The method I've used to model GCD interference is rather "spiky" at present, see new graph function for details. I'm investigating methods of smoothing this out. The issue is that Rawr models GCD interference over a single combat length at present whereas of course EnhSim does it over several thousand hours.

What I need to be able to do is to find a sensible way of averaging out the number of ability uses so that it smooths out the "spikyness". The issue you are seeing is almost certainly because Rawr is hitting one of these spikes and giving odd results.

The spikes affect Hit (gives more MW5_LBs in shorter time), haste (affects GCDs), Intellect (if mana considerations used can adjust when an ability - notably SR - is used).

The fact that these 3 stats affect the order you use abilities in combat means that the total number of uses can change, eg small amount of extra haste means an ability comes off CD a tiny fraction of a second earlier and thus gets used first changing the entire order of abilities that follow, thus affecting dps.

For this reason the model is currently rather reactive to haste/hit/intellect effects. I'm investigating a way to smooth this out.

Re: Rouncer's issue. I could add extra tick boxes to choose what to export. I'll see what I can do in time for next version.

Yeah the downside is that because I exactly mimiced a priority system in a fight you get the extreme spiky nature. Unfortanately this makes Rawr very unpredictable for Enhancement. As tiny changes eg: in Average Lag *can* (not will) radically change gear recommendations. Since this is counter intuitive, I've changed the methodology for v2.2.25 to average out this spikyness over a longer timeframe. Similar to EnhSim running for longer number of simulated hours.

The result is a almost totally smooth graph. It also means the numbers are a lot closer to EnhSim now. I'm planning on doing lots of "sanity checks" soon to check exactly what numbers Rawr is giving and what the same numbers are in EnhSim to see side by side where variations are and to work out why those variations are there.

For instance for my own gear level I'm seeing a 0.24% difference in raw white crits. Its a very constant 0.24% difference across the board, which leads me to suspect that somewhere in the calcs for EnhSim or Rawr there is a difference. Finding it is an issue. What I really need is alternate characters testing to see if this is a consistent 0.24% variance across any character.

If you need help testing Levva, I'd be happy to help as much as I can. It'd be useful if you had a short testing guide though to ensure a standard testing procedure is followed.

In order to test you first need to have the very latest source code and a copy of Visual Studio (Express edition the free one from Microsoft is fine). You cannot test what I need tested if you can't debug the code as I need to know at what point the calculation of crit chance is different in Rawr vs EnhSim.

What you could do without the source, is to run various characters and note the variance between the figures for EnhSim and the figures for Rawr.Enhance v2.2.25 (v2.2.24 is no point using as its radically changed in v2.2.25 with new priority smoothing routines).

What I need to know is not individual variances but patterns. eg: Load one char in Rawr export to Enhsim and note the following : DPS totals, hit/crit percentages, PPM values for each individual ability in both Rawr and EnhSim. Then try other characters each time seeing what the difference between the two sets of data is.

Now I know there will be variances what I need to help pinpoint issues is where there is a consistent variance eg: white crit rate is always 0.24-0.30% lower in one than the other. Or Flametongue dps is always around 10% higher in one than the other, or the PPM rate on Earth Shock is always around 1.2 PPM lower in one than the other.

Its the patterns that matter as these point to where the two sims might be using different calculations. eg: If I note that the reason for a 0.25% difference in crit rate is due to EnhSim using 2.03 times crit damage multiplier rather than 2.06 crit damage multiplier for meta gem bonus crit. That might account for it. Then either Rawr or EnhSim can be fixed to remove the bug.

Do you have any other standards for testing you're looking for? Standard spell priority? Standard hours of simulation time in enhsim? A recent tester showed 0.9 SD for 5000, 0.3 for 10000. Standard combat time rng factor for enhsim? My crude results suggest anything from 35% to 100% is fine (but an agreed standard percentage must be set!); without scripting automation I can't give you the granularity and depth you have asked for to verify.

