Page 2 of 2 FirstFirst
1
2
  1. #21
    Quote Originally Posted by Tickspoon View Post
    Crit is irrelevent since average DPS is what matters.

    For single-target, this graph overvalues Frost Bomb since it doesn't account for increasing the time spent casting.

    eg. 6 minute fight, 25% haste
    NT Cast Time: 360/12 * 1.5/1.25 = 36 seconds

    FB Cast Time: 360/(11.5/1.25) * 1.5/1.25 = 47 seconds

    Difference: (47 - 36)/360 = 3% of DPS time lost

    Since FB usually is ~16% of total DPS and NT ~13%, the 3% more damage is approximately cancelled out by the extra cast time.
    Not only that, but you will most likely have a 0.5-1.25 sec window in bethween Frost bomb coming off cooldown, and recasting it, and with NT while the window is same, as its from 2nd to last tick till recast, so the window of there not being a dot on is much smaller, and doing it every 12 secs with arcane and every 8 or so secs with frostbomb.. that account to quite some damage in the end..

  2. #22
    Deleted
    OK, I will make it simple.
    The chart are DPS. they are not damage nor DPET. So, except for Frost bomb who use a cast time, the GCD of Living Bomb and Nether Tempest don't have to be include.

    I will post the graph for Damage Per Execute Time this afternoon.
    The graph only show what it's made for evolution of the DPS in function of the Haste. If you want to take into account time reaction or lag, you can but I wont include it into the chart because it isn't quantifiable.

    For lag/time reaction, etc, we can say that:
    Frost Bomb's DPS will be reduce by imperfect rotation.
    Living Bomb will be hugely impacted if one mob move just a little out of the 10 yards for the cleave or for the explosion.
    Nether Tempest's DPS will drop if it's refreshed around end due to lag, if a mob move, and particularly be selection time for selecting targets.

    But I'm very curious to know why every one don't comment about NT linear evolution or LV breakpoint just before another tick.

    edit: actually, checking the file and searching a way to share it
    edit#2 : found a error in LV&NT (3 bombs on 3 targets), the DPS is only half of what is show. I'm currently searching an error in FB minimising is damage.
    edit#3 : found it. the sp scaling for the AOE part was the before buff value. I will update it this afternoon.
    edit #4 : no error found on haste, duration and number of tick.

    ---------- Post added 2012-12-20 at 11:09 AM ----------

    The DPET calculation is good.
    For Brain Freeze proc by haste, is it better to show it:

    • proc per second
    • proc per duration (can compare LV & NT but not FB)
    • proc per minute (aka proc/second x60 for a better reading)

    I think I will go for Proc Per Minute, as it can be compared to the new R-ppm for weapon enchants.

    ---------- Post added 2012-12-20 at 03:13 PM ----------

    graph updated in the OP

    I'm still work on BF proc.
    Last edited by mmocccfbebbec4; 2012-12-20 at 09:54 AM.

  3. #23
    Deleted
    Even in theory, I have not been able to go thru logs, expecting anything to be perfect is going to be flawed anyways.

    The liniair progression of NT is logical as there is equal distrubution of damage through out, thus will scale without plataue. Unless you build in such a penalty as I suggest below then each additional tick will be a jump UP instead of down of LB.


    However... in practical theory.... lets try to not do "actual bomb dps" only, but "bomb inside rotation dps"... After all our bombs are only part of our rotation where it represents roughly 10% of our total dps (as fire, depending on the fight)...

