Jump to content


Photo

Mage DPS Spreadsheet


  • Please log in to reply
32 replies to this topic

#1 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11580 posts

Posted 14 September 2006 - 11:26 PM

I don't have time for much description, at the moment.

http://www.sigilguil...hamlet/Mage.xls

The first iteration is here; you can probably extrapolate most of what's going on.
http://forums.elitis...pic.php?id=6825

All this business about "cycles" is elucidated (maybe) here (post 14):
http://forums.elitis...pic.php?id=7732
and here:
http://forums.elitis...pic.php?id=7742

---------------

It's not at all polished or user-friendly.

Basically, you edit anything that's shaded on the first page. There are two parallel columns for you to run a comparison.

Cycles are "casting modes," described by what you do on a Clearcast or non-Clearcast. The one thing I'm currently missing is a way to automatically choose which two cycles (read my links; it's always two) are best for a given fight. So you can input, for each setup, four pairs of cycles, for which it will automatically choose the best. You have to look at pages 2 and 3 to see what the cycle numbers mean, in terms of spells.

A good way to determine your effective casting delay:
See how many Polymorphs you can chaincast on a critter in one minute.
Casting lag = (60 - 1.5*N)/N

---------------

Not modelled:

Cooldowns.
Rolling Ignites.
Cycles in which you use Fireblast every cooldown.
8/8 NW.

#2 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11580 posts

Posted 15 September 2006 - 01:07 PM

Its best use is to compute equal-DPS contours of various stats at your current level of gear. For example, for me, at Patchwerk-length fight, roughly:

1 crit = 1.3 hit = 15 Fire dmg = 11 MP5 = 20 Spi = 28 Int

A point in Arcane Med is worth about as much as a point in Master of Elements, which are both worth about half as much as a point in Elem. Precision. The most important point-for-point talents is Ignite, followed by Fire Power and Imp. Fireball, although it's hard to evaluate Imp. Scorch. Most points in raid Fire are pretty clear-cut.

#3 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 02:07 PM

A point in Arcane Med is worth about as much as a point in Master of Elements.

The truth or falsehood of this statement lies entirely on your crit rate and your spirit value. When I did the Math for my admittedly max-out-damage build, a point in MoE is worth about 1.5 times a point in Arcane Med (for fireball).

#4 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 02:16 PM

11 MP5 = 20 Spi

I also don't see how you could possibly make this statement, unless your Patchwerk attempt are 4 minutes and 39 seconds long and you have 3/3 Arcane Med.

Only under those conditions would the above be equal - if you had anything less then 3/3 Arcane Med and if the fight takes longer than 4:39, 11 mp5 completely dwarfs 20 spirit in practice.

#5 frmorrison

frmorrison

    Protector

  • Allied Members
  • 11427 posts

Posted 15 September 2006 - 02:18 PM

Depending on current gear, talent's power will change for folks, that is the point of creating a spreadsheet.

Yes, if you had more crit than Arawthion has then MoE should be better than Arcane Med, but would it be the same for another Mage? I don't know, that is why spreadsheets are around.

#6 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11580 posts

Posted 15 September 2006 - 02:34 PM

11 MP5 = 20 Spi

I also don't see how you could possibly make this statement, unless your Patchwerk attempt are 4 minutes and 39 seconds long and you have 3/3 Arcane Med.

Only under those conditions would the above be equal - if you had anything less then 3/3 Arcane Med and if the fight takes longer than 4:39, 11 mp5 completely dwarfs 20 spirit in practice.

Are you remembering the effect of Spirit on the size of your Evocation?

#7 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 02:40 PM

11 MP5 = 20 Spi

I also don't see how you could possibly make this statement, unless your Patchwerk attempt are 4 minutes and 39 seconds long and you have 3/3 Arcane Med.

Only under those conditions would the above be equal - if you had anything less then 3/3 Arcane Med and if the fight takes longer than 4:39, 11 mp5 completely dwarfs 20 spirit in practice.

Are you remembering the effect of Spirit on the size of your Evocation?

I most certainly do - and included it in my calculation; thanks for noticing.

20 spirit / 4 = 5 mana per tick x 1500% evocation = 75 mana per tick x 4 ticks = 300 mana added to your Evocation.

I have a Soulkeeper and a wand "of Spirit" to switch out with ItemRack on an Evocation event.

#8 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11580 posts

Posted 15 September 2006 - 03:06 PM


I also don't see how you could possibly make this statement, unless your Patchwerk attempt are 4 minutes and 39 seconds long and you have 3/3 Arcane Med.

