New Affliction DPS Spreadsheet! - Page 2

fallenman · 12/14/08, 10:03 AM

Originally Posted by Jamuka1

Unfortunately, you're math is off.

For a spec as complicated as affliction, spell hit can't be calculated with your simple (finaldps = dps x (.83 + hitchance)) calculation. That works for sac/destro shadowbolt spam, but not this spec.

For instance, if CoA misses, I dont lose CoA's dps for the next 28 seconds. I just cast it again. All I lose is 3/5ths of a shadowbolt and CoA's dps for the GCD.
Likewise, if haunt misses, I dont just lose the haunt damage, I lose 20% dot damage for the next 10 seconds.

Since the value of hit vs spell power is just about the most relevant piece of information an afflic lock is gonna get from a spread sheet, it really needs to be modeled accurately for the spreadsheet to be of any value.

I think this kind of goes to further add to the point that we should be working on getting any issues with Leulier's resolved, since that spreadsheet is much further along. Efforts are appreciated, but there's no need for everyone to be doing twice the work when we already have a spreadsheet that is working correctly for the most part.

Vux · 12/15/08, 1:00 AM

Originally Posted by Jamuka1

Unfortunately, you're math is off.

For a spec as complicated as affliction, spell hit can't be calculated with your simple (finaldps = dps x (.83 + hitchance)) calculation. That works for sac/destro shadowbolt spam, but not this spec.

For instance, if CoA misses, I dont lose CoA's dps for the next 28 seconds. I just cast it again. All I lose is 3/5ths of a shadowbolt and CoA's dps for the GCD.
Likewise, if haunt misses, I dont just lose the haunt damage, I lose 20% dot damage for the next 10 seconds.

Since the value of hit vs spell power is just about the most relevant piece of information an afflic lock is gonna get from a spread sheet, it really needs to be modeled accurately for the spreadsheet to be of any value.

That would only make the value of hit more valuable than it already is. And hit is already the highest valued stat on the spreadsheet. This is why I was reluctant to add the hit calculation in the first place; we all know the important of hit. The main focus is the difference between Crit, Haste, Spell Power, and Spirit.

Originally Posted by fallenman

I think this kind of goes to further add to the point that we should be working on getting any issues with Leulier's resolved, since that spreadsheet is much further along. Efforts are appreciated, but there's no need for everyone to be doing twice the work when we already have a spreadsheet that is working correctly for the most part.

Please, enough of these worthless posts. They contribute nothing of value.

dagee · 12/15/08, 3:03 AM

Vux definitely like the spreadsheet especially since I've seen your initial post in regards in regards to dpct with affliction. Two questions for you.

1) When I changed the time value (something I know we are really not supposed to do) the number really get messed up. So for example. I plug in 300 sec for fight duration and the weight of the various stats goes into negative numbers.

Now it could be that your spread sheet is set up currently so that time cannot be changed but time really should be a variable that the user can change simply because the value of the different specs will obviously change based on duration of fight (haste's value will go go down).

2) It looks to me that the number of cast's is based solely on the cast time divided into fight time. Thus during a fight it is possible to cast a noninteger value. I don't know excel that well but I am assuming there is a way for mat your cells so that they create a step integer function (For example 17/6 would equal 2) thus mimicking the fact that you can only have whole number casts.

Again, great spread sheet. Keep up the great work.

Vux · 12/15/08, 3:39 AM

Originally Posted by dagee

Vux definitely like the spreadsheet especially since I've seen your initial post in regards in regards to dpct with affliction. Two questions for you.

1) When I changed the time value (something I know we are really not supposed to do) the number really get messed up. So for example. I plug in 300 sec for fight duration and the weight of the various stats goes into negative numbers.

Now it could be that your spread sheet is set up currently so that time cannot be changed but time really should be a variable that the user can change simply because the value of the different specs will obviously change based on duration of fight (haste's value will go go down).

Ah yes! That happens since the numbers for Sheet 2 are still for the old length of time. So it's trying to calculate the DPS differences based on different length fights. If you change the fight times on Sheet 2 as well, you'll see the calculations get back on track. : ) You can also just change that little number there for all the Sheet 2 time lengths to feed off Sheet 1 automatically. (I'll add this next version!)

2) It looks to me that the number of cast's is based solely on the cast time divided into fight time. Thus during a fight it is possible to cast a noninteger value. I don't know excel that well but I am assuming there is a way for mat your cells so that they create a step integer function (For example 17/6 would equal 2) thus mimicking the fact that you can only have whole number casts.

Again, great spread sheet. Keep up the great work.

