Wizard CM DPS Simulator - Release Version

  • #21
    That's some interesting analysis, but I think the video you were looking at was made before the meteor coefficient nerf, when it was still 0.25. Now, it's 0.1, the 0.05 is only for meteor shower, I think it's called. Also, FN returns some AP, though it only ends up being something like 1.275 AP per cast on average. I think the majority of the deficit comes from just plain old white attacks used while spamming meteor when out of AP.

    The video I was refering to is this one, after which he says he's only using 17 APoC. In the first 30s of the vid, before he gets knocked back, he casts 9 CR, 7 FN, and 16 meteors.

    Total AP cost is 13*9+16*28=565.
    AP regen is about 300 and starts wso he gains 265 AP through crits.
    CR: 9casts*3tics per cast*1/9 (coef)*17 AP per crit*0.45 crit = 22.9 AP gain
    Meteor: 16casts*4 tics (incl blast)*0.1 (coef)*17 AP per crit*0.45 crit = 49.0 AP gain
    FN: 7*1/6*17 = 19.8 AP gain
    22.9+49+19.8 = 91.7 AP gain total, so we're still 173.3 AP short.

    In 30s he has roughly 45 total attack opportunities, 23 of which are used on FN and meteor. If he lands all of the remaining as white attacks, that's about 22 white attacks. If we assume coefficient of white attacks is 1 (which I really need to test), then we get 22*0.45*17 = 168.3 AP gained, which is right around the defecit. He didn't actually land all the attacks, but he also started with some AP so the two factors mostly cancel out and we have all AP accounted for.


    Regarding my model, I assume that FN is always cast when available, otherwise Star Pact (or whatever main AP spender is being modelled) is used when AP is available, and CR is used after if there is enough AP. I do not account for trying to save AP for star pact by not casting CR since CR does the most damage per AP spent and the relative cost of CR is quite small. If FN isn't cast and there isn't sufficient AP for SP, then I assume a white attack is landed, though I should probably add a factor for white attack uptime, since not every white attack will land.

    The reason my WW and meteor coefficients are not the same is because the meteor tics are done after 3s while the WW is up for twice as long, which gives the mob twice the time to move out of range of it. I do agree that the actual weights used might not be the best possible ones but I felt they were reasonable for the analysis so far.

    Regarding Shocking Aspect, I haven't looked at the skill so I don't know anything about how it functions. If I knew how it worked, I could try including it in my model.


    @ Vibe1320
    I wouldn't use the results to say absolutely how one spec performs versus another, but rather compare how each spec does vs APoC, so just compare the WW changes, or just the MB changes as you increase AP and use whichever between WW and MB you like or feel performs best. As was said in the post above, the uptime coefficients are just very rough approximations.

    Personally I like WW because I just don't like the randomness of the tornados. Actually I prefer meteor because of the higher potential damage to annoying ranged mobs, but I only have an attack speed of 1.5 since I've mostly just gone Crit and Crit damage for Archon. That said, I haven't actually tried the build since the start of 1.0.4 because I never actually liked the spec. I'm only now doing this analysis because it seems like the most viable method for wizards to try and tackle the Uber bosses on MP10, which seems the best way to farm ring mats, if you can kill them before they enrage.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #22
    Quote from Loroese

    Right now, my expected CM spec would be http://us.battle.net...YXSO!YWX!ZcZZZY, assuming survivability isn't an issue. If it is, I'd likely replace Glass Cannon or Astral Presence with Evocation and Force Weapon with either Conduit or Blood Magic, though I might end up replacing Magic Weapon with Teleport since I really liked using teleport as a CM wiz.


    Thanks for all the information. But how do you call this a CM spec if you dont even have the Critical mass passive? Also, given your data the MB spec seems to be the best by far, having almost all the time the highest dps, highest freezetime, highest DS uptime. Why not use it instead of meteor? And again, where is the CM lol
  • #23
    Quote from paulobsf

    Quote from Loroese

    Right now, my expected CM spec would be http://us.battle.net...YXSO!YWX!ZcZZZY, assuming survivability isn't an issue. If it is, I'd likely replace Glass Cannon or Astral Presence with Evocation and Force Weapon with either Conduit or Blood Magic, though I might end up replacing Magic Weapon with Teleport since I really liked using teleport as a CM wiz.