Only under those conditions would the above be equal - if you had anything less then 3/3 Arcane Med and if the fight takes longer than 4:39, 11 mp5 completely dwarfs 20 spirit in practice.

Are you remembering the effect of Spirit on the size of your Evocation?

I most certainly do - and included it in my calculation; thanks for noticing.

20 spirit / 4 = 5 mana per tick x 1500% evocation = 75 mana per tick x 4 ticks = 300 mana added to your Evocation.

I have a Soulkeeper and a wand "of Spirit" to switch out with ItemRack on an Evocation event.

I'll check for bugs--to be honest, that did catch my attention. If you've paid any attention to the inner workings, feel free to check yourself.

#9 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 03:24 PM

Nah, I did my own basic calculation, you can check the code I suppose. Here it is:

Claim: On a fight of Patchwerk duration - 20 spirit = 11 mp5

Argument: Let's compare. As said previously, 20 spirit = 300 mana added onto your Evocation (using the math there). In Patchwerk, you're chaincasting pretty much all the time, so you'll be inside 5 seconds almost always. Therefore, you get 20 / 4 = 5 x 0.3 = 1.5 more mana every 2 seconds. This is 3.75 mp5. (I'll run this calculation again later assuming full Arcane Med).

So, so far we have 300 + 3.75x (where x is the number of 5 second periods). Setting this equal to 11x (which is 11 mp5 times the number of 5 second periods), we get x = 41.38 5 second periods = 207 second fight, or 3 minutes 27 seconds (clearly, Patchwerk takes much longer than this even in the best case).

So, I thought, OK, let's assume he has full Arcane Med. All that does is change the 0.3 above to a 0.45, and we recalculate. We get 55.81 5 second periods = 279 second fight, or 4 minutes 39 seconds (Patchwerk still is much longer than that).

So something is wrong with the calculation if that is the result you get for a 6 minute fight, which I would say is close to the lower bound possible for Patchwerk - I suppose with full Naxx gear you could get down to 5:15 or 5:30, but I don't think 4:39 is really feasible.

#10 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11580 posts

Posted 15 September 2006 - 03:39 PM

Nah, I did my own basic calculation, you can check the code I suppose. Here it is:

Claim: On a fight of Patchwerk duration - 20 spirit = 11 mp5

Argument: Let's compare. As said previously, 20 spirit = 300 mana added onto your Evocation (using the math there). In Patchwerk, you're chaincasting pretty much all the time, so you'll be inside 5 seconds almost always. Therefore, you get 20 / 4 = 5 x 0.3 = 1.5 more mana every 2 seconds. This is 3.75 mp5. (I'll run this calculation again later assuming full Arcane Med).

So, so far we have 300 + 3.75x (where x is the number of 5 second periods). Setting this equal to 11x (which is 11 mp5 times the number of 5 second periods), we get x = 41.38 5 second periods = 207 second fight, or 3 minutes 27 seconds (clearly, Patchwerk takes much longer than this even in the best case).

So, I thought, OK, let's assume he has full Arcane Med. All that does is change the 0.3 above to a 0.45, and we recalculate. We get 55.81 5 second periods = 279 second fight, or 4 minutes 39 seconds (Patchwerk still is much longer than that).

So something is wrong with the calculation if that is the result you get for a 6 minute fight, which I would say is close to the lower bound possible for Patchwerk - I suppose with full Naxx gear you could get down to 5:15 or 5:30, but I don't think 4:39 is really feasible.

Can you check quickly how the numbers work out with 21 Spi and 10MP5, just to estimate my rounding error?

Also, you get 16x regen during Evoc, not 15.

#11 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 03:51 PM

Nah, I did my own basic calculation, you can check the code I suppose. Here it is:

Claim: On a fight of Patchwerk duration - 20 spirit = 11 mp5

Argument: Let's compare. As said previously, 20 spirit = 300 mana added onto your Evocation (using the math there). In Patchwerk, you're chaincasting pretty much all the time, so you'll be inside 5 seconds almost always. Therefore, you get 20 / 4 = 5 x 0.3 = 1.5 more mana every 2 seconds. This is 3.75 mp5. (I'll run this calculation again later assuming full Arcane Med).

So, so far we have 300 + 3.75x (where x is the number of 5 second periods). Setting this equal to 11x (which is 11 mp5 times the number of 5 second periods), we get x = 41.38 5 second periods = 207 second fight, or 3 minutes 27 seconds (clearly, Patchwerk takes much longer than this even in the best case).