For all the dots, it's actually doing the calculations based on having your dots ticking throughout the fight. For Shadow Bolt and Drain Soul, yes, the casts for those are calculated backwards and will therefore have uneven number of casts. However, over a long period of fight time, the DPS that it's calculating based on these number of casts is still a great model for giving information for the purposes of comparing the benefit of different stats. I'm not sure clipping or truncating SB/Drain Soul casts would increase the accuracy of these comparisons.

PSGarak · 12/15/08, 5:13 PM

Originally Posted by Vux

That would only make the value of hit more valuable than it already is. And hit is already the highest valued stat on the spreadsheet. This is why I was reluctant to add the hit calculation in the first place; we all know the important of hit. The main focus is the difference between Crit, Haste, Spell Power, and Spirit.

Incorrect, this actually lowers the value of hit. It certainly can't raise it, because otherwise you wouldn't play in that manner. This decrease is most of the reason why we suspect hit may not be as valuable for affliction spec. The effect of hit on DoT uptime on DoT damage is actually reduced to a factor of about GCD/duration than what you're calculating, at the cost of some filler, which is inherently lower DPS. Again, we're interested in the conversion ratio because many of us believe that the value of hit is more contentious than you do, and I personally don't like dogmatic "of course it's better" answers without verification. (FYI Hit was not by best stat in early TBC either)

The general approximation is that the DoT does the same damage, but for a hit chance of miss chance of x, the cast time is GCD/(1-x) and the duration is extended by GCD(1/(1-x) -1). The 1/(1-x) term is the solution to a geometric series representing the expectation of successive misses. This is an approximation because it ignores the increase of DoT collision when recasting, and any other effect it might have on your cycle.

32103940 · 12/16/08, 9:55 AM

any affliction spell that misses is promptly reapplied, increasing its duration by one cooldown without increasing its damage proportionately. if a 30s dot like SL gets resisted, you recast and it ticks to t=31.5, resulting in 4.76% dps loss. and that is for 50% affliction miss rate. for shorter dots like UA, 50% miss rate results in 9.09% dps loss. 50% miss rate on the good old 21/40 sb spam would result in 50% dps loss.

im sure you see how the very fundamentals of dot spells have innate resistance to dps loss from spell misses. couple that with the superior sp dmg scaling compared to direct dmg spells, and you have a build/rotation that values spell damage over hit at all levels of spell damage.

nikitabanana · 12/16/08, 2:36 PM

Originally Posted by Vux

That would only make the value of hit more valuable than it already is. And hit is already the highest valued stat on the spreadsheet. This is why I was reluctant to add the hit calculation in the first place; we all know the important of hit. The main focus is the difference between Crit, Haste, Spell Power, and Spirit.

Please, enough of these worthless posts. They contribute nothing of value.

Just to let you know, (I can link yet more math/charts/etc which I have done repeatedly in the beta forums to support various things), Talent Calculator - World of Warcraft is not the optimal raid build assuming you are @ 14% hit.

It's actually a rather poorly constructed, silly build. I'm referring to post 6 from World of Warcraft - English (NA) Forums -> [3.0/WotLK] Vux's Guide to Affliction

Talent Calculator - World of Warcraft would be optimal for max single target dps assuming you are at 14% hit. If you are above that hit level you can probably either regear to optimize (which is recommended since none of the lower tier talents the above build has available will up dps).

That would only make the value of hit more valuable than it already is. And hit is already the highest valued stat on the spreadsheet. This is why I was reluctant to add the hit calculation in the first place; we all know the important of hit. The main focus is the difference between Crit, Haste, Spell Power, and Spirit.

Hit has diminishing returns when you start looking at it in regards to destruction spells in an affliction build. I wouldn't, for instance, go voluntarily past 14% hit for destro (aka not speccing cataclysm) in an affliction build.

fallenman · 12/16/08, 3:54 PM

Originally Posted by nikitabanana

Talent Calculator - World of Warcraft would be optimal for max single target dps assuming you are at 14% hit. If you are above that hit level you can probably either regear to optimize (which is recommended since none of the lower tier talents the above build has available will up dps).

Good points about those builds. But I do have 2 questions about yours quick.

I would actually consider fel concentration a higher dps increase considering spell pushback, than the 6% mana savings from suppression. On your average fight, suppression will save you one lifetap (so, 1 GCD) at best. Yes/no?

