EnhSim, DPS simulator - Page 91

Sylvand · 07/14/09, 6:03 AM

All right, I think I have the following implemented:

Trinkets:
Victor's Call
Talisman of Volatile Power
Banner of Victory
Abyssal Rune
Death's Verdict

Totems:
Relentless Gladiator's Totem of Indomitability
Furious Gladiator's Totem of Indomitability
Relentless Gladiator's Totem of Survival
Furious Gladiator's Totem of Survival
Totem of Electrifying Wind
Totem of Quaking Earth

T9 2/4 piece Elemental (regalia) and Enhancement (battlegear) set bonuses
Shamanistic Rage cooldown/coefficient change

I will probably spend some time doing some unit tests on these tomorrow, and I still need to figure out how to share the updates out for other folks to test. Until then, if you know of anything else I should add, either missing now (like the furious totems I've already got) or announced for 3.2 (I think I added all the items from mmo-champion and I also looked over the PTR notes) please post here or PM me.

Pashaman · 07/14/09, 7:39 AM

Originally Posted by Levva

Instead what you should compare is the dps of your gear without anything in that slot, vs the dps with old trinket and dps of new trinket. That then gives you the dps value of old trinket and dps value of new trinket. This then makes a simple comparison possible.

I am not sure why you mention simming with an empty slot.
The matter of simple comparisons is not simple, because of the caps. An item that changes the hit/expertise balance needs regemming/reenchanting in order to make a meaningful comparison. Today, with the tools we have, this is what I do to perform a "simple" item comparison:
1. Simulate dps with current gear
2. In rawr, replace the current item with the one I want to compare
3. Also in rawr, painstakingly regem/reenchant to optimize hit and expertise
4. Export the stats from rawr to enhsim, simulate, compare with results of step 1

Step 3 can be nothing at all or can affect several other items. From a purist minmaxing perspective it is also error prone, since the item change can affect the relative weights of stats (crit vs AP for example) but in my regemming I assume the "old" stat weights. Typically to adress this I have to make "feedback loops" betwen steps 3 and 4, and, as others also noted, real world tests.

Unfortunately I don't have the time to do this for each possible upgrade, and repeat the whole process after each actual upgrade. The compromise I use is ofc EP.

I am sure this is something you guys already thought about, personally I don't see any easy solutions, but having the ability to do this "in real time" (or just faster) would be kinda fantastic.

Originally Posted by Malan

Nah my thought on the process was:
1) Input your current equipment with enchants and gems
2) Input any changes you want to make to that on a second panel. This would include changing out any gems to stay at/under caps.
3) The sim runs a thread for set and shows you the dps differences.
[...]
The whole idea was to simplify the most common uses of the sim, one of which is to anwer "is item 1 better than item 2?"

OK, this is exactly the scenario I was describing as well. From Malan's (possibly simplifying) description, I see that the only improvement over today is that I won't use 2 separate tools (rawr and enhsim) but just jEnhSim. I hope it's more than this, because otherwise trying to recreate Rawr's paperdoll functions (including armory scraper and saving different configs) would not bring any obvious value.

TLDR:
- recreating the functions of EnhSim in Java - kudos, the community is addicted to the tool, tukez seems MIA
- recreating the paperdoll functions of Rawr in jEnhSim - questionable investment of time
- writing a gear optimizer in the future - kudos again, although I lack the apparatus to understand how this will realistically work
- getting rid of EP - only after we have a proven better alternative, please

Levva · 07/14/09, 7:48 AM

Originally Posted by paladyin

I am not sure why you mention simming with an empty slot.
The matter of simple comparisons is not simple, because of the caps. An item that changes the hit/expertise balance needs regemming/reenchanting in order to make a meaningful comparison. Today, with the tools we have, this is what I do to perform a "simple" item comparison:
1. Simulate dps with current gear
2. In rawr, replace the current item with the one I want to compare
3. Also in rawr, painstakingly regem/reenchant to optimize hit and expertise
4. Export the stats from rawr to enhsim, simulate, compare with results of step 1

Step 3 can be nothing at all or can affect several other items. From a purist minmaxing perspective it is also error prone, since the item change can affect the relative weights of stats (crit vs AP for example) but in my regemming I assume the "old" stat weights. Typically to adress this I have to make "feedback loops" betwen steps 3 and 4, and, as others also noted, real world tests.