    Thanks for all the information. But how do you call this a CM spec if you dont even have the Critical mass passive? Also, given your data the MB spec seems to be the best by far, having almost all the time the highest dps, highest freezetime, highest DS uptime. Why not use it instead of meteor? And again, where is the CM lol


    Oops, I messed up when throwing together the spec, but I fixed it by replacing Astral Presence with CM. Good catch and sorry for any confusion. I also swapped in Slow Time for Magic Weapon because it seems like it would do more damage since the uptime on it should be right around 100%.

    For my base case, and most of the data in the second post, Star Pact does more damage than MB, though it has a lower FN uptime. Since the base case is pretty close to my current gear, I looked at the results and the results using Morphos' data and concluded that if the build is viable with those stats using Meteor I would use meteor. The main reason I'd use SP is because I find it more reliable against ranged mobs and I can't help but feel it would have a higher uptime on mobs, plus it does more total damage than WW or MB per cast (though MB does more potential damage per AP). Also, my CM stats are currently pretty low, as indicated by the testing case. With more attack speed I would probably change to WW or MB since those seem to scale a lot better than the meteor specs. I should probably mention that somewhere in the first couple posts.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #24
    Well if that is the case it shows that Meteor Cost Reduction gear (rare sources, skull grasp, stone of jordan) is massively more valuable than APoC for single targets and attack speed is almost a dead stat in terms of DPS gain outside of scenarios where you have 5 or more targets getting rocks dropped on them.

    I'm also curious how Storm Armor (Shocking Aspect) figures in with either of these specs and if using it makes Energy Twister a clear winner over Meteor (for the content that can be cleared without the use of force armor or prismatic armor).

    Edit: Mechanics wise my understanding it is undergoes all of the same processes critical mass does, whenever you crit it rolls against the coefficient and determines if it is triggered. When it is triggered it produces a 35% weapon damage strike against a random single target within roughly the same range as frost nova, weighted towards the closest targets. These strikes can crit, but have an LoH/Crit Mass coefficient of 0. The PTR updated Storm Armor produces a visible lightning strike for the tooltip damage within a longer range roughly every 2 seconds (possibly affected by attack speed) that again can crit but has a coefficient of 0 for LoH/CM.

    On live with the 0.5 coefficient on storm chaser it is a prolific damage dealer, so much so that I have considered at times running builds without Explosive Blast because it doesn't feel like I need it with shocking aspect. I'm not sure how well it will add to damage in 1.0.5 but it still seems like a worthwhile switch if you have the survivability.
  • #25
    That actually sounds pretty easy to model. Do you know the proc coefficient on crit? And if I'm understanding it right, basically everytime you crit, you have a chance to proc the storm armor, which is a 35% single target hit. Also, do you know if there's any internal CD on the ability, or can you trigger it 5 times in 1 second if you get some crazy lucky procs from a high AS WW?
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #26
    I haven't done any serious investigation into its proc rate, I know on live with energy twister it is not possible to read all the damage numbers it generates because it is triggering so often, I'll see what I can figure out this week.

    Edit: and my video card at home decided to quit on me last night, so I guess I won't be doing any testing on this very soon.
  • #27
    So I'm running with a new video card and finally found some time to investigate the behavior of shocking aspect.

