Vontre's Mage DPS Spreadsheet - Page 31

• zeidrich · 02/18/08, 6:31 PM

Can someone explain what exactly the "Build Comparison" is showing me? I've searched the thread, and from people's comments it sounds like some say it should be showing the ideal cycle for whatever talent spec that was chosen.

Looking at the calculations though, it makes talent assumptions. So, say I put an Arcane/Frost spec into the talent URL, and then go and look at the calculations tab, I quickly see that the DPS totals that populate the "Fire" column include improved Fireball as the cast speed is half a second faster than the Fireball cast time that is used to calculate the Frost values.

So then, I expect that it is showing me damage output based on my stats for 3 different "standard builds" of fire/frost/arcane. But then why do those values change when I put in different specs?

I guess in short my question is: I download the spreadsheet as it is with the default stats. Why is it that when I put in a 40/0/21 arcane frost spec it shows 1456 DPS for fire, and 1430 dps for frost, and 1377 dps for arcane, despite showing 841 dps for fireball spam?

And when I put in a 2/48/11 fire spec it shows a 1413, 1383, and 1312 respectively. As well, it says my fireball does 1432 DPS, and would do a total of 429547 damage, despite the fact that the fire column in the build comparison area shows 424009 damage max.

I tried to follow the calculations in the calculations page, but they just got far too complex for me to reverse engineer.

I would expect that a fire build should do better than an arcane/frost build. Or is it just that it is showing me totals for a "typical fire build" using my stats, and those stats are being artificially inflated due to talents in the build like arcane mind and arcane meditation? In which case the only viable comparison would be to put in a 0/0/0 build?

Or is excel just screwing up and not tallying the formulas correctly?

Akuman · 02/19/08, 5:06 AM

Your excel is not screwing up, I get the same results.

It says my fireball does 1432 DPS, and would do a total of 429547 damage, despite the fact that the fire column in the build comparison area shows 424009 damage max

That's probably because scorch is integrated into the cycle.

Maddmage · 02/19/08, 2:03 PM

Originally Posted by Vontre

If you already have the winter's chill talent it's assumed always on.

Oh, I see, thanks for clearing that up.

Rustyshrapnel · 02/21/08, 10:51 PM

I was curious about optimizing my damage based on my current gearset (IE, what spec is best for me given what I have to work with), so I started running some numbers through Vontre's spreadsheet. I got some stuff I expected as well as some stuff that surprised me, and I was wondering if I was perhaps doing something wrong.

Not surprisingly, deep fire beat deep frost. The discrepancy I kept getting was that arcane was still showing very well, so well in fact that it seemed to beat deep frost and deep fire every time even if I was modeling a deep fire or deep frost spec. This simply cannot be.

Reading through this thread, I'm assuming the "Build Comparison" is showing the max dps and damage for the given stats, using an ideal rotation for a cookie-cutter deep fire, deep frost, and arcane build. Under that assumption, it should be showing Fire as the "max" build for a 2/48/11 spec, since that's currently the highest dps raid spec modeled.

Since I'm deep frost at the moment, I used the Gear page to get my unbuffed stats for an ideal Fire build given my available gear (capping my hit, scooting around some gems, etc.). I then input these stats into the Manual Entry on the Main Sheet:

Int: 444
Spirit: 305
Spell dmg (all schools): 1108
Crit Rating: 309
Hit Rating: 164
Mp5: 0
Fire Dmg: 51
Frost Dmg: 0
Arcane Dmg: 0
Haste: 25

Trinket Effects: Serpent-Coil Braid, Icon of the Silver Crescent

For Buffs/Debuffs I went with the default settings of molten armor, 200 vamp touch, 100% JoW, 10% COS and COE, no totems, and no bloodlust. Our guild is low on shamans so I don't usually get totems or bloodlust. Boss Settings were also default, as were Consumables. Misc was set to always Imp Scorch, otherwise nothing was changed.

The Build Comparisons ran thusly: Fire: 1604 dps, 481,091 dmg; Frost: 1530 dps, 458,997 dmg; Arcane: 1622 dps, 486,465 dmg.

