EnhSim, DPS simulator - Page 96

Wootlessberger · 08/14/09, 4:08 PM

Ok, I got the good version now. Still some low numbers on haste.

4 EP values with latests version :

EnhSim version 1.8.3.1

Spell hit + EP range goes past cap by 1.28364% (33.6725 hit rating).
Expertise is past cap, EP expertise test will be skipped.

Calculating EP values...
Calculation 1/14: Baseline DPS calculated.
Calculation 2/14: Attack power calculated.
Calculation 3/14: Crit(physical) calculated.
Calculation 4/14: Crit(spell) calculated.
Calculation 5/14: Hit calculated.
Calculation 6/14: Expertise skipped.
Calculation 7/14: Haste calculated.
Calculation 8/14: Armor Penetration calculated.
Calculation 9/14: Spellpower calculated.
Calculation 10/14: Mh Dps calculated.
Calculation 11/14: Oh Dps calculated.
Calculation 12/14: Mana skipped.
Calculation 13/14: Spirit skipped.
Calculation 14/14: Mp5 skipped.


EP                        value     DPS       total DPS      difference

baseline                                      6837.18
ap                        1.00      0.62      6961.28        124.10
crit rating               2.03      1.26      6874.98        37.80
hit rating                1.62      1.01      6867.39        30.21
expertise rating          0.00      0.00      0.00           -6837.18
haste rating              1.10      0.68      6857.73        20.55
armor penetration rating  1.77      1.10      6870.12        32.94
spellpower                0.99      0.61      6929.38        92.20
mh dps                    5.47      3.39      6862.63        25.45
oh dps                    2.85      1.77      6850.46        13.28
strength                  1.10
agility                   1.82
intelligence*             1.44
spirit                    0.00      0.00      0.00           -6837.18
mana                      0.00      0.00      0.00           -6837.18
mp5                       0.00      0.00      0.00           -6837.18

* part of the EP value is missing because of skipped tests.
Could not open enhsim_ep_exports.txt
elapsed simulation time: 10000.00h
elapsed real time: 40.73s
simulation speed: 883869x

EnhSim version 1.8.3.1

Spell hit + EP range goes past cap by 1.28364% (33.6725 hit rating).
Expertise is past cap, EP expertise test will be skipped.

Calculating EP values...
Calculation 1/14: Baseline DPS calculated.
Calculation 2/14: Attack power calculated.
Calculation 3/14: Crit(physical) calculated.
Calculation 4/14: Crit(spell) calculated.
Calculation 5/14: Hit calculated.
Calculation 6/14: Expertise skipped.
Calculation 7/14: Haste calculated.
Calculation 8/14: Armor Penetration calculated.
Calculation 9/14: Spellpower calculated.
Calculation 10/14: Mh Dps calculated.
Calculation 11/14: Oh Dps calculated.
Calculation 12/14: Mana skipped.
Calculation 13/14: Spirit skipped.
Calculation 14/14: Mp5 skipped.


EP                        value     DPS       total DPS      difference

baseline                                      6836.90
ap                        1.00      0.64      6964.65        127.75
crit rating               1.69      1.08      6869.38        32.48
hit rating                1.77      1.13      6870.76        33.86
expertise rating          0.00      0.00      0.00           -6836.90
haste rating              1.13      0.72      6858.61        21.71
armor penetration rating  1.69      1.08      6869.32        32.42
spellpower                1.00      0.64      6932.34        95.44
mh dps                    5.89      3.76      6865.11        28.21
oh dps                    3.21      2.05      6852.29        15.39
strength                  1.10
agility                   1.70
intelligence*             1.39
spirit                    0.00      0.00      0.00           -6836.90
mana                      0.00      0.00      0.00           -6836.90
mp5                       0.00      0.00      0.00           -6836.90

* part of the EP value is missing because of skipped tests.
Could not open enhsim_ep_exports.txt
elapsed simulation time: 10000.00h
elapsed real time: 40.68s
simulation speed: 885042x

EnhSim version 1.8.3.1

Spell hit + EP range goes past cap by 1.28364% (33.6725 hit rating).
Expertise is past cap, EP expertise test will be skipped.