I'd suggest 5k hours sim, no mana calcs, say 10% combat time rnd factor - I seriously dislike random factors of 35%+ as combat in raids does not last 10 mins +/- 3.5 mins or between 6.5 and 13.5 minutes that's just way too variable. 10% gives 9 to 11 mins which seems within the margins although I'd perhaps suggest a default of 8 mins +/- 10% as that feels closer to what we are seeing now.

Either way it really doesn't matter as long as Rawr is set the same as EnhSim and once set isn't changed for the duration of the tests. What I usually try to do is load up the default base config then setup Rawr and export then compare the results side by side.

From that I'm seeing the 0.25-0.3% variance in crit damage so its an area to focus on, I believe Flametounge damage might be high and Earth Shock on the low side but that seems less consistent.

The point is that regardless of setup in order to establish the cause someone needs to be stepping through the code and checking what actual values are being used in each formulae and comparing and contrasting the formulae to see if anything has been missed or calculated differently in error.

I spent 10:36 on testing MT and watching for ED at the training dummy in Org. in the 10 min, spamming the MT (sr when I was low on mana) whenever the cd was up (there was a few points where I was alt tabbed and missed back to back, I know its not a whole lot of time/testing, but from what I can see, MT is periodic dmg and does not trigger ED. I can test more as I have a lot of time on my hands these days.

WoL: World of Logs - Real Time Raid Analysis

Ok Thanks I'll test myself again this evening. Can anyone assist then in giving a definitive list of things that proc ED please. I verify that against my own testing.

So is there a conclusion? Does FS implement our dmg or is it better to still use ES ? Sorry if i am a bit lazy to read all pages, but it only makes me confused, after certain posts i space out

I love the new graph feature in Rawr. Here's the graph I get in my current gear:

http://img94.imageshack.us/img94/5272/murmorepgraph.png

Does anyone know what might be causing the shape of the graph for hit rating? In case the imageshack link stops working. I'm at 555 hit rating. Going on, the graph shows hit as being marginally the best stat for about 50 more rating. It then craters to having a negative effect, before shooting back up again around 80.

This is the effect of the old v2.2.24 priority queue handling in Rawr that was just far too precise. So when you hit a threshold and suddenly 1 extra hit means you get marginally more LB procs per minute all the priorities shift around and the number of attacks per combat change dramatically. The problem with this approach is that it was just way way too spiky. So for v2.2.25 I've implemented an averaging approach where the number of attacks will not dramatically change by adding 1 hit. This should mean a lot less oddities in gear suggestions, and a lot more stable results. Sadly it also means a lot less interesting graphs.

Still.. i didn't get a proper answer ^_^, is FS more efficient on my dps than ES? What about using FS and then spam ES till you have to renew FS?

errror in posting

This question has been answered many times throughout this and other threads: FS is a DPS increase as long as you're not clipping the DoT. The shock "Rotation" is FS, ES, ES, FS, Rinse and repeat. (of course if you have reverberation this changes)

I apologize if this is redundant, I'm only a part-time shaman and was looking for some answers/maybe don't understand exactly how some elements of the class work.

I've been using Rawr to help with my gearing choices and it seems to always favor me carrying a hit trinket even though I've gone way beyond both soft hit caps (as I've come to understand them). I have essentially all raid buffs marked as active and it's still pushing me to around 483 hit. I understand that hit is still helping my off-hand but I would think that a trinket giving 200+ ap (like fury of the five flights) or another trinket filling in my expertise needs (like the victor's call) would be better since I've passed the threshold of the common hit caps. Instead it seems to prefer using 2 jewelcrafter's gems and an epic to fill in for the missing expertise and to just go on riding right past that hit cap.

Am I missing something or is this proper? Everywhere I look, other stats are written out as favorable to amassing hit this high.