So what it's saying here is if I spec 2/48/11 with my current stats, I'm actually better off going with an arcane build? The first thought was well, I'm not really optimizing the fire spec with the way I've got the conditions set, so I added in a totem and some bloodlust time (10%, is that a reasonable number? Like I said, I know very little about modeling bloodlust). Arcane still came out on top, granted not by much (1739 dps as opposed to 1733 dps). Still, definitely not the "fire trounces arcane" I was expecting.

The only thing I can think of is I have the Tirisfal 2-set bonus but not any other set bonus like Tirisfal 4 or Tempest 4. Is the set bonus really that good that it automatically makes arcane the best, or am I interpreting the Build Comparison feature improperly?

Baruk · 02/23/08, 5:16 AM

answered it myself :/

Cryic · 02/23/08, 11:48 AM

Vontre: I think something is still buggy with the magegraf simulaor for the TLC. When I switch from the Darkmoon trinket to TLC the dps jumps from 2,225dps to 2,341dps. A gain of 116 dps over the Darkmoon.

From:
http://www.magegraf.com/index.php?ha...55b18861e938fd
to:
http://www.magegraf.com/index.php?ha...4e62fd80454ddc

xyu · 02/26/08, 11:57 AM

Has anyone ever come up with the best possible DPS gear setup with this spreadsheet? I'm kind of curious. I know the hunters did it on a thread, i'm wondering if we can make one, or if it's already been discussed.

hypetech · 02/27/08, 4:53 PM

Is there anyway to fix the Band of Eternity listings in the spreadsheet?

Vrangsinn · 02/28/08, 5:23 AM

Originally Posted by xyu

Has anyone ever come up with the best possible DPS gear setup with this spreadsheet? I'm kind of curious. I know the hunters did it on a thread, i'm wondering if we can make one, or if it's already been discussed.

I'm not 100% sure, and it ofc depends if all calcs in the sheet are correct, but on T6-level(MH/BT) pre Sunwell:

I have ele shammy in my group, so can remove some hit for other stats, so I ended up with:

4-set T6: Head, Shoulders, Chest, Gloves. All with 12-dmg gems, except I've been going with Chaotic Skyfire Diamond, and so need two pieces in my gear with blue gems.
Put those in gear where you only need one blue and one yellow and get the bonuses with most +dmg and hit+dmg in the yellow sockets.
I put the blue+yellow in belt and shoulders.

Neck: Hellfire Encased Pendant
Back: Cloak of the Illidari Council
Wrists: Bracers of Nimble Thoughts
Waist: Anetheron's Noose
Legs: Leggings of Channeled Elements
Feet: Slippers of the Seacaller
Rings: 2x Ring of Ancient Knowledge
Trinkets: Skull of Gul'dan + Hex Shrunken Head
Weapon: Zhar'doom(yes, according to the sheet, the haste makes up for the loss in dmg if you have the +hit)
Wand: Wand of the Forgotten Star

This gives 124 hit, so if you don't have an ele shammy, you need to change some stuff around. Weapons are probably the best choice for starters(Chronicle + Tempest)

Note though, I've been off raiding for a while, and on my way back now, so lots of things might have changed. And it's all a waste soon anyway with SW coming up.

edit:
About the Band of Eternity:
It's easiest to just go into the gear section, change the first listing of the band to the stats of the mage ring of your current rep, and pick the top one in the drop down list. Personally I just deleted all the rows except one and updated it as my rep increased.

Soraxis · 03/13/08, 3:41 PM

Is there a way to show what the actual calculations are that you did to figure out the DPS for different spells? Im most interested in fireball DPS...working on my own version of that right now and I want to compare how we did things. Right now Im looking at something like this (copy pasted from excel which Im a noob at :P)

(1165+1.43*C4)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100)))

A4=Crit chance (Decimal)
B4=Spell resist (should be 1 if hit capped)
C4=Spell damage

I talked with Manly about this yesterday, he pointed out a few things that I have yet to account for but am aware of now.

