Page 2 of 4 FirstFirst
1
2
3
4
LastLast
  1. #21
    Quote Originally Posted by Eruionmel View Post
    I'm at 8973 haste and, as you can see above, my haste value is a full 0.5 higher than crit or mastery. So that's clearly not true.
    You missed the Part where I said reforging for Haste Values IN BETWEEN 6050 and 7598 are a DPS loss, this is due to how breakpoints work , There is another breakpoint past 7598, Looking at My spreadsheet really quick its around 10700, IN BETWEEN 7598 and 10700 Is a dps Loss, So its no wonder simcraft is telling you haste has a high statweight, because its trying to get you to that breakpoint. If you cant get to 10700 I suggest you reforge down to 7598, and trade it to mastery

    This is all assuming EOTE and EB talents, as the above poster says Haste will sim higher with PE

  2. #22
    Dreadlord Eruionmel's Avatar
    10+ Year Old Account
    Join Date
    May 2010
    Location
    Washington, USA
    Posts
    788
    Elemental shaman don't care about haste breakpoints. We have a single DoT, and an extra tick of it is next to worthless. Meaning your entire premise is flawed. Those numbers have nothing to do with why haste may not be your best stat.

    And yes, I'm using PE. But the post I quoted said "in every situation," which was not true.

    | PM Signature Requests: Closed | My Work | Signature Tutorial |

  3. #23
    Quote Originally Posted by Eruionmel View Post
    Elemental shaman don't care about haste breakpoints. We have a single DoT, and an extra tick of it is next to worthless. Meaning your entire premise is flawed. Those numbers have nothing to do with why haste may not be your best stat.

    And yes, I'm using PE. But the post I quoted said "in every situation," which was not true.
    I assure you I meant Using EB and EOTE talents, and I am telling you that you are dead wrong about Haste Being a Linear DPS increase, there are levels where taking more haste over another stat is a DPS LOSS over stacking something else. Again I am talking About EB EOTE Only, I have Not simmed PE But I will now when I get a chance

    For EB-EOTE those breakpoints are at 6050 7598 and 10700, If you have Haste in between these breakpoints It is a DPS Loss (Haste from 0-6050 is going to be your best stat). Dont Believe me? Sim it yourself (do the reforging plots it takes a long time)

  4. #24
    Dreadlord Eruionmel's Avatar
    10+ Year Old Account
    Join Date
    May 2010
    Location
    Washington, USA
    Posts
    788
    I never said it was a linear increase. I said that haste breakpoints were not why it scales differently.

    | PM Signature Requests: Closed | My Work | Signature Tutorial |

  5. #25
    Filthe, ilvl 510, http://eu.battle.net/wow/en/characte...ilthe/advanced

    EotE,EB haste > mastery > crit reforges: 121142 DPS
    Int: 4.32
    Haste: 1.47
    Crit: 1.74
    Mast: 1.82

    EotE/EB, mastery > crit > haste reforges: 120754 dps
    Int: 4.28
    Haste: 1.99
    Crit: 1.60
    Mast: 1.76

  6. #26
    Quote Originally Posted by Gendori View Post
    If you Sim the Plots you will see that haste Values in Between 6050 and 7600 are a DPS loss in every single situation. I dont know why exactly But it is. Reforging up to 7598 is Only a .01% DPS gain (yes its that small). Id rather have the chance to get really lucky with the extra mastery at 6050 Haste then to get more Haste and lose some luck
    You should double check the action list for any problems causing this. While it may in some rare cases just be the inherent mechanic of a class, such a behaviour of DPS standing still ( or even decreasing ) when you increase a stat is usually a good indicator of the action list being too rigid, either because it is not clever enough or because it has inflexible, hardcoded numbers in it.

  7. #27
    EotE / PE - http://eu.battle.net/wow/de/characte...omani/advanced

    ilvl 501
    haste: 6630
    mastery: 6665
    crit: 3171

    102868 dps
    Int: 4.14
    hit: 3.59
    haste: 1.78
    crit: 1.54
    mastery: 1.33

    any tipps for me ? wann keep EotE / PE
    Last edited by thehunta; 2013-03-11 at 10:37 AM.

