I was bored for two days and made an Elemental module for Rawr (I made Tree before so I knew what to do so it could be done rather fast) let me know if the results (suggested gear, relative stats values) are good, I've focused on making a good estimation rather than a thorough and 100% accurate DPS estimation (partly because Rawr is not a DPS calculator but a gear optimizer, so if performance can be improved by losing DPS accuracy but keeping gear choice accuracy that's fine)
It was originally meant to give whoever was going to develop it a good starting point, but when I was working on it things were going so fast I thought I could just as well get a working version out before 2.1.9.
The current version uses only the Flame Shock (glyph of flame shock is mandatory at the moment), Lava Blast and Lightning Bolt.
Well, something is majorly up with your stat valuations.
You may also want to make a new thread about rawr.elemental rather than continuing to use this thread.
Last edited by Binkenstein : 02/01/09 at 11:25 PM.
I was bored for two days and made an Elemental module for Rawr (I made Tree before so I knew what to do so it could be done rather fast) let me know if the results (suggested gear, relative stats values) are good, I've focused on making a good estimation rather than a thorough and 100% accurate DPS estimation (partly because Rawr is not a DPS calculator but a gear optimizer, so if performance can be improved by losing DPS accuracy but keeping gear choice accuracy that's fine)
It was originally meant to give whoever was going to develop it a good starting point, but when I was working on it things were going so fast I thought I could just as well get a working version out before 2.1.9.
The current version uses only the Flame Shock (glyph of flame shock is mandatory at the moment), Lava Blast and Lightning Bolt.
It looks to me as if it is massively overvaluing the Valorous & Heroes' gear (by a factor of about 10) - are you sure you have modelled the set bonus correctly? Also, your Chain Lightning is hitting very hard - is it modelled on multiple targets? On the other hand, FS DoT damage looks a little low.
Finally, it seems to be rating MP5 very highly which may be an indication that something is wrong with the Clearcasting model.
let me know if the results (suggested gear, relative stats values) are good
The Lava burst crit calculations are bugged. It's reporting 20k LvB crits normally and 65k crits with the 4 set bonus, which is probably why the set bonus is so highly valued.
OK, are the lava burst crits (20k) without the set bonus correct? (my ele shammy isn't 80 yet)
I've fixed the set bonuses for T7 gear, they were indeed wrong (they were a percentage rather than a floating point so I had to divide by 100 which I forgot)
2-piece modelled by multiplying mana cost by .95
4-piece modelled by increasing crit modifier by .5 * .1 (.5 because spell crit is normally +50% damage)
For example, if you crit modifier was 2 (1.5 + .5 from Elemental Fury) then it will be 2.05 with 4-piece.
Chain Lightning has a tooltip that shows the damage for 3 and 4 targets. The calculation was indeed wrong, it is fixed now (for the next version) (it added the damage on the first target twice)
You can see how the spells are calculated by viewing the Spells.cs file in the Rawr.Elemental directory in the source code.
I will look at the clearcasting model again. I'll add clearcasting %s and mana costs to the tooltips. Right now, I calculate the chance one of the last two spells critted (1-(1-crit*hit)(1-crit*hit)) and apply that, assuming for flame shock and for lava burst the chances for two lightning bolts, and for lightning bolts there's a slightly more complex calculation which includes that the first two lightning bolts have a higher chance due to lava burst always critting.
