I opened the spreadsheet back up and removed the new gems, but it's still giving the calculation errors. I'm not very good with advanced excel stuff so I'm not sure how to track down what's referencing what.
Is there a comment thread somewhere for Magegraf? I am trying to use it to model the performance of the 40/0/21 spec to test out different trinkets, and I am getting absolutely insane results for the lightning capacitor, in the 500 dps range.
Maybe it's assuming this spec isn't worthwhile and didn't include it in its logic? :P after all there's no way pyro can out-do icy veins and no way spell power can out-do empowered fireball and molten fury unless you're intentionally gimping your DPS to stack crit (and yes, you'd be gimping your DPS to make 1 spec better dps than another which isn't very wise as you still do less dps than proper gearing for EITHER spec (yes even arcane/fire would still end up gearing in a way that would favor deep fire)).
On a side note if they reduced the itemization cost of crit by 10-20% and kept the budget invested into crit equal on current items arcane/fire might become a decent spec (although I hadn't actually done the math, it might take more than 20%).
Is there a comment thread somewhere for Magegraf? I am trying to use it to model the performance of the 40/0/21 spec to test out different trinkets, and I am getting absolutely insane results for the lightning capacitor, in the 500 dps range.
Yah i gotta say i am a bit confused about this as well.
When you look at TLC in Rawr the trinket is actually not all that great but with Magegraf it's the best thing in the game by a long shot.
Yeah, I noticed that as well. Rawr and Magegraf seem to have very different opinions about different trinkets.
From what I can tell, it may be due to different spell rotation assumptions. Rawr seems to build in pretty static rotation assumptions, and Magegraf allows for more "on the fly" switching between frostbolt and AB, with a lot more AB.
Hey there, just wondering, im trying to run some results on comparing deep fire to 40/10/11 but im lost in words on the Vontre's spreadsheet, when calculating for intellect do i need to divide/multiply my intellect stats by 1.25 (because of MM) or can i simply just input whatever my stats are including the 25% increase in intellect from MM and just simply imput my talent points and the spreadsheet calculates for me?
Also a similar question with spellfire, do i need to subtract 7% of total intellexxt added to spell damage from my total spell damage or leave it as it is and simply mention i have the spellfire bonus?
Sorry if i have been somewhat confusinf but this has been bugging me for ages as this may/may not make 40/10/11 a higher dps spec with current gear (with scorch debuff up).
Every fight is a food fight when you're a cannibal.
~~Demitri Martin
hi would you be updating the Band Of Eternity's different rep level stats and also Footpads of Madness doesnt have haste rating counted in on the spreadsheet
The increased interest in arcane frost, and the difficulty of modelling the damage of build, I suspect your application is going to get a lot of attention, as its the best method of modelling the build that I have seen so far.
Wonderful job with a very useful app.
A couple suggestions, take them as you willl:
1) It would be handy if the results could show the marginal value of stat increases. 10 more damage gets you 7 more dps, 10 more Int gets you 8 more dps (an oddball result I got from your site that I would love to confirm), etc. Values for damage, hit, crit, haste, int, and spirit would all be valuable.
2) The spell selection model for this spec seems counterintuitive frequently, allowing the AB debuff to fall off, and not fully stacking it back on before switching to frostbolts and letting the debuff fall again. Is it possible for the cast sequence selection heuristics to be more transparent, so we can understand when it switches to different cast sequences and why? I can believe that the model has chosen the dps-maximizing spell selections, but if I can't understand the methods it uses to generate those selection, I can't replicate the results.
I've looked at magegraf again to see if there was any progress made since I last checked and to compare results with Rawr.Mage and I see the same problems as before. From arcane specs the arcane/frost seems to be in the best shape, although it's running below 0 mana for a long time. The other arcane specs however seem to be very inefficient with mana consumables, some for example don't even use evocation.
The best I can do on short notice is open the source code. I'll follow up here in a few minutes.
Oh by the way, spreadsheet 3.0 is coming. I've significantly trimmed the fat with some interface redesign and am potentially introducing a new plugin system which will allow others to publish their own gear files and distribute separately from the theorycraft sheet itself.
2) The spell selection model for this spec seems counterintuitive frequently, allowing the AB debuff to fall off, and not fully stacking it back on before switching to frostbolts and letting the debuff fall again. Is it possible for the cast sequence selection heuristics to be more transparent, so we can understand when it switches to different cast sequences and why? I can believe that the model has chosen the dps-maximizing spell selections, but if I can't understand the methods it uses to generate those selection, I can't replicate the results.
You are looking for the method "arcane3" for arcane/ice builds.