Calculating EP values...
Calculation 1/14: Baseline DPS calculated.
Calculation 2/14: Attack power calculated.
Calculation 3/14: Crit(physical) calculated.
Calculation 4/14: Crit(spell) calculated.
Calculation 5/14: Hit calculated.
Calculation 6/14: Expertise skipped.
Calculation 7/14: Haste calculated.
Calculation 8/14: Armor Penetration calculated.
Calculation 9/14: Spellpower calculated.
Calculation 10/14: Mh Dps calculated.
Calculation 11/14: Oh Dps calculated.
Calculation 12/14: Mana skipped.
Calculation 13/14: Spirit skipped.
Calculation 14/14: Mp5 skipped.


EP                        value     DPS       total DPS      difference

baseline                                      6840.09
ap                        1.00      0.63      6965.66        125.57
crit rating               1.58      0.99      6869.86        29.77
hit rating                1.56      0.98      6869.38        29.29
expertise rating          0.00      0.00      0.00           -6840.09
haste rating              0.92      0.58      6857.36        17.27
armor penetration rating  1.51      0.95      6868.53        28.44
spellpower                0.97      0.61      6931.88        91.79
mh dps                    4.68      2.94      6862.13        22.04
oh dps                    2.29      1.44      6850.89        10.80
strength                  1.10
agility                   1.60
intelligence*             1.41
spirit                    0.00      0.00      0.00           -6840.09
mana                      0.00      0.00      0.00           -6840.09
mp5                       0.00      0.00      0.00           -6840.09

* part of the EP value is missing because of skipped tests.
Could not open enhsim_ep_exports.txt
elapsed simulation time: 10000.00h
elapsed real time: 40.65s
simulation speed: 885587x

EnhSim version 1.8.3.1

Spell hit + EP range goes past cap by 1.28364% (33.6725 hit rating).
Expertise is past cap, EP expertise test will be skipped.

Calculating EP values...
Calculation 1/14: Baseline DPS calculated.
Calculation 2/14: Attack power calculated.
Calculation 3/14: Crit(physical) calculated.
Calculation 4/14: Crit(spell) calculated.
Calculation 5/14: Hit calculated.
Calculation 6/14: Expertise skipped.
Calculation 7/14: Haste calculated.
Calculation 8/14: Armor Penetration calculated.
Calculation 9/14: Spellpower calculated.
Calculation 10/14: Mh Dps calculated.
Calculation 11/14: Oh Dps calculated.
Calculation 12/14: Mana skipped.
Calculation 13/14: Spirit skipped.
Calculation 14/14: Mp5 skipped.


EP                        value     DPS       total DPS      difference

baseline                                      6836.70
ap                        1.00      0.64      6964.72        128.02
crit rating               1.95      1.25      6874.09        37.38
hit rating                1.59      1.02      6867.31        30.61
expertise rating          0.00      0.00      0.00           -6836.70
haste rating              1.07      0.69      6857.29        20.59
armor penetration rating  1.49      0.95      6865.31        28.60
spellpower                0.98      0.63      6931.21        94.51
mh dps                    6.09      3.90      6865.95        29.24
oh dps                    2.67      1.71      6849.53        12.82
strength                  1.10
agility                   1.80
intelligence*             1.42
spirit                    0.00      0.00      0.00           -6836.70
mana                      0.00      0.00      0.00           -6836.70
mp5                       0.00      0.00      0.00           -6836.70

* part of the EP value is missing because of skipped tests.
Could not open enhsim_ep_exports.txt
elapsed simulation time: 10000.00h
elapsed real time: 40.75s
simulation speed: 883348x

I did try with uncapped expertise and hit, still low haste. Maybe haste is bad for me?

Thanks for your time.

Axekilla · 08/14/09, 4:50 PM

Please delete this post. I forgot to include Shamanistic Rage in my rotation, which lead to my OOM issues during simulation.