Yeah I used the spreadsheet too, the only thing I was wondering is where the 0,545 comes from. I use the same formula, except apparently I managed to copy some of the old incorrect code from lightning bolt over to lava burst, pretty strange since that normally doesn't happen :S. Well, we're all human and it's fixed for the next release. Calculating for an average mix of naxx gear buffed by moonkin, pally, priest and mage buffs I get:
Lightning Bolt 3.5k hit 7k crit, 34.77% crit 68% clearcasting
Chain Lightning 3.2k hit 6.2-6.6k crit
Lava Burst 4-4.4k hit 8.5-9.4k crit, 29.77% crit 56.75% clearcasting
Flame Shock 1.2k hit 2.4k crit, 29.77% crit, 591 every 3 seconds, 56.75% clearcasting
BTW those are average numbers and include CC multipliers
FS dot is calculated with 10% from spell power based on Spell power coefficient - WoWWiki - Your guide to the World of Warcraft, increased by Storm, Earth & Fire (adds 60% to dot) Base tick damage is 139, thus we get
1.6 * (139 + sp*.1) with 2110 spell power this is 1.6 * (350) = 560 damage, further increased by buffs on the target.
Regarding MP5, it is only valued high if you go OOM during the fight. Try enabling some buffs and play a bit with the fight duration, you'll notice that there's a point when MP5 suddenly becomes very important, that's when you've gone OOM.
Since it's an estimation, here's how it works: it calculates
- damage done until OOM
- time to burn all mana
- time to regen all mana
From the last two variables, we get the cast vs regen ratio.
Then we get
BurstPoints = DPSuntilOOM * (bvs)*2f
SustainedPoints = TotalDPS * (1-bvs)*2f
bvs is burst vs sustained and goes from 200% burst to 100%/100% to 200% sustained
Obviously a thorough simulation will give much better results, however, the performance will be very bad and it remains a question if the results (gear choices) are different. What about trinkets by the way, are they what you expect?
Yeah I used the spreadsheet too, the only thing I was wondering is where the 0,545 comes from. I use the same formula, except apparently I managed to copy some of the old incorrect code from lightning bolt over to lava burst, pretty strange since that normally doesn't happen :S. Well, we're all human and it's fixed for the next release. Calculating for an average mix of naxx gear buffed by moonkin, pally, priest and mage buffs I get:
Lightning Bolt 3.5k hit 7k crit, 34.77% crit 68% clearcasting
Chain Lightning 3.2k hit 6.2-6.6k crit
Lava Burst 4-4.4k hit 8.5-9.4k crit, 29.77% crit 56.75% clearcasting
Flame Shock 1.2k hit 2.4k crit, 29.77% crit, 591 every 3 seconds, 56.75% clearcasting
BTW those are average numbers and include CC multipliers
FS dot is calculated with 10% from spell power based on Spell power coefficient - WoWWiki - Your guide to the World of Warcraft, increased by Storm, Earth & Fire (adds 60% to dot) Base tick damage is 139, thus we get
1.6 * (139 + sp*.1) with 2110 spell power this is 1.6 * (350) = 560 damage, further increased by buffs on the target.
Regarding MP5, it is only valued high if you go OOM during the fight. Try enabling some buffs and play a bit with the fight duration, you'll notice that there's a point when MP5 suddenly becomes very important, that's when you've gone OOM.
Since it's an estimation, here's how it works: it calculates
- damage done until OOM
- time to burn all mana
- time to regen all mana
From the last two variables, we get the cast vs regen ratio.
Then we get
BurstPoints = DPSuntilOOM * (bvs)*2f
SustainedPoints = TotalDPS * (1-bvs)*2f
bvs is burst vs sustained and goes from 200% burst to 100%/100% to 200% sustained
Obviously a thorough simulation will give much better results, however, the performance will be very bad and it remains a question if the results (gear choices) are different. What about trinkets by the way, are they what you expect?
Are those crit percentages talented? Because they seem very low to me With 10% from Scorch, 3% from ToW, 5% from Thundering Strikes, 5% from Call of Thunder and 5% from Elemental Oath (or the Boomkin Buff), that's 28% on Lightning spells without any gear bonuses which seems to indicate that you are only adding 6.77% from gear. If you haven't included for talents then it seems to be roughly right at 10% lower than talented.
Your FS Dot calc looks right but on Rawr it is giving me 386/DoT with 2372 SP when it should be 602 plus raid damage buffs (698 per tick) so maybe there is an error in how you have entered them into the code. The other spell damages are looking about right though I haven't checked them in detail yet.