Hit doesn't have a special effect on your off hand so not sure what that comment is about. Where hit scores beyond the spell hit cap is that it gives you extra white attacks landing which is more raw damage. However this also means more Maelstrom Weapon procs = more LB = more dps and it means that you get extra AP from more berserking procs, higher uptime on berserking.

It does still seem to over play hit though so I'm trying to fine tune the calcs to find out why.

Your expertise comment seemed odd, as expertise is the number one thing to cap long before you worry about capping hit, if you aren't expertise capped then you have far bigger problems than worrying about hit being over cap. Have a read of *Enhancement Shaman FAQ (3.2.2)*

So I got Rawr 2.2.25 and it doesn't seem to model some trinkets at all in the Enhancement module anymore, namely Comet's Trail. It's sitting at the bottom with Empty. I double checked in 2.2.24 and it was fine. Is anyone else seeing this? I tried out a different module (rogue I believe) and Comet's Trail had some value.

If anyone else has seen this I'll go ahead and log a bug in the tracker.

Please don't log bugs about data issues, unless you have first done a refresh of your data from Armoury and/or Wowhead. Data is entirely local to your PC. Comet's Trail is correctly modelled it just seems to be missing the proc in your data. This suggests one of two things..

1) When you installed v2.2.25 you didn't install to a fresh directory, and kept your old ItemDataCache.XML file.

2) Some data corruption of your ItemDataCache.XML that hasn't recognised the proc effect.

Either way a refresh of item cache from armoury/wowhead will fix most of the issues. You can test this by trying the new 3D modelling I added, under the Tools menu. Does it say you are missing display info? If so then you would be well advised to do a full Update Item Cache From Wowhead.

Version 2.2.26 - New General Release
Revised GCD interference model - no longer use unstable priority system
Added optional filter on professions so that enchants for professions you don't have don't show up
Added Option to view your character in 3D.
Implemented T10 bonuses
Added display of T10 set bonus uptimes
Tweaks to ED uptime calcs
Small tweaks to flurry calcs
Implemented independent uptime displays for MH & OH Enchants
Changed WF to use a stationary distribution model - NB. This is the change reverted for v2.2.27
Added extra uptime stats and tweaked special effect trigger conditions
Added Average AP & Average SP display to front panel tooltips.
Modified abilities to have some GCD interference chances
Added gemming templates that don't respect socket colour
Modified LS spellpower co-efficient
Fixed WF PPM display
Fixed Feral Spirits crit rate
Added a PPM display to assist diagnosing how much GCD intererence affects dps
Altered priorities to use .25 sec queueing
Removed Amplify magic as relevant buff
Added Intellect to stats graph and amended graph to up to 12 stats.

Version 2.2.26.20398 - Special Release to fix WF issue
Fix for WF issue showing fast MHs as good

Download Links

Main Rawr Website for instructions and usage. Download of Rawr main application : v2.2.26 release

Rawr.Enhance.zip - extract to Rawr directory overwriting existing files

For clarity - Download the main application, then download the zip file and extract the files to your Rawr directory overwriting the existing files with the newer versions. Then and only then run the application and confirm the version number on the splash screen and on the Enhancement module page (see graphic below) matches the latest version.

Picture shows how to find the version number.

1) I always install to a fresh directory.

2) This may have been the issue. It wasn't even picking up the static effect though which was especially strange. The reason why I was ready to point towards bug is that other modules were getting the proc and static value correctly. Either way it fixed itself in 2.2.26.

Version 2.2.27 - New General Release
Fix for invalid WF calc that allowed fast MH weapons to appear good
Fix for magma totem not always 100% uptime.
Fix minor display issue for Magma totem hit/crit percentages.
Added Fire Totem Uptime display
Tweak to WF proc rate to bring it more into line with EnhSim outputs
Added Black Magic Enchant (the patch 3.3 haste variant as the old one is poor)

I've added the two new relic slot Totems to Rawr.Enhance, for my kit the Enhancement one is a small upgrade on my current QE totem. The Ele version was similar but slightly behind the Enhance one.