Maybe this is just strategy differences, but I don't understand why people value AoE so highly on Anub hardmode. It's fairly simple to reach the raid AoE dps required for the trash, and I think it's much more important to reduce the length of p3 as much as possible? It just seems a lot more relevant to focus on our single target dps, and let the AoE part of the fight be more of an afterthought. I do maybe 5-6 hurricanes in total in that fight, and I wouldn't swap in 4T9 for that over 2T8.
Is Simcraft taking into account the shortened mean cycle length due to WiseEclipse when you have a higher Lunar crit rate?
Could you go into a little more detail here?
There are a couple Moonkin-related items I'd like to see investigated/fixed in SimC:
(1) Using SimC travel mechanics to properly model Wrath.
(2) The effects of lag and spell queuing for uber-hasted Wrath casts that drop below 1sec.
There are a couple Moonkin-related items I'd like to see investigated/fixed in SimC:
(1) Using SimC travel mechanics to properly model Wrath.
(2) The effects of lag and spell queuing for uber-hasted Wrath casts that drop below 1sec.
Basically, what he's saying is that the common practice for most Moonkin (and what's mainly modeled in WC) is that you will cancel your Lunar Eclipse as the last Starfire cast is going off, in order to guarantee a transition into Solar Eclipse. It reduces the duration of Lunar Eclipse slightly, but causes the overall cycle to be shorter, which increases the relative uptime of both Eclipses in total. It appears that neither SimCraft nor Rawr models this effect.
Maybe this is just strategy differences, but I don't understand why people value AoE so highly on Anub. It's fairly simple to reach the raid AoE dps required for the trash, and I think it's much more important to reduce the length of p3 as much as possible? It just seems a lot more relevant to focus on our single target dps, and let the AoE part of the fight be more of an afterthought. I do maybe 5-6 hurricanes in total in that fight, and I wouldn't swap in 4T9 for that over 2T8.
I think you might be understating how adept we can be at AoE dps. Looking through WoL's, the majority of Moonkins are dedicated AoE dps on this fight with single target being an afterthought.
Of course this is a guild by guild basis, another thing that dictates how it goes is tanking strategies. 2 tanks pulling 2 adds seperately to slow the cast time is definately the more common strategy. However 4 adds stacked reduces the time spent AoE'ing a lot. I still don't see how you can only cast 5-6 Hurricanes in the fight. I cast 3 per add spawn. 4 per spawn in p3 typically.
We also use 7 healers, so that could dictate some differences, however I still can't see myself ever using 5-6 per fight.
Unless of course you mean normal mode, in which I don't think I have ever used Hurricane on.
In my guild there are some aoe-guys and the other ones are single-dps-guys ... and DKs and Mages are much more stronger in aoe than me, so sometimes i am single-dps too, and then i dont hurricane a lot too
It's of course hard mode, I don't think normal mode has been mentioned in this thread at all.
Guess it's mostly raid composition and strategy differences. In p1 we don't AoE the second spawn at all, and I rarely get off more than 2 hurricanes at most vs the first pack. Looking at WoL I see different strategies, some moonkins with hardly any hurricane damage at all and others with hurricane as their by far highest damage ability. I just don't think it's as clear cut that we should focus our gearing towards AoE as this thread might suggest. It really depends on your raid setup.
Basically, what he's saying is that the common practice for most Moonkin (and what's mainly modeled in WC) is that you will cancel your Lunar Eclipse as the last Starfire cast is going off, in order to guarantee a transition into Solar Eclipse. It reduces the duration of Lunar Eclipse slightly, but causes the overall cycle to be shorter, which increases the relative uptime of both Eclipses in total. It appears that neither SimCraft nor Rawr models this effect.
If you cancel the buff while Starfire is in mid-cast it does not benefit from the buff, correct?
Can I model this with a cancel_eclipse action?
actions+=...
actions+=/cancel_eclipse <= if starfire/lunar cast cannot finish before buff ends, just cancel it
actions+=/wrath,eclipse=trigger
actions+=/use_some_trinket
actions+=/starfire,eclipse=benefit
actions+=/wrath,eclipse=benefit
actions+=/starfire
Because of client/server timing issues, you can have a single Starfire
1) Benefit from Lunar Eclipse
2) Proc Solar Eclipse
WiseEclipse consistently does this by doing a /cancelaura as soon as the ?spell-complete? event occurs. CancelAura is not a protected part of the WoW API, so a mod can do this with no action your part.
In sc_druid.cpp, starfire_t::execute, you'd need to change
if ( result == RESULT_CRIT )
{
...
if( ! p -> buffs_eclipse_lunar -> check() )
{
p -> buffs_eclipse_solar -> trigger();
}
}
to something like (new stuff is pseudo-code)
if ( result == RESULT_CRIT )
{
...
if( !p -> buffs_eclipse_lunar -> check() ||
p->wiseeclipse && starfire_t(p).cast_time() > p->buffs_eclipse_lunar->time_remaining())
{
p -> buffs_eclipse_solar -> trigger();
}
}
and with a similar change in wrath_t::execute().
As another twist, for lunar->solar, the reaction time check can typically be ignored. When I see that my SF is going to complete with < 1.5s or so remaining in Lunar, I know that Wise Eclipse will kick in, and I've got an almost 100% chance to proc solar. In this case I will go ahead and queue a Wrath while the SF is is still in progress.
As another twist, for lunar->solar, the reaction time check can typically be ignored. When I see that my SF is going to complete with < 1.5s or so remaining in Lunar, I know that Wise Eclipse will kick in, and I've got an almost 100% chance to proc solar. In this case I will go ahead and queue a Wrath while the SF is is still in progress.
This will be the harder thing to model. Normally, we use sim_t.reaction_time to delay a player's ability to react to a positive buff. For now, I'll probably simply remove the solar reaction-time check in wrath_t when wise_eclipse=1.
I posted about talents and glyphs earlier in this thread, and well.. We are still working on downing him but I thought I could share some things.
Last WoL : World of Logs - Real Time Raid Analysis
Now I did some quick math with this logs and considering my numbers I have to admit that my previous statement about Genesis was incorrect. As you can see, the way my guild handle the fight, Dots+Hurricane(with 5/5 genesis and 0/2 Imp MF) do 40%~ of my dmg, when Starfire+Wrath(with 0/3 Imp IS) consistantly do 50%~.
Really quick(and basic) napkin math shows 0.5%dps increase per point on Imp IS, 0.4%dps increase per point on Genesis and 0.3%dps increase per point on Imp MF (2p T9 - and higher NG upgrade that I can't really put into numbers.. but it may be worth more than Genesis).
I know this is quite innacurate since the initial numbers I'm refering to are with 5/5 Genesis 0/2 Imp MF and 0/3 Imp IS, but it only would favor Genesis in the math. As you can see Genesis is inferior to Imp IS anyway, and at least on par with Imp MF.
So.. unless Hurricane is 40%+ of your damages you should use a standard spec with 2/2 Gale Wind, and no additional points in Genesis.
I was hoping to get some use out of this talent, but I was wrong, so please accept my apologies
Wise-Eclipse is implemented in revision r3748 of SimulationCraft.
Driven by our T9 BiS config it represents a 140dps increase. Now I feel guilty for not implementing it sooner.....
Erdluf, in regards to auto-queuing Wrath after a "wise-eclipsed" Starfire: If the crit rate for Starfire is over 80% I reset the reaction time of the Solar-Eclipse buff. This isn't perfect, but it is a decent start.
Packing up a new release now......
EDIT: simc-322-3 available for download. Toggle wise_eclipse=1|0 to experiment with/without. (on by default)
Last edited by dedmonwakeen : 11/10/09 at 1:25 PM.
Yea 10k iterations on normal runs (50k for scalefactor, 1k if I just want the plot-feature)
Did another run:
T9 => T10
so its 4T10 vs 2T10+2T8 (ilvl 258)
I added WiseEclipse support to Rawr and was playing around with it. I discovered something interesting: The reason why 4T9-258 seems to work out better than 2T8+2T9 is actually because at ilevel 258 levels of gear, you are smashing the crit cap, which devalues the 2T8 set bonus. Downgrading to the 2T8 pieces doesn't necessarily remove enough crit rating to improve the value of the set bonus to where it would overpower the stats on the ilevel 258 gear. So if you're fully kitted out in all ilevel 258 gear and have blown over the crit cap (and my example set is 10% over cap), then obviously, 4T9 will be more powerful than 2T8. If you're flirting with the crit cap line or below it, 2T8+2T9 is better than any combination of 4 ilevel 258 pieces you can put together.
I added WiseEclipse support to Rawr and was playing around with it. I discovered something interesting: The reason why 4T9-258 seems to work out better than 2T8+2T9 is actually because at ilevel 258 levels of gear, you are smashing the crit cap, which devalues the 2T8 set bonus. Downgrading to the 2T8 pieces doesn't necessarily remove enough crit rating to improve the value of the set bonus to where it would overpower the stats on the ilevel 258 gear. So if you're fully kitted out in all ilevel 258 gear and have blown over the crit cap (and my example set is 10% over cap), then obviously, 4T9 will be more powerful than 2T8. If you're flirting with the crit cap line or below it, 2T8+2T9 is better than any combination of 4 ilevel 258 pieces you can put together.
I've been thinking about this, but I don't think it should be quite so dramatic. Crit typically devalues (just using my own gear) from something around 1.25 DPS/point to 0.8 when you cross the Lunar cap. Assuming the difference between 2T8 and 4T9 is somewhere around 100 DPS (from my post above), we can still estimate that you have to be around 200 crit beyond the cap before it tips the scales. But (barring Focus Magic), that's a good bit farther than most people will get in current gear. My various attempts that juggling BiS sets for that post typically sat at around 800 crit rating--right around the cap.
Whereas the full ilevel 258 set I put together with Rawr had a whopping 1019 crit rating, which is so far beyond the cap that the 2T8 bonus loses ~75 dps. It's certainly possible to put together a set with a lot less crit rating and run 2T8+2T9; the Optimizer is far from perfect. I'm just saying, it was an observation that I noticed when playing around with gear sets, and it's certainly achievable with the high item level gear.
I played around with Rawr and the T10-264 gear. It's probably not a surprise to anyone here, but if you're wearing the 258 level T9, assuming these are the stats and not placeholders, you're still going to be wearing 2T8+2T9 for a while yet. The 4T10 set bonus may be the strongest 4-set out of them all, but the 2T10 is showing so weak and the 2T8 and 2T9 bonuses are so strong that it's just not enough. One interesting thing I did notice, though, is that it dropped the [Legwraps of the Awakening] for the T8 legs and swapped in the T9 gloves, because the T10 helm is overwhelmingly strong.
Maybe this is just strategy differences, but I don't understand why people value AoE so highly on Anub hardmode. It's fairly simple to reach the raid AoE dps required for the trash, and I think it's much more important to reduce the length of p3 as much as possible? It just seems a lot more relevant to focus on our single target dps, and let the AoE part of the fight be more of an afterthought. I do maybe 5-6 hurricanes in total in that fight, and I wouldn't swap in 4T9 for that over 2T8.
What you said has some truth but also a lot of emphasis was put on AoE because of the lack of gear people had when they first did the encounter. Now that we've been saturated in >=245 gear to no end, DPS isn't as challenging as it would be in <=239 gear.
So it seems we'll have to wait for 4pc t10 before we start breaking the other set bonuses, is that right? Or do we need the higher level T10 set pieces assuming they exist?