    Storm Armor's base damage is annoying and can get in the way, but here is what I have managed to determine:

    Shocking Aspect's chance to trigger appears the same as the LoH coefficient of the triggering spell, but I don't have enough data points to say conclusively. I tested this against monster power 10 inferno slow ass zombies using a set with a lot of EHP 30% crit chance, and a static damage range (white weapon and +min damage rings). I was easily able to confirm that it triggers on every critical hit from a magic missile through simple observation. For energy twister I used a 1.0 attack speed weapon and no IAS. I noted the hit point total of the zombie, cast a wicked wind and stood there while it took the damage. When the first wicked wind expired I cast another. I do this continuously while recording. Afterwards I watched the recording, counted the lightning strikes from storm armor and their damage, subtracted that from the total damage done to the zombie over the course of the fight. I then calculated the expected damage output of the energy twisters and subtracted that. The total damage divided by the expected damage of a shocking aspect strike gives a number of procs, which when compared to the expected number of crits from energy twister yields a proc coefficient. I did this for a total of 7 zombie kills and came very close to 0.125 every time.

    The Shocking Aspect attack can crit, but does not trigger LoH, Critical Mass, or itself. Base lightning bolts from Storm Armor similarly do not trigger LoH, Critical Mass, or themselves, at least when using the shocking aspect rune.

    The range of the shocking aspect shock is short, roughly the same as the radius for frost nova. Shocking aspect will favor the closest target to you, but this isn't guarenteed.

    I unfortunately didn't think to write down the number of times storm armor struck, just totalled the damage and subtracted it, would be nice to know how the base storm armor works and how much damage it does relative to tooltip DPS.

    This should be very simple to model, even in a closed form rather than a simulation like Matlab. I'm curious for cases of 1 or 2 targets (read here ubers) how much DPS relative to tooltip DPS does shocking aspect provide at a given value of crit. Mainly concerned with the tradeoff between using a Shield and Shocking Aspect versus using a Source and Energy Armor.
  • #28
    Quote from DisposableHeero

    So I'm running with a new video card and finally found some time to investigate the behavior of shocking aspect.

    Storm Armor's base damage is annoying and can get in the way, but here is what I have managed to determine:

    Shocking Aspect's chance to trigger appears the same as the LoH coefficient of the triggering spell, but I don't have enough data points to say conclusively. I tested this against monster power 10 inferno slow ass zombies using a set with a lot of EHP 30% crit chance, and a static damage range (white weapon and +min damage rings). I was easily able to confirm that it triggers on every critical hit from a magic missile through simple observation. For energy twister I used a 1.0 attack speed weapon and no IAS. I noted the hit point total of the zombie, cast a wicked wind and stood there while it took the damage. When the first wicked wind expired I cast another. I do this continuously while recording. Afterwards I watched the recording, counted the lightning strikes from storm armor and their damage, subtracted that from the total damage done to the zombie over the course of the fight. I then calculated the expected damage output of the energy twisters and subtracted that. The total damage divided by the expected damage of a shocking aspect strike gives a number of procs, which when compared to the expected number of crits from energy twister yields a proc coefficient. I did this for a total of 7 zombie kills and came very close to 0.125 every time.

    The Shocking Aspect attack can crit, but does not trigger LoH, Critical Mass, or itself. Base lightning bolts from Storm Armor similarly do not trigger LoH, Critical Mass, or themselves, at least when using the shocking aspect rune.

    The range of the shocking aspect shock is short, roughly the same as the radius for frost nova. Shocking aspect will favor the closest target to you, but this isn't guarenteed.

    I unfortunately didn't think to write down the number of times storm armor struck, just totalled the damage and subtracted it, would be nice to know how the base storm armor works and how much damage it does relative to tooltip DPS.

    This should be very simple to model, even in a closed form rather than a simulation like Matlab. I'm curious for cases of 1 or 2 targets (read here ubers) how much DPS relative to tooltip DPS does shocking aspect provide at a given value of crit. Mainly concerned with the tradeoff between using a Shield and Shocking Aspect versus using a Source and Energy Armor.


    That's good information. If I have time I'll look into adding that to my model since all the top wizards seem to be trying to use shocking aspect.

    The main issue with trying to come up with a simple model to estimate its effectiveness is each spell will add a different amount of dps gain from the armor.