A) Still need to include spell haste
B) I need to include simple procs like the spellstrike set proc*
C) Trinkets/Cooldowns/Gems
D) Advanced Procs like the Sextant of Unstable currents (cant proc more than once every 45 seconds) **
E) He talked a bit about how mages can still get rolling ignites going and I've read a bit about it here, I dont fully understand all of the mechanics of what happens there yet so Im not going to try to include that in what I've written here until I do. I dont think I'd need to erase my final section of the equation to encompass that, it would probably be just another addition based on the spell crit squared and a portion of the ignite damage.

I've got some of the work figured out for the stuff Im missing but its not completely finished yet so Im just posting them below here in some notes so you can see what Ive got so far on the calculations. The only barriers I really see coming in my way are the rolling ignites and calculating more advanced cooldowns like combustion/icy veins as well as different equations for different methods of popping cooldowns. Ie. popping them all as they come up and not trying to merge cooldowns together even though they are on different timers vs. trying to make sure that all cooldowns are popped together.

*Basic procs like this are pretty easy to calculate, using my original template, I'd use something like this (I dont know proc rates here so Im going to pick a random number like 5%) I dont want to include this in my final equation yet because I dont have exact numbers on the proc chances of different trinkets and such, but with the theory down it isnt really that hard to just use different percentages and spell damage bonuses as well as simply adding more spellcasts if the proc is a longer one. If there are two procs being put into play at the same time I would want to simply double the equation and have them run in seperate areas, I dont think many people would actually hold a cooldown of combustion or icy veins for a time when there are two procs up at one time so for the most part what would be required still is a calculation to determine the chance of proc while the mob/boss is at 20% and then just add the molten fury buff in there.

50*(1165+1.43*C4+92)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100)))+
47.5*(1165+1.43*C4+92)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100))+
45*(1165+1.43*C4+92)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100))+
42.5*(1165+1.43*C4+92)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100))+
815*(1165+1.43*C4+92)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100))

For this one, where I got my numbers from are that per 1000 fireballs, you will see 50 procs, that is a guaranteed chance on each first spellcast that you will get the spell damage bonus. On the second spellcast, you have a 5% chance that the first one actually overwrote your second one, so thats still a 95% chance that the original proc is still active (if it is a second proc that would be included again in the first line of the equation). The third fireball you'd have a 90% chance because both the first and second fireball might have overwritten your proc, fourth one there is a 85% chance. The last line is the remaining amount of fireballs cast without procs. One thing to note here that I am not entirely sure about, I have included the fourth fireball in the equation even though the proc lasts for only 10 seconds, reason I have done this is because I have read that fireball damage is calculated when the spell leaves your hand instead of hitting the mob, but I believe the proc becomes active as the spell hits the mob, so if you are far away enough where your travel time is greater than 2 seconds, you should be able to get the fourth fireball off if you are spamming it.

If you had enough spell haste here you could also get more fireball casts in there, Im not at that stage in my equation so Im going to ignore that for now and adjust as that occurs.

**On a side note, I do believe I have determined a way to accurately include a proc such as sextant of unstable currents in my equation while still accounting for the 45 seconds of unproccable time. I dont have any numbers yet on the proc rate for the sextant, so I'm going to make up my own for now and say its a 2% proc rate just so I can show what my calculations would look like.

5*(1165+1.43*C4+190)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100))
60*(1165+1.43*C4)*(1*((100-B4)/100)+1.1*A4*((100-B4)/100))

Basically the concept is that if you have the sextant with a 2% proc rate, you're going to see it proc on average once every 50 spellcasts and then have a 45 second downtime of no proccing, so thats 50 spellcasts trying to get the proc and then another 10 waiting for the proc cooldown to wear off, the other 5 is of course for when the proc is actually up and doing something.

That being said, I know that the main problem calculating this is the fact that sometimes you will see a proc right after the cooldown is up and sometimes you wont see it for ages, in the long run that shouldnt affect the equation, if you take this and put it on a large scale (1000 fireballs for example), the numbers would add up pretty much to what is shown here with some leeway shown on both sides for some of the random factor of procs still in there.

