This project is read-only.

Ep, or simulate?

May 12, 2010 at 4:58 PM

I've been using enhsim for a while, but I'm not sure I'm interpreting the data correctly, or using the program as it was intended. I use it mainly when I need to socket gems to make sure I'm getting the greatest dps increase per new gem socketed.

I start by running a standard “simulate” without changing my stats, which gives me a dps number at the end. Then I change the stats which the new gem would modify, and run another simulate. I do this for each of the possible gems I can socket (AP, haste, hit) and whichever shows the greatest dps increase, that's the one I socket. Is this the correct way to use it, or should I be calculating EP value and socketing whichever stat shows up with the highest ep?

The reason I ask is because the last time I had a free socket, I checked the dps increase, using “simulate” for hit, haste, and ap. Hit showed a slightly higher dps increase.

I then switched my stats back to their original values, and ran the “calculate EP values”. It told me that AP was at 1.0, hit was at 1.59, and haste was at 2.10. Those EP values seem to indicate that I should socket haste.

I then used “simulate” again to check on the dps output for haste and hit, and hit came out clearly higher. Am I doing something wrong? Which one should I use to determine which gem to socket, the dps simulation or the EP value?

May 18, 2010 at 11:27 AM
Running multiple sims is the correct way. Using EP values is very very definitely a VERY wrong way of doing it. You will likely get the WRONG answers if you use the meaningless crap that is EP values.
May 19, 2010 at 12:59 AM
Haha, so this is why my issue was closed. I have a question: EP values are supposed to show the relative value of a stat, correct? They are based on the DPS increase you would see if you added more of that stat, which is done by running the sim with that new value, right? Why are they meaningless crap? If they are crap, why are they still there? Thank you!
May 21, 2010 at 12:16 PM
Edited May 21, 2010 at 12:17 PM

An EP value shows you what you get by adding exactly X of a stat to your existing setup. It does not follow that if you have an EP value of 1.56 for Agi and 1.67 for crit that if you swap an item that removes 30 crit and adds 30 Agi instead that your dps will go up. As the EP value was for ADDING not removing crit. Plus the EP value was for adding 40 Agi not 30. So not only is the value telling you nothing about removing a stat it tells you nothing about the difference between adding 30 of a stat vs 40 of the same stat.

In WotLK we reached a stage where adding slightly more of a stat did not give the same increase as the previous amount added. ie: We discovered lots of stats ended up being stepped, and we encountered various caps. This is particularly true of haste where tiny variations in haste arrived at various "sweet spots" and "sour spots" where adding more had little dps increasing effect, then adding even more suddenly produced a big boost.

EP values just don't work when you have caps, and stepping values like this. They only work when there is a smooth predictable path and adding 20 of a stat gives exactly double the boost as adding 10 of a stat. Similarly removing 10 of a stat should remove the same dps as adding 10 of a stat adds. In WotLK that rule did not hold true at all, largely due to haste effects, caps and proc rates for Maelstrom weapon.

Therefore EP values became somewhat meaningless.

Some people still clung on to them, however they found that if they simmed out the result vs using EP values that frequently the sim results showed a different answer to the EP results. At that point they realised the need to ditch using EP values as an old fashioned unreliable means of gearing.

Where EP values still work is to give a rough guide as to what sorts of stats are better. Ideally you would run EP values with LOTS and LOTS of different step values but it takes forever running it once so people don't and they leave the defaults. This means you see the effect of adding a SINGLE value and if there is a change just before or after that value you don't see it. Imagine a perfect graph of the dps with a line for each stat showing exactly what happens adding all sorts of combinations of adding and removing stats. If you look at that graph you see the whole picture (this graph is available in Rawr). Now imagine that instead of seeing the whole graph you were only allowed a peek at what the value was at a single spot on the graph. This is what you are getting with EP values you see a single dot on the graph and not the whole picture of how the stat changes over time.

This is why thinking EP values can tell you how to gear is wrong, and why I described them as meaningless crap. They don't give you anything like the full picture.

May 21, 2010 at 3:38 PM
Levva wrote:

In WotLK we reached a stage where adding slightly more of a stat did not give the same increase as the previous amount added. ie: We discovered lots of stats ended up being stepped, and we encountered various caps. This is particularly true of haste where tiny variations in haste arrived at various "sweet spots" and "sour spots" where adding more had little dps increasing effect, then adding even more suddenly produced a big boost.

 

Are those "sweet spots" and 'sour spots" consistent with a given amount of haste for players regardless of other stats, or does it vary, depending upon gear?  If it is consistent, is there a graph anywhere that shows these "sweet spots"?

Jul 1, 2010 at 12:27 AM
They are entirely random and utterly dependant on your own gear, what buffs you have at a particular moment etc. So during the course of a fight you WILL hit a sweet spot and you WILL hit a sour spot as various temporary procs happen. The amount of time you are in a sweet/sour spot will vary. ie: At no point do stats increase uniformly now, so at no point does using EP values give you meaningful results.