Also, with points 2 and 3 filled in eradication, which only give another 1.6% increase in spell haste, I would think that another 2 points in molten core would be better. It would increase the MC uptime from 39% to 78%, almost doubling the DPS increase gain on immolate. With your average afflock having immolate be around 9% of their DPS, I could see that being a greater increase over an additional 1.6% haste. But again, your thoughts?

nikitabanana · 12/16/08, 4:23 PM

Originally Posted by fallenman

Good points about those builds. But I do have 2 questions about yours quick.

I would actually consider fel concentration a higher dps increase considering spell pushback, than the 6% mana savings from suppression. On your average fight, suppression will save you one lifetap (so, 1 GCD) at best. Yes/no?

Also, with points 2 and 3 filled in eradication, which only give another 1.6% increase in spell haste, I would think that another 2 points in molten core would be better. It would increase the MC uptime from 39% to 78%, almost doubling the DPS increase gain on immolate. With your average afflock having immolate be around 9% of their DPS, I could see that being a greater increase over an additional 1.6% haste. But again, your thoughts?

Well, right now I'm not exactly getting spell pushback on any fight. I can't think of a single fight I actually am right now.

I did the eradication vs. molten core iterations and found that eradication slightly edged out molten core for the allocation of those two points. I'm going to do more iterations of those two with a few new assumptions just to take care of a couple of edge cases.

fallenman · 12/16/08, 4:45 PM

Originally Posted by nikitabanana

Well, right now I'm not exactly getting spell pushback on any fight. I can't think of a single fight I actually am right now.

Maybe I'm coming up with more pushback in my mind than I'm actually suffering from. I'll have to make a better note this week on it.

I did the eradication vs. molten core iterations and found that eradication slightly edged out molten core for the allocation of those two points. I'm going to do more iterations of those two with a few new assumptions just to take care of a couple of edge cases.

Thanks for the info. I look forward to seeing what you come up if you do any further checking between the two. If you don't mind me asking, do you recall how much difference there was between the two? Also, in your iterations, I presume you are dropping immolate from the rotation once the boss is below 35%?

nikitabanana · 12/16/08, 4:50 PM

Originally Posted by fallenman

Maybe I'm coming up with more pushback in my mind than I'm actually suffering from. I'll have to make a better note this week on it.

Thanks for the info. I look forward to seeing what you come up if you do any further checking between the two. If you don't mind me asking, do you recall how much difference there was between the two? Also, in your iterations, I presume you are dropping immolate from the rotation once the boss is below 35%?

I can't recall the differences - have to look it up later tonight when I get home. I believe I'm dropping immolate sub 30%. I can check that as well.

GorSin · 12/17/08, 5:16 AM

First of all, I want to thank you for a nice and simple tools you made.

Secondly, there are few suggestion that you can made in order to improve this sheet.

Actually, it's not that difficult to precise impact of +hit on dps.

1. Change time spent casting for COA, UA, Corruption, SL, Immolate and SB.

 TimeSpentCasting = NumberOfCast  / ( 0.83 + Hit )

2. Haunt.
There are 2 major impacts of Haunt's misses: lossing 20% of dots damage and necessity to reapply Corruption.
In case of not 100% of hit, Haunt up time will be equal your hit chance and Haunt multiplier to dot's damage will be:

 HauntMultiplier = 1.20 * ( 0.83 + Hit )

Keeping in mind that we have to reapply Corruption 2 times for every 3 Haunt misses, Corruption's number of casts will be

 CorruptionNumberOfCasts = 1 + HauntNumberOfCasts * ( 0.17 - Hit ) * 10 / 15

3. Soul Siphon is different. As we only loose GCD per every SS miss ans it's channeled spell, the best way, imho, to increase casting time for each spell on actual GCD fraction

[code] SoulSiphonCastingTime = 15 * ( 1 - Haste ) + Max( 1, ( 1.5 * ( 1 - Haste ) ) ) * (0.17 - Hit)

4. Also, please change Formula for # of NightFall procs. It must be C9 / 3 * 0.008 instead of 200 * 0.08

Molx · 12/29/08, 11:28 PM

Regarding the Fight Time issue...

First off, I adjusted the second sheet to use the user value from the first sheet and I noticed no change in the following problem I have with your calculations...

The lower the time frame, the higher one's DPS is..?!?

This becomes truely evident with times under a minute... I am pretty sure that the purpose of the spreadsheet is not to gauge one's performance on trash mobs, but it does seem to reveal an underlying issue with the formulas...

And while I wish I could give a possible correction or improvement, I cannot, and for that I appologize...

valheran · 12/30/08, 5:00 AM

Ofc it is.

The shorter the fight, the more % of it is in "powered mode" that produces increased DPS - that means while under trinket uses/ proccs, and most important, heroism.