Unfortunately I don't have the time to do this for each possible upgrade, and repeat the whole process after each actual upgrade. The compromise I use is ofc EP.

I am sure this is something you guys already thought about, personally I don't see any easy solutions, but having the ability to do this "in real time" (or just faster) would be kinda fantastic.

Your step 3 is the problem, you don't need to do this manually. What Rawr allows for is to run the optimiser you can mark your equipment you have with green diamonds and then run the optimiser it then regems/re-enchants for you to optimise your dps. Check the provided video "Tour of Rawr" under the help menu if you aren't sure how to use it.

THIS is what the aim of the GA optimiser is in jEnhSim. You enter your current gear add in extra possible items and it will play around with the possibilities INCLUDING gems/enchants and tell you what is the best setup. Rawr already does this and does it well, however as has been noted actually simming an Enhancement setup rather than modelling it will give more accurate results.

It mimics what most people want from a sim, I have this bunch of gear equipped and this other stuff in my bags. "What combination of this kit and what gems/enchants gives me the best results?" The design of the optimiser is to answer this question.

Originally Posted by Sylvand

I will probably spend some time doing some unit tests on these tomorrow, and I still need to figure out how to share the updates out for other folks to test.

I have given you full project co-ordinator rights to the project this means that you can create a release - just mark it as a beta. BTW I'd recommend version numbering it as v1.8.0. To create a release you simple zip up the package for release and use the release page to upload the zip file. Enter some comments about it and hey presto. Don't make it the default though until patch 3.2 hits and testing is completed.

Rouncer · 07/14/09, 9:51 AM

Originally Posted by Sylvand

I will probably spend some time doing some unit tests on these tomorrow, and I still need to figure out how to share the updates out for other folks to test. Until then, if you know of anything else I should add, either missing now (like the furious totems I've already got) or announced for 3.2 (I think I added all the items from mmo-champion and I also looked over the PTR notes) please post here or PM me.

Mind updating the Orc option to include fist weapons? Also the engineering and tailoring bonuses will need to be updated.

Engineering

haste gloves - 340 haste for 10 seconds with a 1 minute cooldown to 340 haste for 12 seconds with a 1 minute cooldown
rocket gloves - 1440 to 1760 Fire damage on a 45 second cooldown to 1654 to 2020 Fire damage on a 45 second cooldown

Tailoring

Lightweave - 250 spellpower to 295 spellpower
Swordguard - 300 AP to 400 AP

Vespasian · 07/14/09, 1:20 PM

Originally Posted by Sylvand View Post
I will probably spend some time doing some unit tests on these tomorrow, and I still need to figure out how to share the updates out for other folks to test. Until then, if you know of anything else I should add, either missing now (like the furious totems I've already got) or announced for 3.2 (I think I added all the items from mmo-champion and I also looked over the PTR notes) please post here or PM me.

Can we please have Comet's Trail added to the trinket list in the coming update? Tyvm in advance.

Ulduar BiS list
I have recently begun using RAWR as a means of generating stat pages for export into Enhsim to identify a final BiS list. I am of the understanding that many people on this forum do not see value in a BiS list due to a) availability of gear and b) relative stat weightings and their volatility based on gear combinations. It is precisely this volatility though that I am looking to address in defining a BiS set. I am approaching BiS set up currently, and am tired of simming individual pieces of gear/enchants/gems etc in search of the highest dps, whilst always knowning that there is likely a different combination available to me that could result in 50-100 more dps. With Ulduar drawing to a close, I think it would be valuable to develop a finalised BiS set that people can look to in their remaining weeks of raiding.

I for one am particularly interested to explore the benefit of Comet's Trail/Dark Matter measured against the crit loss required to maintain hit cap, the ideal off-set tier item (Gloves or Shoulders?) and the ideal ring combination. I am certain that these are the main questions that other Ulduar 25 geared shaman are asking themselves, or will be asking themselves soon.

Since the sim provides us with a clear indication of our maximum dps potential, I am simply looking for the single gear/gem/enchant combination that provides me with the highest potential DPS.

I have done my own tests but given the number of potential combinations I think a collaborative approach would be best. I am envisaging:

a) an agreed set of parameters for any tests
b) the upload of paperdoll sheets (Rawr or other paperdoll interface) with accompanying config files demonstrating results.
c) correction and re-upload of sheets/configs until an optimal setup is reached. The integrity of the config pages will be important since they buff/lag variables must be consistent.