I found one strange thing on first glance. Maybe I just don't understand everything yet, but it seems like there is a bug in mana_left function. From the use of it it seems it's supposed to return how much mana is still available till the end of fight. What is strange is you compute how many times a certain effect can still be used, but then you never use $uses, you just add one ammount. Maybe it should be $mana_left += $uses * $manaamt; ?
I found one strange thing on first glance. Maybe I just don't understand everything yet, but it seems like there is a bug in mana_left function. From the use of it it seems it's supposed to return how much mana is still available till the end of fight. What is strange is you compute how many times a certain effect can still be used, but then you never use $uses, you just add one ammount. Maybe it should be $mana_left += $uses * $manaamt; ?
Nice catch, I'll take a look at the functionality.
I updated the mana calculations on magegraf again. These appear to be working considerably better. Link me any problems or oddities.
Edit: These fixes seems to put Arcane/Ice squarely above Arcane/Fire and Arcane/Arcane. Not surprising. Actually looks like arcane/ice is a little above fire now (except for the tiny detail of keeping 2 pieces of T5).
I'll take another look at the code and see if I find something else. Also a big issue with pure arcane builds is that you never let the AB debuff drop. I don't see how a rotationless approach with pure arcane would give good results. While that is not that important in Arcane/Frost, longer ABs have a much better effect paierd with AM as it plays with OO5SR mana regen. Which function is used for 48/0/13?
arcane3 is used for all the arcane specs currently. arcane1 and arcane2 are deprecated.
I updated the logic again. The problem was the AB/AM cycle wasn't actually burning all the available mana, so I changed some things specific to AB/AM that hopefully improve mana usage for that spec.
[e] arcane2 is the traditional rotation logic, however I disabled it a while back due to uncertainty on how lag affected the AB ramp up. Additionally the AB spam logic for arcane3 was giving results that were just as good if not better. There is a lot working against rotations and in the end I decided I didn't want to risk misleading people by including them in the main calculations.
Regarding the actual calculations in arcane2; currently it is not scripted to burn excess mana, for more normal gear sets (550 int, 300 sp, 100 haste) this wasn't a problem as the rotation itself was burning about the right amount. I don't feel like putting in the effort to update arcane2 and test it as I doubt the results would be any better. It makes sense, casting scorches and unramped ABs is not at all ideal, you give up a lot to get that overlap bonus.
Any plans to work Sorcerer's Alchemist Stone into the spreadsheet?
Personally I'm pretty curious to see how it compares to DC:C when taking into account that there are many fights where the buff will drop, either by the nature of the encounter or occasional player error. I currently don't have access to Malacrass' trink (the guild as a whole has never really put much time into ZA) or Illidan's trink, so right now I'm stuck with Icon and DC:C.
Any plans to work Sorcerer's Alchemist Stone into the spreadsheet?
Personally I'm pretty curious to see how it compares to DC:C when taking into account that there are many fights where the buff will drop, either by the nature of the encounter or occasional player error. I currently don't have access to Malacrass' trink (the guild as a whole has never really put much time into ZA) or Illidan's trink, so right now I'm stuck with Icon and DC:C.
If only the 40% increase effect on potions worked for destruction potions :p As far as DC:C, it stacks so fast with imp scorch that downtime isn't a huge factor for most situations.
If only the 40% increase effect on potions worked for destruction potions :p As far as DC:C, it stacks so fast with imp scorch that downtime isn't a huge factor for most situations.
True, but if it fades when there's still a decent amount of time left on the scorch debuff (~15s?), I wouldn't imagine it would be best practice to fire off a few scorches to help ramp the card up. With fireballs, it'd probably take 3 to get the DC:C buff back in full force, and by then, that's three fireballs at a fairly big drop in damage (granted each is more powerful than the previous, but they're all still <+80)
arcane2 is the traditional rotation logic, however I disabled it a while back due to uncertainty on how lag affected the AB ramp up. Additionally the AB spam logic for arcane3 was giving results that were just as good if not better. There is a lot working against rotations and in the end I decided I didn't want to risk misleading people by including them in the main calculations.
Regarding the actual calculations in arcane2; currently it is not scripted to burn excess mana, for more normal gear sets (550 int, 300 sp, 100 haste) this wasn't a problem as the rotation itself was burning about the right amount. I don't feel like putting in the effort to update arcane2 and test it as I doubt the results would be any better. It makes sense, casting scorches and unramped ABs is not at all ideal, you give up a lot to get that overlap bonus.
You're right, the difference is not that big. Doing rotation-less sequence for Arcane/Frost you lose less than 1%, for deep Arcane/IV about 2%. I guess the remaining differences are explained by what cooldowns are stacked with which spells.
03/25/08, 1:33 AM
Similar Threads