    In practice you will never stop an ongoing cast, which even in theory is very valid... /stopcast is always a dps loss, right?.... Allowing for an average ongoing cast of 1 second is I would say fair.
    Even with the 1 second ongoing cast, keeping LB up near 100% of the time is credable as there is a 3 second window to get it right.
    Nether Tempest is very likely you either refresh it to early (before the pan-ultimate tick) or to late (after the last tick), sometimes you will get it perfect but a lot of times you will refresh to early or to late... losing 0.5 tick on average per NT refresh is not a bad assumption I think
    Frost bomb, kinda simular in a perfect world you would always cast it at ( 10 + 1.5 ) / Haste intervals, but factual is that will hardly ever happen... even theory will know this. Again having a 1 second window would increase "absolute theory" : "real world values". Also the addition of this one second to Frost bomb (after haste) would kind off line up the duration of all 3 bombs to around 12 seconds and reduce/remove the discussion of "How does it impact the rest of your rotation becuase frost bomb is cast more".
    Worse yet, Frost bomb cant always be cast at the time you want it to be cast because for whatever reason in the fight you have to move (Attunuation, Unseen strike, Chains, you think of more) or are stunned or interrupted, contrary to the other two bombs Frost bomb has a cast time. Even in theoretical practice this will cost you sometimes seconds on reapplying your bomb, it will depend on the fight how hard this will hit, but 1 second every 4 casts? More? Less? But yeah some additional penalty due to movement is valid IMHO, my suggested time 1 second every 4 casts or 0.25 seconds every cast.
    Living Bomb simularly should get some kind of penalty to it as haste grows, as your LB duration drops just before the extra tick you do have to apply it more and your DPS of other sources will drop.... Every 120 seconds
    00.0% haste LB 10.00 times (12.00 second duration)
    12.4% haste LB 11.24 times (10.67 second duration)
    12.6% haste LB 09.01 times (13.32 second duration)

    Or just before the haste plataue you lose 1.2 x 1 second of other source dps
    Just after the haste plataue you gain 1 x 1 second of other source dps
    That is roughly a full cast Fireball every 2 minutes between being just under that plateau or just over... Or to be more exact a LOSS of other source dps by 1/2 a fireball just before the plateau to a GAIN of other source dps by 1/2 a fireball. Assuming a 60k fireball, that swings a 30k penalty and a 30k bonus dps around depending on its current durtion over 120 seconds (30k / 120 = 250 dps). Starting at 0 bonus/penalty at 0 haste, building to a 250 dps PENALTY at 12.4% haste and a 250 dps BONUS at 12.6% haste

    Nether Tempest only swings by 1.5 cast every 240 seconds (0.75 down and 0.75 up) or 0.75 * 1.4 second gcd per 240 seconds
    0.75 * 1.4 seconds (GCD) /2 second (cast time of Fireball) * 60k / 240 seconds = 131 penalty/bonus swing

    The only remaining point then to tackle is the 1 second GCD of LB vs the normal GCD of NT and FB, this can be "faked" by giving both NT and FB a penalty on their dps values. 4 Casts of any bomb will result is a ~2 second discrapancy between LB and NT/FB, what kind of damage can we do in roughly 2 seconds? For example 60k Fireball cast, thus even in theory we know we will lose that 60k damage every 4 casts therefor deducting the 15k dps per application (or perhaps 10k but atleast some penalty), which will average to about 1k dps loss per application of 12 seconds.

    Also the addition of the ideas above will line up the Mage bombs much more to AND reality AND to eachother AND overall DPS as a result of using these bombs.

  4. #24
    Why do people expect everybody to fail on NT mechanics? You can track the timer, you can track the ticks, you can see your own cast time, you can see when your current cast ends, if it doesnt line up with last tick of NT, you simply cast another spell and go on your merry ways..

  5. #25
    Deleted
    Quote Originally Posted by mrgreenthump View Post
    Why do people expect everybody to fail on NT mechanics? You can track the timer, you can track the ticks, you can see your own cast time, you can see when your current cast ends, if it doesn't line up with last tick of NT, you simply cast another spell and go on your merry ways..
    It's very true.
    A lot of people forget that a gains of 0.5s every 12s isn't a direct dps gain. It takes one or two minute to gain the equivalence of a GCD. At the same time, it's all this little optimisation that give the difference between a standard mage and the most powerful.
    Theoricrafting only tell how things function, not how to use it.

    Even if it helps one person, I will be happy.

    to note: I will correct the 3 targets graph tomorrow morning (dawm copy/paste)

  6. #26
    Deleted
    Quote Originally Posted by mrgreenthump View Post
    Why do people expect everybody to fail on NT mechanics?
    How can you expect to NOT fail at NT compared to not fail on LB, NT you have <1 second to get it right, LB you have 3 seconds to get it right

    Seriously there is so much more room for error on NT, unless you want to use a /stopcast to refresh it perfectly losing much more dps on your normal rotation than you gain on NT perfection.

