is available from the Google Chart API. I try very very very hard to limit 3rd-party dependencies. Also, having a format where the entire data is enclosed in the html output report is much more convenient that managing separate image files.

However, at the very least we could provide options to generate N-dimensional data using smooth_rng and limited iterations. The smooth_rng code improves precision at the expense of accuracy. For the purposes of generating a bunch of relative data points, it should be okay to use.

wasniahC · 11/24/10, 3:48 AM

Gems will no longer be an issue - A blue post stated they were going to change the meta gem requirements back, but that it couldn't be hotfixed, and would be changed in a patch.

Berthold · 11/24/10, 4:17 AM

Originally Posted by dedmonwakeen

Very pretty pictures!

At the moment, SimC charting restricts itself to whatever is available from the Google Chart API. I try very very very hard to limit 3rd-party dependencies. Also, having a format where the entire data is enclosed in the html output report is much more convenient that managing separate image files.

However, at the very least we could provide options to generate N-dimensional data using smooth_rng and limited iterations. The smooth_rng code improves precision at the expense of accuracy. For the purposes of generating a bunch of relative data points, it should be okay to use.

Thanks. If simcraft has an option to generate variations of crit/haste/mastery,calculate the dps, and then create a CSV with those values, it would be really easy to use a standard spreadsheet application to generate meaningful figures. I am really looking forward to that and am especially interested how the surfaces look for other classes :-)

p.s. I added a suggestion on the CSV-format in the googlecode project.

marky111 · 11/24/10, 2:48 PM

Thank you Berthold, awesome work!

One Question: Regarding Destro specc, if I go to the right it always goes up. Does that mean haste is best stat regardless being at a haste threshold?

wasniahC · 11/24/10, 10:17 PM

Originally Posted by VoidStar

TL;DR version: it does appear worth casting Immolation Aura twice during Metamorphosis under certain circumstances, but the conditions change very quickly and the benefit is relatively small.

Then it might be worth putting a higher priority on it, perhaps? Another thing to consider is that if its DPET gets lower by delaying it, its priority may in fact be higher than BoD/Corr where maximising DPS in long run is concerned? (since their damage would be delayed, not reduced, by waiting)

Ehvyn · 11/25/10, 1:45 AM

Originally Posted by marky111

Thank you Berthold, awesome work!

One Question: Regarding Destro specc, if I go to the right it always goes up. Does that mean haste is best stat regardless being at a haste threshold?

No.

Once you get to your threshold - in this case 1406 - you will see no gain from haste until you get to the next threshold. The only reason why we have these Thresholds is to get the extra tick from our Immolates and thus stacking haste past the threshold is nothing but wasted stats unless you can get to the next tier.

Berthold · 11/25/10, 3:53 AM

Originally Posted by Ehvyn

No.

Once you get to your threshold - in this case 1406 - you will see no gain from haste until you get to the next threshold. The only reason why we have these Thresholds is to get the extra tick from our Immolates and thus stacking haste past the threshold is nothing but wasted stats unless you can get to the next tier.

This not entirely true; the graphs were generated for level 85 so the rating would not be 1406 as it now for level 80.
What you can find out using these graphs is that a) haste is a very powerful stat (for all three specs) b) hitting the first cap is really a good idea c) stacking haste after the first cap is still better than crit/mastery (because the hill is steeper to the right) and d) after hitting the second cap (although I doubt that this is possible in the first tier) you might want to stop stacking haste and consider others, e.g. crit, because you benefit more from those than from additional haste.(the hill goes down)

p.s. there may be single points where it might be better to use crit or mastery, but the very sparsely generated terrain does not allow for more conclusions.
p.p.s. As a rule of thumb for Destro: stack haste till second cap (should be around 3000-3500 rating, you could calculate where immolate gets another tick) and afterwards do crit or mastery, they are not very different.

VoidStar · 11/25/10, 4:17 AM

Originally Posted by wasniahC

Then it might be worth putting a higher priority on it, perhaps? Another thing to consider is that if its DPET gets lower by delaying it, its priority may in fact be higher than BoD/Corr where maximising DPS in long run is concerned? (since their damage would be delayed, not reduced, by waiting)

I was wrong earlier when I claimed that Corruption was better DPET than 5 ticks of Immolation Aura: I must have read the DPE number as it's actually slightly worse (it's still still better than 4 ticks though).

The challenge of working out the proper place in the priority list for Immolation Aura is that delaying the first cast devalues the second cast and the second cast is always going to be only a fraction of the worth of the first.

If we were aiming to get benefit from a second cast of Immolation Aura we would want to cast both of them at the earliest possible moment as if we delay by more than a couple of seconds then it isn't worth casting the second aura at all. We should be able to squeeze 4 or 5 ticks out of the second cast which is significantly better use of the time than our filler spells. In most cases casting the aura instead of one of the higher DPET spells merely delays those spells and thus the aura effectively replaces some filler spellcasting time, almost certainly a good move.

If casting the Aura would cause you to miss an opportunity to refresh Immolate with HoGd then this is worse than simply delaying the casts, but in a fairly complex manner. I think that it is worth investigating as more intelligent aura use should be a dps increase, and will slightly increase our mastery scaling.

• Zakalwe · 11/25/10, 4:30 AM

My gut (and lack of free time this week) tells me it's not really worth investigating the second immo aura cast except possibly as a simple "cast it a second time if you happen to have all dots up and cooldowns down at that time" priority, which would be a single line in the action list, just above shadow bolt.

But since the new version of the sim is out and downloadable, you can do all the investigating you want:
Downloads - simulationcraft - Project Hosting on Google Code

wasniahC · 11/25/10, 5:34 AM

Ah, if only I had a clue how to actually use the program. Time to start figuring that out, I guess
On the point of clashing immo aura/HoG, while I'm not sure why hand of gul'dan to refresh immo is currently higher priority than metamorphosis , since meta is off the GCD and HoG isn't exactly weak, it would prevent a first immolation aura from clashing with HoG (though the second one could still clash, yes)

One thing Zakalwe - If you are not going to test modelling with immolation aura being used twice atm, then shouln't it be put below all rotational abilities and proccs on the priority list, since having it higher than them doesn't realistically increase the amount of times it will be used in a fight? All it does is delay using rotational abilities, effectively increasing their CD, and give a chance to waste MC proccs, unless you manage to get a full 21 seconds with no filler used

Caltiom · 11/25/10, 8:01 AM

The complicated line for Immolation Aura comes from myself, I tested this about 2 months ago with the same thoughts in mind.

It turned out that 1 Immolation Aura per Metamorphosis did slightly more dps (it was something like 10-20 points). Another important thing that differs between the 2 options is that with 1 Immolation, you can wait on Tier10 4pc Buff for Immolation.

On a !T10 Set it could be better to use 2 Immolation's, but only a exact simulation can tell. Now that we release a beta version of simcraft, feel free to do so.