I can see how you are calculating the mana situation but in practice an elemental shaman using no Chain Lightning in his/her rotation will struggle to go OOM (actually they would struggle to go below 90% mana) in a raid buffed situation and if they did they would T/Storm rather than opt for MP5 gear. I did play with the fight length and at by reducing the fight length from 5 mins to 4 the gear priorities did seem to slip into a more expected order. I'm not convinced there isn't something wrong in there somewhere though because even using Chain Lightnings in my rotation I am unlikely to go OOM in 5 mins.
Regarding the trinkets, I think there is something not quite right. It seems to be value crit trinkets higher than I would expect and the Thunder Capacitor lower than I would expect. I think these are things you might expect if your overall crit percentage was lower than it actually is which is one thing that made me query whether your crit values included talents or not earlier. Also, have you accounted for The Thunder Capacitor charging on Lightning Overload crits? I'm not saying you haven't, just throwing some potential queries out there.
Anyway - keep it going - I've been dying for a Rawr Elemental for so long now
Oh if I turn all those crit buffs on (I did a very fast calc) I actually get 47% crit
I'll add an option for Thunderstorm.
Maybe you need to check if the MP5 values in basic stats are what you have too and that the mana costs of the spells are right.
Though, as soon as your mana regen and your mana burn per second are equal, you'll see you stay at 90% or higher. With buffs like BoW and replenishment this isn't too hard.
Finally, I'll check if the Lightning Overload calcs are correct. Does a Lightning Overload proc consume clearcasting?
Is there any reason to evaluate spirit on gear, since we have no use(excluding OS5SR, which we virtually never get)? We have no spirit -> spellpower and no '% regen while casting'. I have no idea how much weight you're giving to spirit and whether or not it would change the values, so this may be a silly question.
If you go out of mana, you could do two things
1. cast a spell as soon as you have mana. You might stay inside the 5SR rule this way.
2. wait for a decent amount of mana, then burn that mana. You will go outside the 5SR rule, spirit regen will have an effect.
If you don't go out of mana, spirit and mp5 will have no value. With good gear and using thunderstorm everytime it's up, you don't go out of mana at all at.
Maybe I could add a slider of some kind that says "Outside 5SR time% when regenning mana" which you could set to 0 (e.g. when regenning, you cast a spell every 5 seconds or something) so spirit really would have no value at all.
Besides, spirit has such a low value if you go OOM, that gear really needs to have TONS of spirit to be better than the rest, which it doesn't. It doesn't mean it's bad to give spirit a value. In fact, the value is probably pretty accurate.
One more thing that might clarify things for you. The way Rawr works is like this: we calculate all stats and use some kind of model to predict how good your gear is. This model is based upon values like expected dps, expected healing done, expected damage absorbed. For these calculations, we use actual formulae instead of some kind of "stat X is worth Y" trick. Sometimes you see a spreadsheet or a website use formulae like "every crit% is worth 10 points and every AP is worth 1 point" or something similar, and calculate gear value by adding up these numbers. This is not how Rawr works. The Rawr.Elemental model uses expected average dps and dpf (damage per fight) as an indication of how good a piece of gear is. Hence you will see that if you selected the right (strange) talents, even strength and agility can increase dps.
If you go out of mana, you could do two things
1. cast a spell as soon as you have mana. You might stay inside the 5SR rule this way.
2. wait for a decent amount of mana, then burn that mana. You will go outside the 5SR rule, spirit regen will have an effect.
If you don't go out of mana, spirit and mp5 will have no value. With good gear and using thunderstorm everytime it's up, you don't go out of mana at all at.
Maybe I could add a slider of some kind that says "Outside 5SR time% when regenning mana" which you could set to 0 (e.g. when regenning, you cast a spell every 5 seconds or something) so spirit really would have no value at all.