Sylvand · 08/14/09, 5:58 PM

Originally Posted by Axekilla

When running Enhsim 1.7.0.3b or 1.8.0, I was able to successfully simulate mana usage w/o strange results like 40% oom time.
With the latest 1.8.3.1, regardless of the values I put in, If I check simulate mana, my dps drops to 5400 or so. If I uncheck it, it'll shoot back up to 7250 or thereabouts.

I don't have my exact config as I'm at work. I'll edit this post and add that later if that will help.

Is there a reason why simulate mana doesn't work anymore? Am I missing something?

My spec is 18/53 (2 pts reverb, 0 in imp ss) as per the BIS sheet. sim time and settings are identical to that same first post in the BIS sheet.

I don't know of any reason that simulate mana wouldn't work anymore. Feel free to PM me your config or post it here.

Thaekor · 08/14/09, 8:01 PM

I have tried looking on here, and have looked through the readme file from the .zip download. I keep getting this error:

Could not open enhsim_ep_exports.txt

This file isn't in either the 1.8.2 version, or the beta 1.8.3 version I just downloaded. I tried creating this text file, since it also, stopped showing that it would output the ep values to the output.txt file. Still to no avail. Also, it is showing my windfury / flametongue as having a 0% uptime, in what it actually did output. I am just stumped as to what is going on.

I have tried it using both the EnhSimGui, and the batch files in the zip download. The only portion of the config file I changed was the info that is copied to the clipboard, from Rawr.

Jessamy · 08/14/09, 8:19 PM

When you download and run enhsim on a computer that already had a previous version, make sure to point the enhsim gui to the location of the new executable version.

Sylvand · 08/15/09, 12:20 AM

Originally Posted by Thaekor

I have tried looking on here, and have looked through the readme file from the .zip download. I keep getting this error:

Could not open enhsim_ep_exports.txt

This file isn't in either the 1.8.2 version, or the beta 1.8.3 version I just downloaded. I tried creating this text file, since it also, stopped showing that it would output the ep values to the output.txt file. Still to no avail. Also, it is showing my windfury / flametongue as having a 0% uptime, in what it actually did output. I am just stumped as to what is going on.

I have tried it using both the EnhSimGui, and the batch files in the zip download. The only portion of the config file I changed was the info that is copied to the clipboard, from Rawr.

enhsim_ep_exports.txt is not needed; it hasn't been included for a while though I know there was a version somewhere that included it. Basically it's used to build e.g. a lootrank URL based on the calculated EP values. Here's the one I use, you can build your own similar to it:

http://www.lootrank.com/wr.asp?Cla=64&s7=4&Max=15&n1=1&Str=$strength&mh=$haste&dps=$mhdps&mcr=$crit&odps=$ohdps&spd=$spellpower&Agi=$agility&mhit=$hit&Int=$intelligence&Spi=&map=$attackpower&arp=$armorp&Exp=2&Ver=6&i3=0&i9=0&j8=0&t1=46212&t2=45480&t3=46203&t5=46205&t8=45553&t9=46208&t10=45249&t6=45108&t7=45444&t11=45503&t31=45525&t12=45931&t32=45263&t4=45224&t14=45449&t15=45463&t16=40267

Thaekor · 08/15/09, 1:16 PM

I have the gui pointing to the new executable. I already checked that, and I ran it from the batch files aswell. I guess I should have been a little clearer on what I wanted. After running the EP calculation batch(or gui version for that matter), the output.txt file doesn't have the pawn string like it did in previous versions.

Recently, I have found myself playing resto far more often than enhance, and I roll on enhance gear when no one else wants it. Now i have several pieces of gear, and I want to see which of them are truly upgrades and which are sidegrades at best. Rawr is great, but it doesn't load the items in your inventory, without alt-tabbing to load each item individually (at least as far as I know). With pawn and Enhsim I used to run the sim, load the pawn string and pick the biggest upgrade item, and run it again, until I got the gear sorted out.

Levva · 08/16/09, 6:12 PM

Originally Posted by Thaekor

I have the gui pointing to the new executable. I already checked that, and I ran it from the batch files aswell. I guess I should have been a little clearer on what I wanted. After running the EP calculation batch(or gui version for that matter), the output.txt file doesn't have the pawn string like it did in previous versions.