    A rough guess at damage it does is if it procs off crit chance and spell coefficient than
    CC*ProcCoef*DmgCoef*Tics = approximate weapon damage gain from spell

    For WW with 45% crit and 2 attacks per second base, WW tics about 24 times per cast so you'd gain
    0.45*0.125*0.35*24 = 0.4725 weapon damage per target per WW or about 47% of weapon damage per cast. This ends up being about a 18.9% dps gain when accounting for the 252% base weapon damage of WW.

    For CR, it tics 3 times with a coefficient of 11.1% so we get 5.25% weapon damage gain compared to the 291% weapon damage total or about a 2% dps gain, so not as useful for CR.

    For FN and DS, it's a pure dps gain. If you cast FN every 2s and use shards with DS and cast that every 4s you'd end up with about a 1% overall dps gain.

    For white attacks, the net dps gain is about 15.75% since the proc coefficient is 100% (I have verified that in case anyone is worried).

    In short, WW damage increases by 18.9%, CR increases by 2%, white attack damage increases by 15.8%, plus a net dps gain of 1% from FN and DS. Overall my guess is about a net dps gain of 15% or so for 45% crit and 2 APS, but that doesn't include the Storm armor contribution, just the shocking aspect.


    Over the last day I actually started collecting some data on different specs, though my methodology was pretty terrible and just last night came up with a much better solution. Later today, if I have the time, I'll make some data collection runs to compare the effective dps of some of the different skills. I want to do that to see how efffective shocking aspect is as well as to try and verify that my simulator is working reasonably well for the different spells. I'll post my results when they're complete.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #29
    I have some interesting data that some might find interesting. What I did was kill Ghom a couple times on MP4 and record the times, using different specs. In particular I alternated between Storm Armor, Shocking Aspect rune, Pinpoint Barrier, Diamond Shards, and Prism. The test isn't very good for Prism because of the frequent aoe damage from his storm clouds, and with Shocking Aspect and either prism or shards I had some problems not dying, despite having almost 1300 LoH and 1.5% life steal, so that data might be a little off.


    I started with a full defensive CM build with Crystal Shell, Prismatic Armor, and the CD reduction rune on frost nova, as well as Cold Blooded, CM, and Evocation as passives that were unchanged throughout. I took the fight time and my char sheet dps to compute the DPS Multiplier, which is the effective dps gain of the build. For example, at 100k dps using the full defensive build I would expect an effective dps of 247k. I also computed relative DPS of each spec meaning switching from Prismatic to Storm Armor increased dps by about 36%.

    And in case there's any confusion, 'x' in the table means that skill was used.


    Storm Armor Shocking Aspect Pinpoint Shards Prism DPS Multiplier Relative DPS
    2.47 1
    x 2.78 1.13
    x 2.99 1.21
    x 3.01 1.22
    x 3.36 1.36
    x x 4.06 1.65
    x x x 4.25 1.72



    The interesting information is that Storm Armor is about a 36% dps gain over prismatic armor. When you add Shocking Aspect it's a 65% dps gain over prismatic armor and about a 36% dps gain over pinpoint barrier. Adding Diamond Shards further increases the dps gain but it's only about a 4% dps gain by itself when you use Shocking Aspect. The dramatic decrease in survivability for the fight makes it very not worthwhile, but that could easily change depending on the fight. Rak for example, would probably be just fine with DS or prism.

    I also expect the values for Shocking Aspect to be dependent of crit and attack speed, since it only procs on crits and higher attack speed should lead to more procs from WW, so with 50% crit, I'd expect the damage to increase by about 16%, or I'd expect the total dps gain to be about 90%, for a relative dps of 1.9. With a crit % of 52% I would expect Shocking Aspect to double your dps over prismatic armor, roughly speaking, with the same APS of 1.965 or so.

    For reference, my stats at the time of the data collection were 104k dps buffed (110k with pinpoint), 1.965 attacks per second, 20 APoC, 1298 LoH, 1.5% Life Steal, 641 resist all and 4588 armor unbuffed, with 40.5k hp and 43% crit with Scoundrel. I also had about 400 life per second regen, just to complete the stats.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #30
    Awesome! Thanks for all the work, doing the test runs and writing this up. Extremely informative.