We can begin the process now and make amendments when Comet’s Trail is added to the sim (I am running latest exe through the 1.7.0 GUI btw)

I will start the ball rolling. A few words on the parameters:
1) I used 10m combat durations- most encounters run for approx 10 mins, and either way it certainly is not too short an amount of time to arrive at reliable data
2) I used 50-60 lag since this is the average US ping (yes I know this does not account for human reaction error, but I am interested in potential).
3) I used JC and LW for the professions as they both provide generic +64 AP.
4) I used a Tauren since they do not receive any dmg bonus from their racials. I am aware that the BiS gearing will be moderately affected by Draenei racial and slightly affected by Orc racial.
5) I used Deadly Gladiator Totem as it is the best available on current version of Enhsim. Obviously Furious will provide superior DPS.
6) I used a generic priority order: SW>MW5>ES_SS>SS>ES>LL>MT>LS. I am aware that there are different opinions on ideal priority list, but for the puproses of this I am sure we can agree tha this priority order will suffice.
7) I used a generic spec 16/55/0- as above.
8) I am fully raid buffed- because if you aren't, then this list probably has little relevance to you in the first place.

I managed to achieve a report of 7121dps with the following:

Rawr sheet: http://uploading.com/files/V8LJMJL1/...50709.xml.html

Config

### Saved by version 1.7.0 of EnhSimGUI

simulation_time                 1000
simulation_time_combatlog       30
combat_length                   10
report_count                    80
threads                         2
min_lag                         50
max_lag                         60
simulate_mana                   0

ep_precision                    2
ep_base_stat                    ap
ep_ap                           200
ep_crit_rating                  30
ep_hit_rating                   30
ep_expertise                    4
ep_haste_rating                 30
ep_armor_penetration_rating     30
ep_spellpower                   150
ep_dps                          7.5
ep_mana                         0
ep_spirit                       0
ep_mp5                          0

mh_auto_attack                  1
oh_auto_attack                  1
wait_ss_with_wf_cd              0.00
cast_ll_only_if_wf_on_cd        0
bloodlust_casters               1
sync_bloodlust_with_trinkets    0
cast_lvb_only_if_ed_left        15.0
cast_lvb_only_if_fsdots_left    4
cast_lvb_only_if_fs_active      1
cast_fs_only_if_dots_left       0
cast_ls_only_if_charges_left    0
cast_sr_only_if_mana_left       750
use_mana_potion_if_mana_left    3000

rotation_priority_count         9
rotation_priority1              SW
rotation_priority2              SR
rotation_priority3              MW5_LB
rotation_priority4              ES_SS
rotation_priority5              SS
rotation_priority6              ES
rotation_priority7              LL
rotation_priority8              MT
rotation_priority9              LS

miss                            8.0
dodge                           6.5
glancing                        24.0
armor                           10645
spell_miss                      17.0
nature_resistance               0
fire_resistance                 0
frost_resistance                0
arcane_resistance               0
shadow_resistance               0

armor_debuff_major              20.0/20.0
armor_debuff_minor              5.0/5.0
physical_vulnerability_debuff   4.0/4.0
melee_haste_buff                20.0/20.0
melee_crit_chance_buff          5.0/5.0
attack_power_buff_flat          687/687
attack_power_buff_multiplier    99.7/99.7
spell_haste_buff                5.0/5.0
spell_crit_chance_buff          5.0/5.0
spell_crit_chance_debuff        5.0/5.0
spell_damage_debuff             13.0/13.0
spellpower_buff                 165/165
spell_hit_chance_debuff         3.0/3.0
haste_buff                      3.0/3.0
percentage_damage_increase      3.0/3.0
crit_chance_debuff              3.0/3.0
stat_multiplier                 10.0/10.0
stat_add_buff                   51/51
agi_and_strength_buff           178/178
intellect_buff                  60/60

replenishment                   1
water_shield                    0
mana_spring_totem               0
blessing_of_wisdom              1
judgement_of_wisdom             1

flask_elixir                    flask_of_relentless_assault
guardian_elixir                 -
potion                          haste_potion
food                            fish_feast