  7. #27
    Deleted
    If you look closely to NT dps, BF's proc rate, etc, compared to the ones of LV, you will found that the difficulty for refreshing it is taken into account.
    If we look to BF proc by bomb:

    • Frost bomb: 1*100% = 100%
    • Living Bomb: 4*25% = 100%
    • Nether tempest: 12*9%= 108%
    OMG, NT is far ahead but you can have more proc than it's humanly possible to launch.

    But at the same time, Living Bomb have an 1s GCD but NT only have 1.5s.

    Is this unfair?
    not really, it's here to give all bomb strength and flaw.

    ---------- Post added 2012-12-21 at 09:46 AM ----------

    Quote Originally Posted by namliam View Post

    Living Bomb (...)

    00.0% haste LB 10.00 times (12.00 second duration)
    12.4% haste LB 11.24 times (10.67 second duration)
    12.6% haste LB 09.01 times (13.32 second duration)
    Good, I have the same numbers.

    In 60s, you will have:

    • at 12.4% haste, 4 ticks every 2.669 for a total of 22.48 ticks per minute with 5.62 bombs
    • at 12.6% haste, 5 ticks every 2.664 for a total of 22.52 ticks per minute with 4.50 bombs

    Conclusion,
    A gains of +0.04 ticks per minute cost 1.12 explosions per minute.
    with 1 explosion's damage ~ 4 ticks damages, the total loss is equivalent of :
    (1.12*4)-0.04 = 4.44 ticks.

    So in 1 full minutes, more than one Living bomb can be gains or loss. If DEPT is taken into account, Living Bomb is always the most priority except for Pyro! and AM (6 stacks).

    ---------- Post added 2012-12-21 at 10:25 AM ----------

    I just finish the graph for Brain Freeze ppm. It's just 3 lines that grow linearly at the same rate.

    @0% haste:
    • LB 5.000
    • FB 5.217
    • NT 5.400

    haste for 6/7/8 ppm :
    • LB 20% / 40% / 60%
    • FB 15% / 34.2% / 53.4%
    • NT 11.2% / 29.7% / 48.2%

    I can post it but it don't show anything interesting
    Last edited by mmocccfbebbec4; 2012-12-21 at 11:04 AM.

  8. #28
    Deleted
    Quote Originally Posted by Nathyiel View Post
    • at 12.4% haste, 4 ticks every 2.669 for a total of 22.48 ticks per minute with 5.62 bombs
    • at 12.5% haste, 5 ticks every 2.664 for a total of 22.52 ticks per minute with 4.50 bombs

    Conclusion,
    A gains of +0.04 ticks per minute cost 1.12 explosions per minute.
    with 1 explosion's damage ~ 4 ticks damages, the total loss is equivalent of :
    (1.12*4)-0.04 = 4.44 ticks.
    Note the bolded part, this means you are casting one less LB per minute yes, so you lose 1 explosion yes....
    But .... Every 2 minutes you save 2 LB applications/refreshes, which is 2 seconds saved in Globals, which in turn allows you to cast an extra Fireball.
    So in the one minute you do lose the explosion, but you gain half a Fireball in return....

    Self buffed on a dummy, my fireball hits for 42.5k + 15% Ignite, the LB explosion hits for 30k, though you only gain half a fireball
    So that is 48,875 * 50% = 24.5k, i.e. you do lose the explosion of 30, but you gain 24.5k in return.