Recently, I have found myself playing resto far more often than enhance, and I roll on enhance gear when no one else wants it. Now i have several pieces of gear, and I want to see which of them are truly upgrades and which are sidegrades at best. Rawr is great, but it doesn't load the items in your inventory, without alt-tabbing to load each item individually (at least as far as I know). With pawn and Enhsim I used to run the sim, load the pawn string and pick the biggest upgrade item, and run it again, until I got the gear sorted out.

If you have Character Profiler addon loaded RPG Outfitter | Character Profiler: WoW Addon then you can use that to import into Rawr it will import your character and all the items in your bags and bank and mark them all as available.

Izuul · 08/17/09, 4:54 PM

I'm using the sim for the first time today (shaman is an alt), and i noticed that the EP values tend to vary a great deal each time you run them.

What's the best way to handle this to get accurate numbers?

Rouncer · 08/17/09, 5:04 PM

Originally Posted by Izuul

I'm using the sim for the first time today (shaman is an alt), and i noticed that the EP values tend to vary a great deal each time you run them.

What's the best way to handle this to get accurate numbers?

Increase the number of hours. For accuracy you'll want it set to at least 5000 hours and the larger the number better. Also keep in mind that the bigger you set your EP test values the more accurate the results, even if they do lose some precision in the process.

Sylvand · 08/17/09, 5:49 PM

Originally Posted by Rouncer

Also keep in mind that the bigger you set your EP test values the more accurate the results, even if they do lose some precision in the process.

I think you have these reversed. If you set an EP value of 300 (for something), it should reduce the effect of RNG on the result but it may not be representative of what will happen if you add only 20 or 40.

Rouncer · 08/17/09, 6:25 PM

Originally Posted by Sylvand

I think you have these reversed. If you set an EP value of 300 (for something), it should reduce the effect of RNG on the result but it may not be representative of what will happen if you add only 20 or 40.

A bazooka being used against a small target is pretty damn accurate even if it isn't very precise. Large EP values will give you very accurate results, they just won't have any precision at all.

True, I could just be screwing up the terminology but that is how I've always thought of it working.

Synkronos · 08/18/09, 4:32 AM

I think what everyone's talking about here is consistent results

If you want the sim to give (nearly) the same results every time you run it for a given input set, you'll need to increase the number of hours, as Rouncer says.

For most things you don't really need the smoothing provided by large EP values. Haste is the obvious exception here with massive swings in value over a very short range. The rest behave pretty much as expected and change in value very gradually.

Katine · 08/18/09, 12:48 PM

I searched this thread to see if Blood of the Old God is implemented and i would appear from previous posts that it is as long as i use the item id (45522). However when i add that item ID im still getting Trinket 45522 not found when i run the sim.

Am i missing something?

Edit: Oops i forgot to update my path to the new enhsim.exe.

Sylvand · 08/18/09, 1:01 PM

EnhSim 1.8.3.2 is out

EnhSim - Release: v1.8.3.2

Adding 6 second cooldown to Totem of Electrifying Winds proc.
Hyperspeed Accelerators will now work with on use trinkets. They will trigger a 10 second trinket lockout and will be used with higher priority than trinkets.
Thunder Capacitor and Reign of the Unliving proc fixed; they will now discharge on the 4th and 3rd proc respectively, and their cooldown applies between each proc, not just after a discharge.
Added the following trinkets from triumph badges
- Mark of Supremacy (mark_of_supremacy, 47734)
- Shard of the Crystal Heart (shard_of_the_crystal_heart, 48722)
- Talisman of Resurgence (talisman_of_resurgence, 48724)

The EP exports file Levva added is also included.

Right now the next thing I foresee adding will be the new Deathbringer proc. If anyone manages to get their hands on this, please add as much information as you can here (or in the 3.2 thread, or PM me).

Alandovos · 08/18/09, 5:40 PM

As an FYI I've started working on decompiling and getting a new clean source copy of the EnhSimGUI so we can move forward. Once I've got a fully working version of the last release I'll start on getting updates done. Nothing is available for testing yet but I'm 90% on having the display source where it needs to be and just need the config reading/writing fixed up so it can compile. My estimate is about a week or so to get that done.

If you have suggestions for improvements beyond just getting new items to show up by name in the drop downs (vs the code #) either send me a PM or add an item to the work list at the codeplex site (EnhSim - Issue Tracker). PLEASE DO NOT REPLY IN THE THREAD.

mjgunn · 08/19/09, 4:12 AM

Just to triple check here....

To put an item that isn't in the list in the gui yet into the sim, say the mjolnir runestone, or the totem of electrifying wind, I just put the item ID #, say 47666 for the totem, into the totem spot? Want to make sure I'm really right on this because doing that shows an almost 200 dps increase from my deadly gladiator's totem of indomitability which seems like a huge boost for a single piece (not that I'd be complaining!).

Mengus · 08/19/09, 5:07 AM

If you look at config.txt file that comes in the latest version of EnhSim, you can see all the various totems, set bonuses, trinkets, etc. that are commented out.

If you're using the EnhSimGui, it won't be able to access them, but use the gui for what you can, then go in and manually edit a config file (make sure you backup the original) and you can insert the appropriate items in there.

Levva · 08/20/09, 9:56 AM

Originally Posted by Rouncer

A bazooka being used against a small target is pretty damn accurate even if it isn't very precise. Large EP values will give you very accurate results, they just won't have any precision at all.

True, I could just be screwing up the terminology but that is how I've always thought of it working.

However this approach is fundamentally flawed as you cannot say that the EP values for say +300 Haste is anything like the EP values for +10 Haste.

EP Values VARY depending on your gear. The most certainly VARY if you inflate their value. You CANNOT reliably use wildly varying EP values and have a hope of them giving accurate results for your current gear.

For example: we know that typical early Naxx EP values are very different from typical post Ulduar values. If you started with Early Naxx gear and adding huge EP steps what you are doing is pushing the results well past even T9 values. Do you seriously imagine that you are going to get good results?

EP values if they are ever of any use are only good for adding a small delta from your current gear and checking what effects they have. This is especially true of haste and the irritating effects of simulator precision where micro variants in hasted GCD effects can mean wild variants in what is theoretically possible for priorites. eg: dipping a GCD so that 14.00001 Magma totem drops is possible during a fight rather than 13.99999. A tiny tiny amount of extra haste can give a dps boost that you'd never actually experience in reality.

Using EP values should only ever be a manual short cut to actually trying out the current setup vs a proposed setup.

ie: if you sim your current gear get a set of realistic EP values and a dps total. If you use those EP values to suggest re-gemming/enchanting you SHOULD ALWAYS feed the proposed setup AFTER regemming/enchanting back into the sim to ACTUALLY CHECK if the dps went up.

Way to many times I've seen someone get EP results and say "EnhSim told me to re-gem for haste, so I removed all my AP gems and replaced with haste ones". They never tested the sim with the new gem config as often if they had the result would be LOWER dps. You cannot rely on EP values to give you anything other than general guidance as to what to go after. Actually running the sim with the new config is vital.

Rouncer · 08/20/09, 10:35 AM

Originally Posted by Levva

However this approach is fundamentally flawed as you cannot say that the EP values for say +300 Haste is anything like the EP values for +10 Haste.
[/b]

You are just saying exactly the same thing just with more capital letters. edit Simming with +300 crit rating will be pretty precise but your accuracy will be in the toilet so the results will be useless, just as simming with +1 haste rating would be incredibly accurate but the precision /edit would be garbage since any variation would be lost in general RNG so those results would also be garbage.

The concept is to find a compromise point between those two extremes and this is aided by increasing the length of time the sim is run for, to attempt to limit the amount of general randomness that enters into the results.

I did have an interesting thought though, based on your haste example. Maybe it would be a good idea to change combat length from a specific value to instead give it a range of values. So that instead of having the sim run 10 minute blocks of combat simulations for 10000 hours, it would run blocks that vary randomly in length from say 9:30 to 10:15 for those 10000 hours. That would more accurately simulate combat conditions where we have a general idea how long a fight will last but can't be that precise about it. This possibly could resolve some of the irregularities we are getting with haste values as well as limit any skewing of the results due to the influence of activated abilities. For example think about something like the hyperaccelerators. They have 12 second duration and a 1 minute cooldown or a 20% uptime but if the combat cycles were set to say 3:30 then they actually end up with a 23% uptime increasing their value. This gets even more skewed with something like the wrathstone or the Feral Spirits with their longer cooldowns.

Edited to fix as the concept was fine just had the terms mixed up.

Levva · 08/20/09, 11:58 AM

Random combat legnths an EXCELLENT idea, sorry couldn't resist given your caps comment.

It would be easily possible to randomly add say +/- 10% of the combat length. So if you set for 10 mins it varies between 9mins & 11 mins. If you set for 6 mins it varies between 5m 24s and 6m 36s. This would indeed smooth out the random effects of tiny haste value changes, due to fixed combat lengths.

Oh whilst I'm at it. Sylvand I was thinking that for EP value outputs it would be a good idea to include "Best Gem values" at the bottom of that so you list ...

Bright Cardinal Ruby (40 AP) : 40 EP
Delicate Cardinal Ruby (20 Agi) : X EP
Quick King's Amber (20 Haste) : X EP
Smooth King's Amber (20 Crit) : X EP
Runed Cardinal Ruby (23 Spellpower) : X EP etc.

The idea being that it smooths out the effect of the different item budgets and makes comparing easier.

Even better if you say listed top Red socket, Yellow socket, Blue socket gems by EP value. Which is possibly the most FAQ we get.

Yo! · 08/20/09, 1:48 PM

Originally Posted by Levva

Random combat legnths an EXCELLENT idea, sorry couldn't resist given your caps comment.
It would be easily possible to randomly add say +/- 10% of the combat length. So if you set for 10 mins it varies between 9mins & 11 mins. If you set for 6 mins it varies between 5m 24s and 6m 36s. This would indeed smooth out the random effects of tiny haste value changes, due to fixed combat lengths.

Random combat length makes simulation more "real" but may bring additional fluctuation to EP values and will not smooth EP results produced by tiny haste value changes that are caused by borderline cases of "swing time/windfurry cooldown" ratio.

Sylvand · 08/20/09, 1:57 PM

It shouldn't be difficult to add random combat length, though it will be an extra (optional) parameter in the config file so it might be tricky to use it with the GUI. Are you guys thinking uniform distribution or something like normal/etc.?

Accuracy vs. precision; I don't want to push to point all too much because it's coming down to a matter of semantics, but I would mention that the wikipedia article Accuracy and precision - Wikipedia, the free encyclopedia uses them in the same way Levva and I have. There's a good chance other people would use them the same way.

Malan · 08/20/09, 2:55 PM

Originally Posted by Rouncer

Simming with +300 crit rating will be pretty accurate but your precision will be in the toilet so the results will be useless,

I think you've mixed up those terms. As noted above in the wiki link, precision relates to the repeatability of the data. Repeated simulation with the same inputs produces fairly precise data - the output values of dps and EP values aren't going to dramatically change from one run to another as long as the inputs are fixed.

dedmonwakeen · 08/20/09, 8:03 PM

Originally Posted by Levva

Random combat legnths an EXCELLENT idea, sorry couldn't resist given your caps comment.

It would be easily possible to randomly add say +/- 10% of the combat length. So if you set for 10 mins it varies between 9mins & 11 mins. If you set for 6 mins it varies between 5m 24s and 6m 36s. This would indeed smooth out the random effects of tiny haste value changes, due to fixed combat lengths.

I've found that using random latency values accomplishes the same thing.

Also, if you are running multiple iterations of the same fight length, I would be very wary of how you end the fight.

DPS = Total_Damage / Time_Spent_Attacking

Time_Spent_Attacking can be defined in a couple different ways. Granted, this is more of an issue for characters with long cast-time nukes......

Another way to get slight variation in combat length is to limit the sim by target health.