###############################################################################
### Everything in the section below can be replaced by information obtained ###
### from your paper doll stats or exported by the ShockAndAwe addon         ###
###############################################################################

race                            tauren
mh_speed                        2.7
oh_speed                        2.7
mh_dps                          188.1
oh_dps                          178.9
mh_crit                         35.23
oh_crit                         35.23
mh_hit                          12.66
oh_hit                          12.66
mh_expertise_rating             136
oh_expertise_rating             136
ap                              4430
melee_haste                     13.16
armor_penetration               2.44
str                             135
agi                             883
int                             541
spi                             155
spellpower                      1329
spell_crit                      27.16
spell_hit                       15.82
spell_haste                     10.13
max_mana                        12231
mp5                             0

mh_imbue                        windfury
oh_imbue                        flametongue

mh_enchant                      3789
oh_enchant                      3789

mh_weapon                       -
oh_weapon                       -

trinket1                        45931
trinket2                        45522

totem                           42606

set_bonus1                      worldbreaker_battlegear_4
set_bonus2                      -
set_bonus3                      -

metagem                         41285

gloves_enchant                  -
cloak_enchant                   -

glyph_major1                    feral_spirit
glyph_major2                    stormstrike
glyph_major3                    windfury_weapon

glyph_minor1                    -
glyph_minor2                    -
glyph_minor3                    -

ancestral_knowledge             2/5
improved_shields                3/3
mental_dexterity                3/3
shamanistic_focus               1/1
flurry                          5/5
elemental_weapons               3/3
unleashed_rage                  3/3
weapon_mastery                  3/3
dual_wield_specialization       3/3
mental_quickness                3/3
improved_stormstrike            2/2
static_shock                    3/3
maelstrom_weapon                5/5

convection                      0/5
concussion                      5/5
call_of_flame                   3/3
elemental_devastation           3/3
reverberation                   0/5
elemental_focus                 0/1
elemental_fury                  5/5
call_of_thunder                 0/1
unrelenting_storm               0/3
elemental_precision             0/3
lightning_mastery               0/5
elemental_oath                  0/2
lightning_overload              0/5
lava_flows                      0/3
storm_earth_and_fire            0/3
shamanism                       0/5

I simmed Dark Matter and crudely simmed Comet’s (using elemental focus stone) and both came up less dps than the Mjolinir Runestone when used with Blood of the Old God. I did a quick sim replacing Blood with Dark Matter but it required a lot of gear swapping and hit gemming to maintain cap. The result was a dps loss. I am not convinced that these results are fully accurate though and would welcome comments.

This config serves as the baseline. If you are able to find a gear/gem/enchant combination that sims higher than this in the current version of enh sim, please post the paperdoll sheet and the config. Furthermore, if you find errors in my own paperdoll, config or methodology please post as I welcome any suggestions. Tyvm in advance.

P.S I think the process would be made easier by an online paperdoll interface, but I was unable to locate one. Is someone able to suggest one?

Malan · 07/14/09, 1:38 PM

Originally Posted by paladyin

- recreating the paperdoll functions of Rawr in jEnhSim - questionable investment of time

Yes a few of you seem intent on hammering that point and yet nobody can say why. We also have no intent of "recreating" Rawr in the sim, this seems to be some weird leap that everyone is making from my statement that we are "representing gear on a per slot basis."

snakeyez · 07/14/09, 2:01 PM

Originally Posted by Vespasian

P.S I think the process would be made easier by an online paperdoll interface, but I was unable to locate one. Is someone able to suggest one?

chardev.org v6 ~ a World of Warcraft character planner is what you are asking for? i use it for simming out my wish list gear granted i don't think it accounts for professions. (like skinning crit ect ect)

Sylvand · 07/14/09, 3:03 PM

Originally Posted by Vespasian

Ulduar BiS list
I have recently begun using RAWR as a means of generating stat pages for export into Enhsim to identify a final BiS list. I am of the understanding that many people on this forum do not see value in a BiS list due to a) availability of gear and b) relative stat weightings and their volatility based on gear combinations. It is precisely this volatility though that I am looking to address in defining a BiS set. I am approaching BiS set up currently, and am tired of simming individual pieces of gear/enchants/gems etc in search of the highest dps, whilst always knowning that there is likely a different combination available to me that could result in 50-100 more dps. With Ulduar drawing to a close, I think it would be valuable to develop a finalised BiS set that people can look to in their remaining weeks of raiding.