    The adding of haste to either NT or LB is a up and down around the 12 second durution, while the duration is > 12 seconds you gain globals because you dont have to refresh your Bomb as much.
    While the duration is < 12 seconds you losing globals because you have to refresh your Bomb too much.
    Or if you are assuming the optimal DPS for LB to be at 12.4% (the maximum dps for the bomb itself) you lose Bomb DPS by going to 12.5%... But you cannot ignore the gain for the extra application

    So your bomb "rotational dps" going from 12.4% haste to 12.5% haste (using your numbers)
    My tick damage: 6755
    My Explosions: 30k
    My Fireball: 48,875 (including Ignite)

    12.4%: 22.48 ticks + 5.62 explosions = 22.48 * 6,755 + 5.62 * 30,000 = 151,852 + 168,600 = 320,452 damage or /60 for 5,340 dps
    12.5%: 22.52 ticks + 4.50 explosions = 22.52 * 6,755 + 4.50 * 30,000 = 152,122 + 135,000 = 287,122 damage or /60 for 4,785 dps
    To be very exact 1.12 seconds saved for a 2 second Fireball (a little more than half a fireball)
    Or..................................................... 1.12 / 2 * 48,875 + 287,122 = 27,370 + 287,122 = 314,492 damage or /60 for 5,241 dps.

    So even at my method the jump from 12.4% haste to 12.5% haste that still constitutes a dps loss, but a much smaller one than if you do not take the saved Global into account.

  9. #29
    Deleted
    All this calculation was made for only 1 targets.

    With 3 targets and 3 bombs, the gap is enormous :
    12.4% : 67.44 ticks + 16.86 explosions (for 50.58 explosion's damage) = 455.6k+1517.4k = 1973 k damages
    12.6% : 67.56 ticks + 13.50 explosions (for 40.50 explosion's damage) = 456.4k+1215.0k = 1671 k damages
    Time gains : 3.36s if cast & 2.8s if spread.

    It's a 301,600 damages loss for a 2.8s/3.6s gains, every minutes.
    3.6s/2 *48875 = 82110 damages
    2.8s/2 *48875 = 68425 damages

    To note: it's 12.6% haste, not 12.5% but it won't change anything.
    @12.4% haste, each ticks last 2.67s
    @37.4% haste, each ticks last 2.18s

  10. #30
    Deleted
    That is assuming you insta spread your LBs, which quite frankly you do not do.... I have not dug deep into should we force spread LB + Ignite + Pyro dot for example on Windlord every time we can (i.e. apply LB > IB)

    This will cost a good amount of Pyro procs, dont think the gain of (mostly) ticks on LB will make up for the loss of Pyro's
    Considering my HS-Pyro hits for 90k (including ignite excluding Pyro dot), compared to 48k of Fireball the 40k difference constitutes roughly 7 ticks of LB that you need to gain vs the HS-Pyro you are losing.

    And then there is a whole mess about how you should spread your ignite and Pyro-dot... So you want to use IB on HU (i.e. before you can cast HS-Pyro), but also immediatly after to spread the "love" you just put on your target with the Fireball/HS-Pyro
    Last edited by mmoc980c3dc910; 2012-12-21 at 12:15 PM.

  11. #31
    Deleted
    I take into account each solutions : 3 cast Living Bomb that takes 3*1x GCD and 1 Living Bomb and a Inferno Blast resulting in 1s+1.5s.

    I don't take into account fight specificity or class mechanics since it isn't the discussion here.
    And I have already mentioned what is prioritized over LB for DPET. There's spec forum to discuss if it's better to cast IB or to refresh LB, etc.

    But mathematics calculation show that Living Bomb do better damage just before a haste breakpoint than just after. It's true whatever you choose to spread it or to cast it manually.

  12. #32
    Quote Originally Posted by Nathyiel View Post
    edit (21/03/2013):
    [...]
    First of all thank you for your effort. I've got some remarks: please label your axis and chose colours + line thickness which can be seen more clearly.

  13. #33
    Deleted
    All graph are damage/haste, it seem obvious but I will do it my next update (110% haste).
    For color, I will see what I can do.
    For thickness, sorry but it will make it unreadable.

  14. #34
    Quote Originally Posted by Nathyiel View Post
    For thickness, sorry but it will make it unreadable.
    Hm... Have you tried gnuplot? You can easily save any Excel file in a simple text format (usually csv) and it produces diagrams in a very clear layout. Unfortunately it needs some time to get used to this kind of plot program if you haven't had any experience with such a program.

  15. #35
    Just save them as gifs - the high jpeg compression is what is rendering things difficult-to-read.
    Just another bad mage playing his class badly...
    Worst Mage EVAR!

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •