This project is read-only.

Trinket interaction unmodelled?

Topics: Rawr.Base, Rawr.Base.Items
Jan 3, 2010 at 4:15 AM

So I equip mjolnir's and whispering fanged skull. I am not at ArP cap so I can take the whole ArP proc.

It would seem to me that I would always want these two trinkets to proc together so that the increased AP is used when I have loads of bonus ArP.  Them being desynced should lower DPS.

When I move the trinket offset slider the expected DPS remains constant.  Are these sorts of interactions not modeled?

Jan 3, 2010 at 4:29 AM

Correct; the offset is only for syncing or desyncing procs which share a cap. ArPen and AP don't scale significantly non-linearly with each other, so averaging them out as far as each other are concerned is fine.

Jan 3, 2010 at 8:08 PM

Did some napkin math since it was still bugging me heh...

With a set of gear I was testing with Mjol and WFS I was 9554 DPS.

So then I removed the proc effects from both trinkets.  Now I show 9003 DPS. 

Then I added the procs as static values to show what I would be with them proced and am at 10697 DPS.

Given that the two procs "should" line up exactly throughout the fight so my average DPS "should" be 9003*2/3+10697*1/3 = 9568 DPS.

That is assuming instant up and always in sync of course.  Assuming only 30% up time instead of 33% gets you 9511 DPS.

So the net answer is that you are right.  The approximation done in the modeling is good enough. It could be over valuing or under valuing but it's in the right ball park regardless. Chances are, actually, it's overvaluing.  Only on the "perfect" case of 33% and always lined up do you beat the listed DPS and then only barely.  A couple percentage points off and you start to plummet.

Once you add in the ICC ring procs, berserking, various buffs that might be on or off at a given time, whether or not Bloodlust lines up with the're in the land of approximation anyhow I guess heh.


Jan 4, 2010 at 7:53 AM

Our "averaging" is more complex than your napkin math.  We consider how long it takes to refresh based on the ICD *and* the trigger chance (ie 50% on crit, or 15% on hit, etc).

Also note that the fight duration means a lot here -- if you're looking at 15sec effects and 45sec ICDs constant refresh, your uptime over a 60sec fight is going to be 50% (two procs occur).  Over a 300sec fight, we expect to see 35% uptime (7 full procs), but at 315 we expect 33.333%.