  8. #28
    Quote Originally Posted by Caltiom View Post
    You should double check the action list for any problems causing this. While it may in some rare cases just be the inherent mechanic of a class, such a behaviour of DPS standing still ( or even decreasing ) when you increase a stat is usually a good indicator of the action list being too rigid, either because it is not clever enough or because it has inflexible, hardcoded numbers in it.
    The problems that are causing this is simple. Haste loses its value past 6050 because past that point Lava Burst can become GCD Locked During Hero, at the other breakpoints flame shock get an extra tick, making up for the loss of DPS by doing More damage, AND that extra tick giving you more chances for Lava Surge Procs. Yes this is a Very very Minor difference however it is still there, hence why haste from 0-6050 is a Linear Increase in DPS but after that Level you need to be near 7598 and 10500 to see any further increases.As I say again, if you are taking EB/EOTE, and have a haste value In between 6050 and 7598 or 7598 and 10500 You are giving yourself a very minor dps loss then if you would just reforge to one of those numbers. If you dispute my numbers sim it yourself, Im looking at the spreadsheet I put out right now, and I again assure you the sim is spot on with the right action priority and everything

    In any Case These Increases are VERY VERY VERY VERY Minor. Were talking a .03-.05% dps difference, maybe smaller

  9. #29
    Quote Originally Posted by Gendori View Post
    I assure you I meant Using EB and EOTE talents, and I am telling you that you are dead wrong about Haste Being a Linear DPS increase, there are levels where taking more haste over another stat is a DPS LOSS over stacking something else. Again I am talking About EB EOTE Only, I have Not simmed PE But I will now when I get a chance

    For EB-EOTE those breakpoints are at 6050 7598 and 10700, If you have Haste in between these breakpoints It is a DPS Loss (Haste from 0-6050 is going to be your best stat). Dont Believe me? Sim it yourself (do the reforging plots it takes a long time)
    Can you please post your results. I'm very surprised that you've observed break points, especially with your other claims.

  10. #30
    Sure thing, I sorted the excel spreadsheet that Simcraft made for me. Ill try and format it to this post thats something readable, but basically how it works is this, the first column if its a Positive number is what happens when you add that amount of haste from 6050 and the negative numbers are what happens when you subtract haste from 6050. The 3rd column is the simmed DPS for that Reforge. These are the top 30 reforges from 6050 haste, youll see the best reforges are right around the 7600 and 10600 mark, while the reforges In between are lower DPS than 0 (which is 6050 haste). These show definite Breakpoints where there is a dps gain over the values in between the breakpoints

    haste_rating mastery_rating DPS
    1500 -1500 113867.4148

    1550 -1550 113860.5526

    1450 -1450 113859.4231

    350 -350 113856.1842

    4650 -4650 113854.4305

    4700 -4700 113836.2078

    -50 50 113835.5849

    1600 -1600 113831.1521

    4850 -4850 113825.3532

    4750 -4750 113817.3911

    1650 -1650 113816.8356

    1400 -1400 113815.3872

    0 0 113813.8194

    100 -100 113811.7435

    -100 100 113810.5265

    4800 -4800 113809.2717

    300 -300 113806.7969

    4350 -4350 113800.929

    -200 200 113793.0175

    900 -900 113785.4505

    -250 250 113785.2379

    1350 -1350 113784.9214

    3900 -3900 113783.5789

    5000 -5000 113782.0924

    3950 -3950 113780.9286

    450 -450 113780.9242

  11. #31
    Dreadlord Eruionmel's Avatar
    10+ Year Old Account
    Join Date
    May 2010
    Location
    Washington, USA
    Posts
    788
    Ok, the variations in DPS amounts are so incredibly minute that there is no way you could actually draw any serious conclusions from them. Try running all of those tests against and you'll probably get wildly different results. I mean, you have things sorted that have a variation of 1 DPS. The first three "increases" above the 0 mark are 15, 16, 17 respectively. Do you know how much of a DPS difference that is?

