I like this App and i just wanted to let you know a bug.

When i check Leader of the Pack in the Retribution Model there is no increase in DPS under the Stats Tab.

Click on the little diamond shape next to each piece of gear. It will turn to a green diamond when you have selected something.

I managed to crash Rawr reliably by loading my mage's file while in the Ret model from playing with another char. When I swapped to Mage model first, I was fine. Since I don't have that problem with swapping back and forth between Mage and Cat models, I'd guess it's Ret-specific. One thing that would be cool is if the model type is saved in the character files.

In Rawr.Mage, I realize it may be a fair amount of work, but the spell rotation tooltip is really complicated to read, and I think part of it is cause the linear solver just tells you how much time is spent in each thing, right? It'd be really really handy if there's an easy way to make a graph of cooldown usage, like the x-axis is time and there are categories on the y-axis for each CD, so that it'd be easier to see when/where it's saying to line up the CDs.

The problem is this: Rawr needs to calculate the DPS of a particular set of stats several THOUSAND times a second, when doing the optimizer. Because it has to calculate DPS so fast, it does not actually simulate the timeline... it uses a linear solver, an equation with constraints, to figure out cooldowns, stacking bonuses, etc. Unlike some other DPS models, this model is very fast to calculate, and pretty accurate, but not quite as accurate as actually modeling cooldowns, spell times, etc on a timeline. In fact, it's quite likely that the rotation you see is not even possible, or very improbable, due to ability recoils.

So Rawr.mage can't show you the timeline, because it doesn't make a timeline, and its DPS may not even be correct if you actually created a timeline. It's pretty close, but not perfect... sacrificing a little accuracy for a lot of speed. Otherwise, the Optimizer might take an order of magnitude longer, or be able to check an order of magnitude fewer combinations.

It'd be difficult to come up with a timeline given the output, for sure, but I think it may be possible if you're only looking at relatively short-fights and use some tricks and/or if you do the conversion as a constraint satisfaction problem. I don't know C# at all, but I'll see what I can do about getting a Perl script to make SVGs of it when I can find some time.

I found a bug in the Rawr.Mage model. I was fiddling with the model to simulate a TK-Solarian fight (lots of AoE), and all of a sudden it told me that the only stat that mattered was mp5, and that didn't matter much. Turns out the spell cycle it was choosing was 600 seconds of idle time, and 5 water elementals. Odd. The stat I'd been fiddling with was mp5 from our shadow priest... with a little prodding, I determined that at 128mp5, it was fine... at 129mp5, It had me wanding the whole fight.

Obviously some edge condition was triggered. Could you pass this on to the Rawr.Mage TCer to check on?

Here's my Katikk.xml Rawr file. This is with the 128mp5 from shadow priest. Bump that up one point, and you'll see the bug appear.

I've been working on LP solution => timeline conversion for a long time. It's almost ready, but I didn't feel like it was good enough to release to public in b13. It's still missing some exception cases like Cold Snap, SCB, but general framework is in place. As was said you can't assume all constraints can be satisfied, so you want to produce a timeline that "makes the most sense". The basic idea is I created a scoring for a given timeline conversion that calculates how many impossibilities it has. I take a set of cooldowns and partition the LP solution into segments based on cooldown constraints. Then I compute the tightest packing of those segments given their constraints using a heuristic that the best solution will be the one that has the most fudge space for repositioning the cooldowns. After I have that ordering I start positioning them and insertinging non-constrained segments based on mana constraints and positioining mana regen items. If the tightest packing is longer than fight duration I drop out "least important" cooldown and recompute the thing. At the end the conversion with highest score is selected for display. It's probably more code already than all the other stuff . You can check out the latest build if you want to play with it. If anyone has a less "black art" solution I'd be interested to hear it.

I'm curious about the cat modeling. You state on the Wiki page that Cursed Vision was the best helm (pre 2.4), yet the Thunderheart Cover comes out on top. Is this due to hit rating being neglected? The same issue arose with Nether Shadow Tunic versus T6 chest.

Looking at the belt options, it suggests 10 strength and 5 strength 5 crit gems as the best option for Don Alejandro's Money Belt. It did not show two 10 agility gems as an upgrade, even though that would score higher.


Does Band of the Eternal Champion rating consider the proc?

Excellent work overall, and I plan to use your program to beg my GM for more gems.

Most likely, it listed T6 helm as better than CVoS for you, because it would give you a set bonus. Same with T6 chest vs Nether Shadow.

Regarding the belts, I'm not sure what you're doing there; do you have a gemming of it with 10agis? It doesn't pick gems (other than the Optimizer of course), you do.

Yes, Band of the Eternal Champion includes the proc.

The optimizer is still consistently giving me lower values than I currently have equipped. Maybe I have the best optimized set already chosen, but then shouldn't the optimizer find that same set?

Nah, I've seen optimizer give me a set ~40 dps lower than the currently selected one at full thoroughness.

I marked everything I was wearing plus a few extras I had in my bags as available. I had just assumed it was due to T4 piece selection to maintain the 2 piece bonus (I'm using Shoulders and Gloves, it might have come up with 2 alternate albeit lower dps options).

edit: I just tried it again. I currently use S3 helm, T4 shoulders, Nether Shadow Tunic, T4 gloves and Skulker's Greaves for DPS. S3 helm has a Glinting Noble Topaz, but aside from this it's all Delicate Living Ruby (meta requirements are fulfilled in other slots). The set that optimizer came up with is: S3 helm (same gemming), T6 shoulders, T4 chest, T6 gloves and T4 pants (all gemmed with Delicates). It's a 63 dps deficit this time. So it's pretty much due to choosing suboptimal pieces for maintaining the 2T4 set bonus (T4 legs are horrible).

In those 5 slots, items enabled for optimizer to use:
Helms: S3, T4, T6, Grimgrin Faceguard
Shoulders: T4, T5, T6, Razorfury Mantle
Chest: T4, Nether Shadow Tunic
Gloves: T4, T5, T6, Gloves of Dextruous Manipulation
Legs: T4, T5, T6, Skulker's Greaves

edit2: I used the known gemmings only option. I have 2-3 copies of all items, 1 gemmed with Delicate Crimson Spinels in all slots, the other with Solid Empyrean Sapphire in all slots, as well as any different gemmings used by my actual gear.