Wizard CM DPS Simulator - Release Version

  • #41
    Quote from AudiR8GT

    is it the same as DPS calculator?


    No, this is a simulator that attempts to determine effective dps of a CM wizard based on actual rotations of skills over a fixed time interval (like 60s fight against a boss), regeneration of APoC, etc. As far as I've seen, the DPS calculators only use the dps stats to compute the char sheet dps, and that value is very misleading for CM wizards since effective dps is highly dependent on factors such as IAS, APoC, and crit chance, among others.

    An example is crit chance. It will increase your char sheet dps but based on my results above, it also increases your effective dps, so you basically 'double dip' with an increase in crit.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #42
    Loroese

    What is your thought about 'Shards vs 3 targets'?

    I think 'Shards vs 3 targets' will get you more DPS gain than when you face just one target. I'd like to hear your opinion on this
  • #43
    Against multiple targets I expect shards to out perform prism because with enough targets you basically get AP saturated, i.e., you can't spend AP faster than you regain it. At that point Prism losed its effectiveness but shards continues to be useful.

    Otherwise, prism and shards both scale about the same for multi targets and as long as you can gain benefit from the reduced AP cost of spells, I think prism will out perform shards. This mostly seems to apply to lower gear levels though, so those with really high crit, attack speeds, and APoC likely don't get more from shards, even single target. I haven't gottan around to simulating that yet though.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #44
    Quote from Loroese

    Against multiple targets I expect shards to out perform prism because with enough targets you basically get AP saturated, i.e., you can't spend AP faster than you regain it. At that point Prism losed its effectiveness but shards continues to be useful.

    Otherwise, prism and shards both scale about the same for multi targets and as long as you can gain benefit from the reduced AP cost of spells, I think prism will out perform shards. This mostly seems to apply to lower gear levels though, so those with really high crit, attack speeds, and APoC likely don't get more from shards, even single target. I haven't gottan around to simulating that yet though.


    I have two sets of Chantodo's Will; one with a socket and the other with 3% Life Steal. With the one with a socket, I have 3.04APS, 55.5CC, 20ApoC and 225K DPS, and with the one with Life Steal, Only DPS goes down to 180K but I can use Shards instead of Shell(I should run Shell to deal with Reflected Damage)

    I am wondering that 180K DPS with Shards 'might' outperform 220K DPS with Shell against 3 targets. Against a single target, for example, 180K DPS with Shards records 1min and 7 sec average and 225K DPS with Shell records 1 min and 5 sec average(MP9 Ghom)
  • #45
    I'm thinking shards would probably outperform shell in that scenario, since they are pretty even for single target fights, based on your results and mine previously. With more targets you'll get faster refreshes of DS which means more dps.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #46
    I guess my biggest question would be at what aps and crit chance can you run shards with 20 or less ap on crit and outperform a wizard hat with additional ap on crit on a single target scenario
  • #47
    I'll look into that but I doubt I'll have any time to run simulations for a few days.

    Regarding the use of shards: from my experience, if you can survive your current difficulty with shards, do so. If not, use crystal shell. I suppose in your scenario switching to a new hat with no APoC would allow you to raise survivability sufficiently to drop shell for shards?
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #48
    I normally run storm armor and shards on mp 7 with minor deaths, but im finding if i try playing on 8-10 i struggle without changing to crystal shell or energy armor. Im unsure if its worth playing on those higher mp if i have to give up dps to survive. Im not sure if shards with energy armor grants more dps than storm with crystal shell or if its even worth playing those difficulties
  • #49
    I've been working on the simulator a bit today. I've got some good data to use to verify the simulator, which seems like it needs a bit of a tweak because my simulation effective dps multipliers are a bit low.

    On a side note, and probably of more interest, I discovered a couple things today. First, Storm Armor periodic damage is simply one strike per second on average, as I stated in another forum thread. Second, I just now realized the Chain Reaction tics are based on attack speed. At 1 APS they tic once a second so I assume the tic rate is basically 1/APS, so with 2 APS it's 0.5s between tics, though I don't think it will have much of an impact on my simulator because it still only tics 3 times. Still, I haven't seen that fact mentioned so I thought I'd share it.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #50
    Some new results, with accompaning simulation comparison.