    It's 0.00088%. You could lean forward to fart while DPSing and see a greater loss than that.

    If you want to draw any conclusion from your data, it should be that haste and mastery are so close in value at this point that there is no difference whatsoever and you can take whatever the hell stat you want. The difference between your "top" DPS and "bottom" DPS is 87 fucking DPS. 87. That's a total damage difference of 52,200 over the course of ten minutes.

    | PM Signature Requests: Closed | My Work | Signature Tutorial |

  12. #32
    Quote Originally Posted by Eruionmel View Post
    Ok, the variations in DPS amounts are so incredibly minute that there is no way you could actually draw any serious conclusions from them. Try running all of those tests against and you'll probably get wildly different results. I mean, you have things sorted that have a variation of 1 DPS. The first three "increases" above the 0 mark are 15, 16, 17 respectively. Do you know how much of a DPS difference that is?

    It's 0.00088%. You could lean forward to fart while DPSing and see a greater loss than that.


    If you want to draw any conclusion from your data, it should be that haste and mastery are so close in value at this point that there is no difference whatsoever and you can take whatever the hell stat you want. The difference between your "top" DPS and "bottom" DPS is 87 fucking DPS. 87. That's a total damage difference of 52,200 over the course of ten minutes.
    And this is all assuming perfect play,latency and a patchwork fight (those still exist right) ?

  13. #33
    I Had already said In One of my posts that the dps Gains are very very small, however, if you look at the spreadsheet you can clearly see that all the numbers around those breakpoints are clearly higher DPS than the numbers In between, and I have ran this twice, the first spreadsheet got overwrote by the second, the results were similar. For the record I ran 25k Iterations each time. This list is definitley accurate

    Also, This isnt the whole spreadsheet, I only took the top 30 reforges (there are actually over 200 of them), and the difference from first to last is 2k DPS

    Ill pick out a couple Nice ones though to prove my point. These are all haste values In between 6050 and 10800, Youll see that These points are statistically significant (.4% DPS LOSS)

    3100 -3100 113332.32
    3300 -3300 113358.0856
    3150 -3150 113356.2304
    3250 -3250 113354.7591

    Yet More .3-.4% DPS Losses

    2900 -2900 113444.7604
    3400 -3400 113435.4234
    2950 -2950 113421.5697
    2800 -2800 113414.582
    3450 -3450 113402.5531
    3000 -3000 113401.5437
    3050 -3050 113398.2082
    3350 -3350 113397.1264

    That 8500-9500 Haste range is Definitley a statistically significant dps loss, theres just too much data around those values, they cant ALL be wrong
    Last edited by Gendori; 2013-03-11 at 09:29 PM.

  14. #34
    Quote Originally Posted by Gendori View Post
    That 8500-9500 Haste range is Definitley a statistically significant dps loss, theres just too much data around those values, they cant ALL be wrong

    Yet looking at Elementals like Brightleaf from BL and Inane from UT you see them sitting at 8K+ haste. The numbers your talking about are so small that in practice with so many outside variables calling it a dps loss is an exaggeration.

  15. #35
    Dreadlord Eruionmel's Avatar
    10+ Year Old Account
    Join Date
    May 2010
    Location
    Washington, USA
    Posts
    788
    Quote Originally Posted by Gendori View Post
    I Had already said In One of my posts that the dps Gains are very very small, however, if you look at the spreadsheet you can clearly see that all the numbers around those breakpoints are clearly higher DPS than the numbers In between, and I have ran this twice, the first spreadsheet got overwrote by the second, the results were similar. For the record I ran 25k Iterations each time. This list is definitley accurate

    Also, This isnt the whole spreadsheet, I only took the top 30 reforges (there are actually over 200 of them), and the difference from first to last is 2k DPS

    Ill pick out a couple Nice ones though to prove my point. These are all haste values In between 6050 and 10800, Youll see that These points are statistically significant (.4% DPS LOSS)