I don't see availability in gear as a reason. Sure, some folks might not get all the hard modes but a lot of people do, so it becomes simply a matter of picking from that gear. For me I think it's more a matter of simulation variance. If the difference between BiS and second BiS is 0.5 dps, and your simulation stddev is 5 dps, how do you figure out which would be better? We could use Rawr's approximations as a deterministic metric, but then the question becomes fairly trivial enough. Beyond that, 10 minutes is not necessarily a reasonable measure, as a lot of fights are not this long (Hodir hard of course comes to mind). I would tend to say 6 minutes might be more reasonable.

Beyond that, I would say you are restricting far too many variables for your test. It doesn't make any sense to pick Tauren for a couple reasons. First, you may not have freedom on your faction, depending on who you want to raid with. Second, and Orc would never ever reroll Tauren to squeeze out more dps, but a Tauren might reroll Orc. The hunter BiS list accounted for this by having a BiS alliance and horde version, though in theory you could have only one which showed the best race overall.

Also, things like rotation and talent spec should not be fixed, as they are practically free to change. If your spec and/or rotation are always optimal, let it be shown in the highest dps config.

Lastly, if you think there is merit to this project, you might want to start a new thread. I do plan to have an EnhSim Beta out probably tomorrow night that will include the items you mention, though it will also include the 3.2 changes mentioned so 3.1 results might be a little off.

clliche · 07/14/09, 5:08 PM

Regarding genetic algorithms:

Levva, based off of your explanation of a genetic algorithm, wouldn't the same problems that surface with EP occur? You give the specific example of swapping out an ap gem for a stamina gem and having the algorithm realize that AP is more beneficial. However, how does an algorithm deal with hit and its various soft caps? More importantly, can an algorithm like that deal with fluctuating values? x haste for instance might seem better than ap at one point in the "breeding" but at another point it may indeed be worse. Basically, if a genetic algorithm consistently chooses the most sucessful stats and combines them with other combinations of the most sucessful stats, does it realize that stacking for a certain stat may not be beneficial due to fluctuating values? I feel like its a distinct possibility that an algorithm like this will try to stack certain stats because the initial batches will all produce combinations where a certain stat is the best (hit for example). This ignores that a mix of stats is usually better. I guess this depends on how the algorithm chooses which chromosomes are the most fit within each batch however.

Sylvand · 07/14/09, 5:36 PM

Originally Posted by clliche

Regarding genetic algorithms:

Levva, based off of your explanation of a genetic algorithm, wouldn't the same problems that surface with EP occur? You give the specific example of swapping out an ap gem for a stamina gem and having the algorithm realize that AP is more beneficial. However, how does an algorithm deal with hit and its various soft caps? More importantly, can an algorithm like that deal with fluctuating values? x haste for instance might seem better than ap at one point in the "breeding" but at another point it may indeed be worse. Basically, if a genetic algorithm consistently chooses the most sucessful stats and combines them with other combinations of the most sucessful stats, does it realize that stacking for a certain stat may not be beneficial due to fluctuating values? I feel like its a distinct possibility that an algorithm like this will try to stack certain stats because the initial batches will all produce combinations where a certain stat is the best (hit for example). This ignores that a mix of stats is usually better. I guess this depends on how the algorithm chooses which chromosomes are the most fit within each batch however.

Ideally the breeding/mutation algorithm wouldn't be skewed toward particular stats, or make any EP-like inferences. Rather the offspring/mutations would be some random combination of the ancestors and have its performance independently evaluated. If the evaluation ranks it high, the only special treatment is that it's less likely to get culled. Offspring and mutations that perform poorly for any reason (e.g. they put too many hit gems in) will get culled because they perform poorly, not as a direct result of any rating cap issues.

One can certainly imagine a strain that survives several generations even with an excess of expertise/hit rating, on sheer virtue of its otherwise superior gear. However the idea behind GA is that this strain should eventually produce offspring or mutations that replace some of that expertise/hit with a better stat, and as a result will perform even better.

Schwolop · 07/14/09, 9:23 PM

Generally speaking, all optimisations can be thought of as the minimisation of some function, by variation of the parameters. The issue with GAs and many other techniques is that the optimisation can be caught in a local minima. GAs try to avoid this by peturbing each generation with some random genes. However, the larger the "valleys" due to local minima, the larger the random peturbations need to be to escape them. As these random peturbations increase, the time taken to compute a solution increases.

There are solutions to this too, including reducing the peturbation size over time, etc, but AFAIK there's no general method. Each problem needs to have a highly tuned GA associated with it. I'm sure this is possible for jEnhSim to achieve, but it's certainly non-trivial, and I'd agree that this should probably be the last item on the to do list.

bobxii · 07/14/09, 10:37 PM

Originally Posted by Schwolop

Generally speaking, all optimisations can be thought of as the minimisation of some function, by variation of the parameters. The issue with GAs and many other techniques is that the optimisation can be caught in a local minima. GAs try to avoid this by peturbing each generation with some random genes. However, the larger the "valleys" due to local minima, the larger the random peturbations need to be to escape them. As these random peturbations increase, the time taken to compute a solution increases.

Welcome to protein crystallography

. When it gets to the computational power requirements that GA requires, it may become simpler (relatively) to use it to come up with the most accurate closed-form solution possible instead of directly solving for the optimum condition; even if it is a 14-dimensional set (rank 14+ tensor, IIRC), the benefit of being able to explicitly solve for the best solution (using a closed-form equation) likely outweighs the trouble to get to it numerically.

Actually on that note, a few of the stats (dimensions) can be trivialized with statistical approximations to reduce the dimensionality (AP and ArP, for example) of the overall equation. Several require simulation to arrive at an accurate model (haste, crit, hit, 4pc t8, weapon speeds, most procs... that's it(?!))

Of course, a simpler way would be to put together the set of all possible gear combinations (within a certain gear range and that meet certain criteria (hit and exp capped), to be sure), and sim them all. It would be (ridiculously) computationally expensive, but isn't terribly difficult to code for, and it only needs doing once. Then simply host it on a server where the Rawr package can make direct comparisons between gear pieces.

Vespasian · 07/15/09, 12:39 AM

Originally Posted by Sylvand

If the difference between BiS and second BiS is 0.5 dps, and your simulation stddev is 5 dps, how do you figure out which would be better?

The sim will offer a standard deviation of approx 5 dps per simulation, so perhaps our results should be lists in ranges rather than absolute figures i.e 7112-7117 dps depending on the highest and lowest dps that you observe after running multiple simulations. Perhaps then, new data must demonstrate a dps increase in the range of 3-5 dps so as to accommodate this deviation, as well as recognise only actual increases, not statistical variances.

Beyond that, 10 minutes is not necessarily a reasonable measure, as a lot of fights are not this long (Hodir hard of course comes to mind). I would tend to say 6 minutes might be more reasonable.

I chose 10mins since it is the length of XT-002 hard mode, which I feel is the best available dps environment for testing the dps potential of an enh shaman in Ulduar (that and perhaps Algalon). Iron Council is 'similar' depending on your strat. If this list is maintained going forward, perhaps this duration can be revised in light of the nature/duration of 3.2 encounters. Either way, I think a longer time range will provide more accurate data than a shorter range when looking for dps potential.

It doesn't make any sense to pick Tauren for a couple reasons. First, you may not have freedom on your faction, depending on who you want to raid with. Second, and Orc would never ever reroll Tauren to squeeze out more dps, but a Tauren might reroll Orc. The hunter BiS list accounted for this by having a BiS alliance and horde version, though in theory you could have only one which showed the best race overall.

I think we can generally assume that at this stage of an expansion, it would be highly unlikely that anyone interested in these lists would have the time or inclination to abandon their Ulduar geared character so as to re-roll a race with superior racials. I do not think it would be appropriate to generate a list for the 1% of players who might be willing to do this. This list must be broadly practical in its application.

Since this is a gear analysis and not a gear + race analysis, I think it would be appropriate to remove racials as a variable from the equation. Anything that other races are able to achieve over the listed Tauren results is a bonus for them.

Finally, my gut feeling is that racials will not affect gearing choices as much as we might expect. In the optimal set up that I was able to discover, expertise was uncapped and hit was well over cap. Efforts to cap expertise and dump hit in favour of crit resulted in a dps loss. My point is that expertise and hit values are highly dynamic, and I do not expect the Draenei or Orc racials to greatly affect the gearing results (orcs/trolls obviously would achieve higher dps due to Command, Zerker etc). Still, it would make sense for people to share any results they find that disprove this hypothesis.

Also, things like rotation and talent spec should not be fixed, as they are practically free to change. If your spec and/or rotation are always optimal, let it be shown in the highest dps config.

I think this is completely correct. I am tempted to say that the same be should of professions, but again in the interests of practicality I think we can assume that with the coming homogenisaiton of professions (JC nerf) it is unlikely that someone would spend 3-6k gold on a profession change that would net a 5dps increase.

Lastly, if you think there is merit to this project, you might want to start a new thread.

I will do just that so long as I meet with no objections from the mods.

I do plan to have an EnhSim Beta out probably tomorrow night that will include the items you mention, though it will also include the 3.2 changes mentioned so 3.1 results might be a little off.

Since mana is not simmed the SR change should not affect results. We will likely see people putting more points into reverb though in 3.2 as mana will be less of an issue, but the sim would support both current and future talents specs in line with this.

Vespasian · 07/15/09, 4:31 AM

I have completed the first part of this project and am keen for some feedback. I will post what I have done here for discussion and feedback, and if the feedback is generally positive I will then open a new thread in the Shaman forum. Since this project is about finding the single highest dps set up as indicated through Enhsim, I am sure that the information is still relevant to this thread. The next two posts are as they would appear in a new thread.

Edit: I have started a new thread to house this discussion. It can be found at [Enhancement] BiS gear and set-up discussion

The following posts/responses relate to posts that have been moved to the above link.

Vespasian · 07/15/09, 4:33 AM

Post removed.

Vespasian · 07/15/09, 4:34 AM

Post removed.

bobxii · 07/15/09, 5:54 AM

I simmed ~7239 with MT above LL, by the way.

Vespasian · 07/15/09, 6:43 AM

Originally Posted by bobxii

I simmed ~7239 with MT above LL, by the way.

I get roughly 7141. Can you please PM me your config? I am eager to avoid back and forth discussions on this thread as I would prefer to save them for the actual thread (if and when it is formed).

Peterle · 07/15/09, 7:28 AM

Maybe your simulation time is a little bit to low (even if you're simulating 10 times). I would prefer a simulation time of 100000 hours.

Sylvand · 07/15/09, 8:05 AM

I'm happy to announce EnhSim v1.8.0 test release is available for download.

I've documented the changes as best I can, so even if you don't run the build please take a look and see if there's anything off. For now I expect the proc chance on the trinkets and totems may be off, but I have limited data so far. If folks have access to these items on PTR and can get some more numbers it would be much appreciated.

I've also done some basic sanity checks on pretty much all the items, and things seem to be consistent with what you would expect from the rest of the simulator. If anyone can take time to play with it and let me know how it goes that of course is also appreciated.

Again feel free to PM me or reply here. I'll also be watching the 3.2 thread for relevant updates.

EnhSim - Release: v1.8.0

p.s. I've also attempted to fix the Mac binary in there (though it doesn't have my changes, not sure what version it is really) so if there are any Mac folks that can try it out, let me know.

Mjoedvitnir · 07/15/09, 9:01 AM

my BiS equip reaches 7186-7187 dps. if you use the pvp totems it might be 10dps higher - i dont use it. also i use tuskarr's vitality (/cat's swiftness) since it's better than anything else in reality imo, if you want best possible dps in sim change it - i go for best possible dps ingame.

i've used v1.8.0 since comet's trail isnt implemented in 1.0.7.3b. in 1.0.7.3b i get 7036 without the procc from comet's trail.
so here it is:

rawr xml
enhsim cfg

//edit i removed the comet's trail and tested if 1.0.7.3b and 1.0.8 are showing the same results, they do.

Deathicle · 07/15/09, 12:17 PM

@Vespsian: Is there any particular reason you elected to use Relentless Assault and Haste potion over Endless Rage and potion of speed?

Erysha · 07/15/09, 1:25 PM

I noticed that you're using 2 Balanced Twilight Opals. Wouldn't the BiS be better served by using Enchanted Tear (+6 to all Stats)?

Vespasian · 07/15/09, 1:27 PM

Peterle, I ran the sim using 10000 hour simulations and the results were identical. If there is a reason why you think the extende simulation hours would provide different/more accurate data, can you please post at [Enhancement] BiS gear and set-up discussion

Mjoed, thanks for this. I took your work and expanded upon it by tyding up the enchants and talent spec. I was able to achieve roughly 7280 dps.

Deathicle, that was just me being bad. FixT.

suicidalkatt · 07/16/09, 2:56 AM

I'm seeing posts of EnhSim Version 1.8.0

where is this downloadable from?

I've only see 1.7 available on any major links =/