    Storm Armor Shocking Aspect Pinpoint Shards DPS Mult Rel DPS Simuated dps Rel Sim DPS
    2.49 1 2.484 1.00
    x 2.73 1.10 2.766 1.11
    x 2.48 1.00 2.545 1.02
    x x x 4.34 1.74 3.569 1.44
    x 2.81 1.13 2.81 1.13
    x x 3.77 1.51 3.286 1.32


    Based on that table, it seems my simulator is decent at approximating Diamond Shards, Pinpoint barrier, and Storm Armor. The shocking aspect damage seems to be off by a decent amount, though I'm not quite sure why. I suspect it has to do with my method of key pressing. Instead of pressing 123 all at once, I kind of roll my fingers back and forth, which could lead to differences between my model and the game play. My model, for comparison, assumes I press all at the same time at a frequency equal to the ReactionTime variable. I probably won't have time to test it for a few days though.

    I set the ReactionTime number to 0.2388 to make my base dps match that of the simulation. My other stats: 51.5% crit, 19 APoC, 2.6904 APS. I had 16517 base dps, 17013 with pinpoint, and killed Belial on MP10 hell. I intentionally dropped my dps to make the fight last 1 min or longer, so I bought a very low dps dagger with +18% IAS (white dagger) that dropped my dps from 90k to 16k. I found Belial much more consistant of a fight because he never moves, and at Hell MP10 I had enough defensive stats that I didn't have to worry about dying. I dismissed the scoundrel and didn't use a storm crow. I think I did have 1-2 thorn items but the damage from them should be neglegible.

    I only ran each once though, instead of twice like I did previously.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #51
    Do you have any idea what may be cuasing my chain reaction to not be casted even when it's on CD? or maybe it is casting, but refreshing so fast that you can't tell?

    Sadly I don't have matlab atm to see how close my kill would match a sim to see if I am missing casts.

    It's rather annoying when I have the AP to cast it, but for some reason it just won't show the cooldown cycle animation.

    (fight starts 1 min in)
  • #52
    I've seen similar behavior. I'm not really sure what it is but sometimes button clicks just don't seem to register, and I can clearly see that happening in your vid, though there are plenty of times where the spell gets cast and the CD refreshes VERY fast. The same is true of DS and is also noticable in the vid.


    For reference, the kill time was around 72s, with 25.7 mil hp for an effective dps of around 357k dps. My simulator doesn't model Slow Time correctly, since it uses an attack animation and I assumed it didn't, but if I set your Reaction Time to 0.1 (i.e., 100 ms), my simulator gives an effective dps coefficient of 3.568, so with your 100k dps that's 356.8k effective dps, which is pretty close to your actual value.

    The main issue with my simulator at the moment is the ReactionTime variable. The effective dps can change pretty significatly when the value is tweaked. For example, if I change it to 0.2s for your stats, it gives an effective dps coefficient of only 2.61. I imagine it would be easier to match if I I used autohotkey or something.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #53
    Hypothetically, what would your sim show if I had hypothetically had used auto hot key at 50ms intervals between casting 1, 2, and 3?
  • #54
    Using ReactionTime=0.05 (i.e., 50 ms), the dps coefficient is 3.616. That's assuming all 3 buttons are pushed at the same time, every 0.05s, and WW is also pushed at the same time. Basically it prioritizes FN > WW > white attack, then casts EB if there's enough AP and activates DS if available. Then it adds back the appropriate amount of AP from crits over that time frame, and the amount from AP regen over that time step, then repeats. However, that doesn't account for the casts of Slow Time, which will take away 1 twister every 8s, resulting in slightly lower actual dps, so it's still pretty close to being on track. I might get around to adding it to the priority list if I find the time over the weekend.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #55
    Quote from Loroese

    Some new results, with accompaning simulation comparison.


    Storm Armor Shocking Aspect Pinpoint Shards DPS Mult Rel DPS Simuated dps Rel Sim DPS
    2.49 1 2.484 1.00
    x 2.73 1.10 2.766 1.11
    x 2.48 1.00 2.545 1.02
    x x x 4.34 1.74 3.569 1.44
    x 2.81 1.13 2.81 1.13
    x x 3.77 1.51 3.286 1.32


    Based on that table, it seems my simulator is decent at approximating Diamond Shards, Pinpoint barrier, and Storm Armor. The shocking aspect damage seems to be off by a decent amount, though I'm not quite sure why. I suspect it has to do with my method of key pressing. Instead of pressing 123 all at once, I kind of roll my fingers back and forth, which could lead to differences between my model and the game play. My model, for comparison, assumes I press all at the same time at a frequency equal to the ReactionTime variable. I probably won't have time to test it for a few days though.