    Just to make sure: I'm assuming your two main spells were WW+EB? It would be interesting to see how this works with the non-APoC-CM builds (Blades, Shock). I'll see if I have time to run similar tests this week... probably not too soon.

    I felt that Shocking Aspect is quite a DPS increase, but that much? Wow. The increase in DPS also means increased life steal - which could almost justify the decrease in survivability. If you have 4.5%, it could work out quite well...

    Do you have any idea for similar benchmarks to test for survivability? It's a shame that we don't have log files about incoming damage or so, it would make things so easy. I don't like that I have to rely on my subjective feeling which defense spells work best...
  • #31
    Yes, I used the standard WW+CR. I've been tempted to try blades, because it would let me swap my helm to a visage for more crit and attack speed, but I didn't think to try it on Ghom. I really wish Rak were easier to test but you have to run through a couple corridoors to get to him, which is a bit annoying, whereas for ghom you can just teleport outside his room. Otherwise Rak would be good to use to test shards vs prism, though he can teleport mid fight which is a bit annoying when you're trying to collect data like that.

    As far as testing survivability, I think Ghom is a good test since most of the things that really hurt are extra affects like Arcane Sentries, electrocute, desecrator, etc. Melee hits are few and far between when you're 1 on 1 with a boss or elite because of the frost nova. From my test runs, Shocking Aspect with Crystal shell worked just fine but DS or Prism fell off far too fast. In fact, I died the first time I tried with shards and got pretty scared the other two times I tried. I didn't even try prism because I didn't feel it was worthwhile seeing how low the uptime would be.

    I'm a little curious how his damage in Act 3 compares to the uber boss version though.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #32
    Awsome work. Here is the interesting part to me. Storm Armor with no rune gave you a 36% boost, while adding in shocking aspect brought that up another 29% to a 65% increase. The suggests that the base storm armor does more damage than the shocking aspect triggers, which is something I would never have guessed. With that in mind, it raises a next question in my mind. We can conclude from tooltips that in the same test, Storm Armor (Thunderstorm) would yield a 46.8% DPS gain leaving it still inferior to shocking aspect, Reactive Armor and and Scramble aren't terribly relevant to what we're trying to do, but what about Power of the Storm?

    Prism gave us a 13% DPS gain with very minimal uptime, while the base storm armor is worth 36%. This suggests that power of the storm might compete with Shocking Aspect in terms of damage gain, and is very likely to outpace Thunderstorm.
  • #33
    As a rough guess, I would expect Power of the Storm to increase dps by around 12%. My thought is that AP reduction of 3 means WW costs 32 which is around 10% cost reduction, and CR goes from 20 to 17 which is 15% cost reduction, so if you split the difference it's around 12-13%. That would put the total damage boost around 1.35*1.12 = 1.52 or around 50-55% damage increase for Storm Armor plus Power rune.

    What might be a good next step is to figure how precisely how Storm Armor works, so we can know how it scales with gear. I do have a fair amount of IAS gear I can examine since I was only using like 2 pieces for my test. I was going to use a 3rd but I swapped to a LoH ring when my first attempt with Shocking Aspect and Shards left me dead with Ghom around half health. I might look at that later, after i've had a chance to get some sleep.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #34
    I'll see if I can do some testing on this with different speed weapons. Tonight is a raid night for me so I won't be able to do much testing until tomorrow unless I lose the roll (we're overstaffed on melee DPS right now =().

    Also, despite around a 12% overall cost reduction, the effect has accelerated returns with APoC because more tornados also means more AP per second from APoC, so it is a reduction in cost and an increase in AP regen overall.

    Cost reduction gear does absolutely amazing things for Meteor builds, I wouldn't be surprised to see Power of the Storm compete with Shocking Aspect in the right circumstances.
  • #35
    Quote from Loroese

    As a rough guess, I would expect Power of the Storm to increase dps by around 12%. My thought is that AP reduction of 3 means WW costs 32 which is around 10% cost reduction, and CR goes from 20 to 17 which is 15% cost reduction, so if you split the difference it's around 12-13%. That would put the total damage boost around 1.35*1.12 = 1.52 or around 50-55% damage increase for Storm Armor plus Power rune.