    3100 -3100 113332.32
    3300 -3300 113358.0856
    3150 -3150 113356.2304
    3250 -3250 113354.7591

    Yet More .3-.4% DPS Losses

    2900 -2900 113444.7604
    3400 -3400 113435.4234
    2950 -2950 113421.5697
    2800 -2800 113414.582
    3450 -3450 113402.5531
    3000 -3000 113401.5437
    3050 -3050 113398.2082
    3350 -3350 113397.1264

    That 8500-9500 Haste range is Definitley a statistically significant dps loss, theres just too much data around those values, they cant ALL be wrong
    Turn the overall spreadsheet into a chart and post a picture of it.

    | PM Signature Requests: Closed | My Work | Signature Tutorial |

  16. #36
    I am really bad with excel if you could explain how to do it id be more than happy too

    Edit: Figured it out, enjoy

    Keep in mind where the 0 is is 6050 haste

    http://img837.imageshack.us/img837/8392/elechart.jpg

    When I get a chance Ill run one for PE
    Last edited by Gendori; 2013-03-11 at 10:41 PM.

  17. #37
    Dreadlord Eruionmel's Avatar
    10+ Year Old Account
    Join Date
    May 2010
    Location
    Washington, USA
    Posts
    788
    Quote Originally Posted by Gendori View Post
    I am really bad with excel if you could explain how to do it id be more than happy too

    Edit: Figured it out, enjoy

    Keep in mind where the 0 is is 6050 haste

    http://img837.imageshack.us/img837/8392/elechart.jpg

    When I get a chance Ill run one for PE
    If you're going to spend the time doing another one, stick to using the same talent setup, but use someone else's toon, preferably one of the people Intrepid mentioned. I want to see whether it's actually a statistically predictable phenomenon or whether it's toon-specific.

    Also, the difference between the top of the lower spectrum (~7600) and the bottom of the upper spectrum (~9200) was still only 500 DPS, which is well within the realm of variation. Most of the time your DPS variation is in the 10,000 realm, let alone 500. Even if it is significant to your toon, I'm willing to bet it won't be the same for every ele with your talent setup.
    Last edited by Eruionmel; 2013-03-12 at 12:12 AM.

    | PM Signature Requests: Closed | My Work | Signature Tutorial |

  18. #38
    Once I sort out a few things I'll be looking at a large haste plot (although I have to restrict it to 1000 plot points as LibreOffice only has 1024 columns) under a few conditions (with/without T15 4pc, Burning vs Sinister meta)

  19. #39
    Quote Originally Posted by Eruionmel View Post
    If you're going to spend the time doing another one, stick to using the same talent setup, but use someone else's toon, preferably one of the people Intrepid mentioned. I want to see whether it's actually a statistically predictable phenomenon or whether it's toon-specific.

    Also, the difference between the top of the lower spectrum (~7600) and the bottom of the upper spectrum (~9200) was still only 500 DPS, which is well within the realm of variation. Most of the time your DPS variation is in the 10,000 realm, let alone 500. Even if it is significant to your toon, I'm willing to bet it won't be the same for every ele with your talent setup.
    You quite dont understand how simcraft Plots reforges. Basically how it works is it Sims every single Reforge Plot and runs them through the full 10k iterations, so the reforge plot at 7600 haste has been run 10k times, and the plot at 7650 is run 10k times etc, Looking at the graph its easy to see that there is clear convincing evidence of every single plot in that range being a dps lot, 400-500 DPS is roughly .3-.4% If you were plotting a single point you could make a case that it is in error (simcraft puts the margin of error on each plot at roughly 72-74 dps), But you cant make the case when every single plot comes up a loss, they cant all be wrong

    This is also why Simming this stuff takes a LONG time (the one I linked took Over 3 hours)

  20. #40
    has anyone had a V shape result from their haste v mastery plot? Can't post the image sorry. I'm reading it as keep my haste and any gear I get now, keep that haste and reforge/gem mastery.

Posting Permissions

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