Looking back on this, I see I havent accounted for spell resists interfering with how often the procs land even though I do have them affect the DPS on the fireballs, Im not going to adjust this yet but what Im thinking I'd want to do is have an adjusted proc rate for the trinket and go from there, instead of saying 2% I'd say its 1.98% (99% of 2%) and use that instead.

galzohar · 03/13/08, 4:36 PM

Here's something to make your life easier (and I'm sure it was mentioned many times before but it's pretty hard to look it up and pretty easy to miss it as it's probably burried in the threads):

U=uptime
P=proc chance
T=cast time

chance for a proc to be up is 1 - (chance for proce to not be up).

Chance for proc to not be up is chance for all spells casted during the last U to have not procced, which is approximated very well (due to small variance of U/T) by:
(1-P)^(U/T)
Therefore the chance for a proc to be up is:
1 - (1-P)^(U/T)
which is equal to the uptime %.
This approximation is only good if (while fighting) U/T is approximately constant. It gets inaccurate if you stop casting very very often, but works great if you cast most of the time and incorporate very short interruptions into your "effective" cast time.

For cooldown procs, since they don't overlap, it gets even easier. What you're looking for is how much time on average you need to wait between procs.
C = cooldown
Procs/second = P/T
seconds/proc = T/P = time between procs with no cooldown
Since a proc triggers a cooldown during which you cannot proc, you will start counting the average time until the next proc only once the cooldown is over, and therefore the average time between procs with a cooldown will be:
seconds/proc = T/P + C
procs/second = 1 / (T/P + C)
Since procs never overlap on cooldown items:
uptime% = uptime * procs/second = U / (T/P + C)
This approximation is only valid when the "no casting time" in fights is either neglicible or is at least much smaller than the time spent casting and incorporated into an "effective" cast time. This is true for most fights (as in, having high DPS uptime compared to downtime).

About cooldown pairing along with procs, as long as you don't actively do anything to combine cooldowns with procs, you will on average gain from stacking them just as much as you lose from not stacking them. For example if I have a 10% damage proc up 10% of the time it will increase all my DPS by 1% as well as the DPS gained from a +spell damage proc. I do not know of any proc items that are actually worth waiting for before you use all your cooldowns, although I suppose you could possibly have some minimal gains if you know you won't be able to use your cooldowns again during the fight but know you have time to wait for the proc and still use your cooldowns and gain more from them with the proc.

Secondary effects are usually quite incredibly small (although you may want to include them for completion especially when it comes to cooldown stacking). 1% spell resist for example will turn a 20% proc on spell hit into a 19.8% proc chance, so if that proc gave effectively 100 spell damage your model "overestimates" its benefits by less than 0.4 spell damage.

Remember if you're not intentionally pairing effects, averaging their benefits is fine. That's why the gains from stuff like icon of the silver cresent would be at minimum damage*duration/uptime because that's what you would gain, on average, if you mindlessly hit it every time it's up - any smarter use and you could gain more out of it. Not to mention that if the fight isn't a multiple of 2 minutes (from first activation) you gain more out of it even before adding the "playing smart" factor.

psbassplayer · 03/14/08, 5:04 AM

i'm very tired right now, and cannot read everything so i don't know if someone posted this already, but is there a version of this that includes the 2.4 gear? trying to figure out what badge loot i want to get first, and i'd rather not run the numbers through myself. if there isn't can you guide me to one please? thank you in advance.

Monolih · 03/14/08, 11:48 AM

I could not find a option for Misery so im wondering if its missing or if its accounted for anyway in some other option?

Docjowles · 03/14/08, 1:04 PM

Vontre doesn't really seem to actively maintain the spreadsheet anymore; his attention is focused on MageGraf.com. I'd recommend folks give that a whirl, along with Rawr.Mage.

Vontre · 03/14/08, 2:10 PM

I still maintain the spreadsheet but adding in new gear is so incredibly tedious. There's not really anything else to be done on it.

Misery is always counted, there is no option for it unless you modify the Game Stats tab directly.