    This could become very effective when you use Mistral Breeze. Hm... anyone ever tried Mistral Breeze + Prism + Power of the Storm? 10 AP Twisters all over the place... I'm gonna try this tonight.
  • #36
    Quote from Bagstone

    Quote from Loroese

    As a rough guess, I would expect Power of the Storm to increase dps by around 12%. My thought is that AP reduction of 3 means WW costs 32 which is around 10% cost reduction, and CR goes from 20 to 17 which is 15% cost reduction, so if you split the difference it's around 12-13%. That would put the total damage boost around 1.35*1.12 = 1.52 or around 50-55% damage increase for Storm Armor plus Power rune.


    This could become very effective when you use Mistral Breeze. Hm... anyone ever tried Mistral Breeze + Prism + Power of the Storm? 10 AP Twisters all over the place... I'm gonna try this tonight.
    If you want to go with the near zero cost build, I would advise Meteors. In addition to the 10 AP cost reduction from skills, you can get up to -19 meteor cost from gear (5 from source, 5 from SoJ, 5 from Mara's, 4 from Skull Grasp) leaving star pact with an AP cost of 6.
  • #37
    I did a test with Power of the Storm and got a relative dps coefficient of 1.43, compared to 1.36 for just storm armor for a net gain of only about 6% dps from the rune. I imagine it has some to do with my poor stats, resulting in a lot of white attacks that are unaffected by the rune.

    I also tried running with Spectral Blades - Deep Cuts, but even with prismatic armor and crystal shell, I died before I could get the boss below 40%. The dps was fairly competetive with the WW build over the 32s or so I was alive, but slightly less. The huge reduction is survivability was astonishing though.

    Lastly, I tried Mistral Breeze, with my base skill setup, and it took me about 10% longer to kill the boss than with WW. I just don't like MB because those tornados wander far too much for a spec that likes to stand still.

    I might have some time to kill tonight so maybe I'll try a few different gear configurations to see if there's any significant difference in Storm Armor dps at different attack speeds.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #38
    I have some results from testing a couple different gear configurations to try and see how attack speed affects Storm armor, but they're a little odd.


    Storm Armor DPS Mult Rel DPS APS
    2.61 1 1.652
    x 3.45 1.32 1.652
    2.47 1.00 1.9647
    x 3.36 1.36 1.9647
    2.54 1.00 2.2644
    x 3.04 1.20 2.2644


    The methodology is probably flawed, as I just kind of swapped gear around to get different attack speeds, so my char sheet dps was diffrent for each (though I normalized the storm armor dps with the prismatic armor dps for each case). My char sheet dps was something like 97.9k for the 1.652 case, 104.2k for the 1.9647 case, and 114k for the 2.2644 APS case.

    I think I need more data points before making any significant conclusions about the data. I find it odd that the first two attack speeds give a pretty similar dps gain for the armor but with the highest APS it decreased significantly. It might have to do with the fact that I used a different weapon for the low APS case, so I should probably use the same weapon for all cases and just swap my armor pieces to remove the IAS on them.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #39
    It just occured to me that I should have dismissed my follower for the tests since he adds a bit of a dps bias. It probably also explains the differences between my base prismatic armor case and my dps simulation results since he has a cold bow so he increases the effectiveness of my Cold Blooded passive. I'm a little reluctant to repeat my tests, but I might do that sometime in the next few days. I'm going to test the base case now though, because I'd really like to confirm that the simulator is working reasonably well. For reference, if anyone was curious, my simulator was telling me my effective dps should be around 2.0 as opposed to the 2.47 I obtained from my test.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #40
    is it the same as DPS calculator?
  • To post a comment, please or register a new account.
Posts Quoted:
Reply
Clear All Quotes