So, I thought, OK, let's assume he has full Arcane Med. All that does is change the 0.3 above to a 0.45, and we recalculate. We get 55.81 5 second periods = 279 second fight, or 4 minutes 39 seconds (Patchwerk still is much longer than that).

So something is wrong with the calculation if that is the result you get for a 6 minute fight, which I would say is close to the lower bound possible for Patchwerk - I suppose with full Naxx gear you could get down to 5:15 or 5:30, but I don't think 4:39 is really feasible.

Can you check quickly how the numbers work out with 21 Spi and 10MP5, just to estimate my rounding error?

Also, you get 16x regen during Evoc, not 15.

Without Arcane Med: 264 seconds, 4 minutes 24 seconds

With 3/3 Arcane Med: 391 seconds, 6 minutes, 31 seconds (this is the only one of the 4 cases which seems correct, but the real question is: did you really round? did your spreadsheet include 3/3 Arcane Med? and if so, why are you specced that way? lol)

#12 Maklar

Maklar

    Von Kaiser

  • Members
  • 46 posts

Posted 15 September 2006 - 03:55 PM

1 crit = 1.3 hit = 15 Fire dmg = 11 MP5 = 20 Spi = 28 Int

I am assuming this is based off of the mana value and does not take into account the added crit from the int?

#13 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 04:04 PM

1 crit = 1.3 hit = 15 Fire dmg = 11 MP5 = 20 Spi = 28 Int

I am assuming this is based off of the mana value and does not take into account the added crit from the int?

If it was based off the mana value, then 15 fire damage would do absolutely nothing for you, except possibly shorten the duration of the fight, but it would only be by a negligible amount and surely isn't part of the spreadsheet.

#14 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11580 posts

Posted 15 September 2006 - 04:15 PM

1 crit = 1.3 hit = 15 Fire dmg = 11 MP5 = 20 Spi = 28 Int

I am assuming this is based off of the mana value and does not take into account the added crit from the int?

Nope, it computes crit% based on Int. Static mana bonuses just have small effects over fights of nontrivial length.

#15 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11580 posts

Posted 15 September 2006 - 04:17 PM

1 crit = 1.3 hit = 15 Fire dmg = 11 MP5 = 20 Spi = 28 Int

I am assuming this is based off of the mana value and does not take into account the added crit from the int?

If it was based off the mana value, then 15 fire damage would do absolutely nothing for you, except possibly shorten the duration of the fight, but it would only be by a negligible amount and surely isn't part of the spreadsheet.

Correct, the spreadsheet makes the assumption that fight duration is a constant.

#16 Maklar

Maklar

    Von Kaiser

  • Members
  • 46 posts

Posted 15 September 2006 - 04:24 PM

Yeah, I had a brain lapse and was thinking ~30 int = 1 crit instead of 59.5

Does this take into account race ( <3 Gnome)?

Also, I believe using a mana gem triggers GC. Does it factor that in? *

*Clarify: What I actually mean is that I usually use it when I'm on GC (so after NW proc, or Fireblast). That way the time associated with using the command does not impact DPS. So assuming .27sec lag it would need to calculate using a gem/pot as .27sec instead of 0. Does it do this?

#17 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 04:26 PM

Did you find out if you had that rounding error or if your code was wrong?

I'm not about to pore over Excel here at work...especially since half my time is using Excel anyways.

:(

#18 snape

snape

    Great Tiger

  • Members
  • 777 posts

Posted 15 September 2006 - 04:27 PM

Yeah, I had a brain lapse and was thinking ~30 int = 1 crit instead of 59.5

Does this take into account race ( <3 Gnome)?

Also, I believe using a mana gem triggers GC. Does it factor that in?

Using Mana Gems does NOT trigger the global CD. That's the real beauty of them - Mana Potions are the same way. You can hit gem and then fireball right away (I use hotkeys to do so).

#19 Maklar

Maklar

    Von Kaiser

  • Members
  • 46 posts

Posted 15 September 2006 - 04:28 PM

Yeah, I edited my statement to clarify what I meant. I'm sure you could macro it much like a combustion/fireball macro to help nullify the lag but I do not do so.

#20 Chimp

Chimp

    Piston Honda

  • Members
  • 150 posts

Posted 15 September 2006 - 04:28 PM

Not modelled:

Cooldowns.
Rolling Ignites...

I once had a pop at modelling rolling ignites, and failed miserably. I wouldn't recommend trying ;D




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users