    I set the ReactionTime number to 0.2388 to make my base dps match that of the simulation. My other stats: 51.5% crit, 19 APoC, 2.6904 APS. I had 16517 base dps, 17013 with pinpoint, and killed Belial on MP10 hell. I intentionally dropped my dps to make the fight last 1 min or longer, so I bought a very low dps dagger with +18% IAS (white dagger) that dropped my dps from 90k to 16k. I found Belial much more consistant of a fight because he never moves, and at Hell MP10 I had enough defensive stats that I didn't have to worry about dying. I dismissed the scoundrel and didn't use a storm crow. I think I did have 1-2 thorn items but the damage from them should be neglegible.

    I only ran each once though, instead of twice like I did previously.


    For fun, I added calculations to determine the dps contribution of each skill in the build. My results are shown below for the same case above and for another run where I tried to match the SNS dps multiplier. Please note the individual percentages are expected to vary greatly depending on individual stats (crit, APoC, IAS, overall dps in the form of simulation duration, etc.).

    DPS multiplier = 3.57, ReactionTime = 0.2388
    CR = 21.3365%
    White = 1.0723%
    WW = 37.7865%
    SA = 12.0307%
    StormA = 10.4272%
    DS = 17.3468%


    DPS multipler = 4.44, ReactionTime = 0.01
    CR = 22.0691%
    White = 1.5102%
    WW = 37.8826%
    SA = 12.1463%
    StormA = 8.4235%
    DS = 17.9684%


    It's interesting to note that the relative contribution of each skill doesn't really change much despite the increase in the DPS multiplier with the change in the ReactionTime variable. I still think the SA contribution is a bit low, since the in game data indicates it should be more like 30% contribution,. The DS and SA contributions seem consistant with in game data.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #56
    Loroese, what can I actually do with the file that you made available in the OP if I don't have MATLAB? Just open it in a text editor?
    I'm starting to spec my wizard to a CM build (I only use self-found gear or trades with friends, so it'll take ages), just for fun, and would like to see what actually changes when I switch some skils/runes around.
  • #57
    Quote from maka

    Loroese, what can I actually do with the file that you made available in the OP if I don't have MATLAB? Just open it in a text editor?
    I'm starting to spec my wizard to a CM build (I only use self-found gear or trades with friends, so it'll take ages), just for fun, and would like to see what actually changes when I switch some skils/runes around.


    Unfortunatelly, unless you have access to MATLAB, you can't really do anything with it, aside from try to convert it to a C++ program or something. One reason I'm posting the matlab file is in the hopes that someone with more familiarity with C++ can convert it to a more easily usable program. My experience lies mostly with working with MATLAB. When I get the Shocking Aspect portion closer to the expected results, I'll also post it on the official forums since there seems to be more traffic there, which should translate into a higher likely hood of someone being able to and interested in converting it.

    If you have some specific configurations you'd like me to simulate, you can post the stats and I'll try posting results when I have the time to run it, but that might take a few days because I've got other code to work on and run. As a rough approximation you could use the approximate relative dps values in my table above to approximate how much you gain from each skill. However, those results are highly dependant on the actual base stats, so the approximation might not apply for your gear set.

    On a side note, you can open it in a text editor, but without the MATLAB program itself, it's just a text file.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #58
    That's pretty much what I thought.
    Thanks for going through all the trouble, this is a great thread.

    EDIT: my cousin has a duplicate copy of a 2011 matlab he doesn't use. If I can get hold of it, is it worth it? Can I (fairly easily) just load up the m file and edit the fields? I've only used matlab a handful of times, so I'm quite the scrub with it.
  • #59
    Quote from maka

    That's pretty much what I thought.
    Thanks for going through all the trouble, this is a great thread.

    EDIT: my cousin has a duplicate copy of a 2011 matlab he doesn't use. If I can get hold of it, is it worth it? Can I (fairly easily) just load up the m file and edit the fields? I've only used matlab a handful of times, so I'm quite the scrub with it.


    Yes, you can basically just run the m file in matlab without any hassle. If you got your hands on the copy you probably already figured that out but it's still worth mentioning for anyone thinking of trying it who also has access to MATLAB. Major universities and colleges tend to have MATLAB on school computers for those in US.
    Wizard DPS and EHP Spreadsheet, mostly useful for wizards.
    Crusader DPS and EHP Spreadsheet, meant for Crusaders
    My Wizard
  • #60
    If you have time could you plug in another Ghom MP5 run for me?

    Mp5 Ghom
    144443.3 tooltip dps
    5443 dps from follower
    17APOC
    3.14APS
    56% CC
    Spec:
    Shards, Snap, Chain, Venom, Wicked

    46 second kill


    Namely I want to see if the ticks of venom weapon had the ability to contribute to CM procs and thus inflating my dps that way.

    Also where do you find how much HP these bosses have?
  • To post a comment, please or register a new account.
Posts Quoted:
Reply
Clear All Quotes