Besides, spirit has such a low value if you go OOM, that gear really needs to have TONS of spirit to be better than the rest, which it doesn't. It doesn't mean it's bad to give spirit a value. In fact, the value is probably pretty accurate.
One more thing that might clarify things for you. The way Rawr works is like this: we calculate all stats and use some kind of model to predict how good your gear is. This model is based upon values like expected dps, expected healing done, expected damage absorbed. For these calculations, we use actual formulae instead of some kind of "stat X is worth Y" trick. Sometimes you see a spreadsheet or a website use formulae like "every crit% is worth 10 points and every AP is worth 1 point" or something similar, and calculate gear value by adding up these numbers. This is not how Rawr works. The Rawr.Elemental model uses expected average dps and dpf (damage per fight) as an indication of how good a piece of gear is. Hence you will see that if you selected the right (strange) talents, even strength and agility can increase dps.
Actually, even with less than medicore gear(Blues, old 70 gear ect) You'll probably never see your self going out of mana if you use thunderstorm when it's off CD, maybe mana pot and don't use chain lightning.
It's starting to be a while, but I can't remember having any mana problems when I hit 80 in my 70 gear and started raiding again.
I'm pretty sure there are issues with calculations - exactly what I can't work out but Lava Burst appears to be the culprit.
In good, but not BiS gear I'm listed at doing 11k DPS, which I'm pretty sure is wrong. Examining the data shows the Lava Burst crits doing 10k-20k, when they do more like 8k-11k.
I'm also pretty sure that it's not calculating mana usage correctly, as under normal raid conditions it's pretty much impossible for me to OOM, and that's without using Thunderstorm or a Mana potion. 1 mana spring totem, and a FS, LvB, LBxn, LvB, LBxn, rotation.
Pewsey has heard about tact and discretion, but tends to regard them much as children view vegetables.
There are only two kinds of MMOs: the ones people complain about and the ones nobody plays. (inspired by Bjarne Stroustrup)
pewsey, the Lava Burst calculations have already been fixed.
I'm also considering adding an optional "rotations" simulator, but this involves a bit more work.
(Most of that work involves trinkets, procs and keeping high performance; simulating spellcasts, EM and CC is very easy and in fact already implemented)
Awesome, what is the best way to update my 2.1.9 to include this new functionality. I've browsed the RAWR website, but there appears no way to update individual DLL files other than by compiling from the patched sources.
Pewsey has heard about tact and discretion, but tends to regard them much as children view vegetables.
There are only two kinds of MMOs: the ones people complain about and the ones nobody plays. (inspired by Bjarne Stroustrup)
Today, I've done a lot of work on a rotation simulator. It's pretty much complete (not checked in yet for those of you who compile source code), I'd like some feedback on some observations, since I do not have the luxury of a lvl80 shaman myself yet.
Rotation is like this:
1. Try to use any trinket
2. Try to cast Elemental Mastery
3. If under 60% mana, try to cast Thunderstorm
4a. If Flame Shock DOT is up, try to cast Lava Burst
4b. Try to cast Lava Burst
5a. If Flame Shock DOT is not up, try to cast Flame Shock
5b. Try to cast Flame Shock
6. Try to cast Lightning Bolt
7. Wait 100 milliseconds (apparently we're out of mana)
The decision between 4a and 4b is configurable.
The decision between 5a and 5b is configurable.
I get the best results with 4b and 5a, using Glyph of Shocking, Glyph of Lightning Bolt and Glyph of Flametongue/Lava. However, after adding many raid buffs, the best results are with Glyph of Lightning Bolt, Glyph of Lava and Glyph of Flame Shock
...
4. If flame shock DOT is not up anymore: try to cast Flame Shock
5. If flame shock DOT is not up for more than 1.5 sec: try to cast Chain Lightning
6. Try to cast Lava Burst
7. Try to cast Lightning Bolt
...
This is the priority that many elemental are using.
A little bit more simple is :
...
4. If flame shock DOT is not up for more than 1.5 se: try to cast Flame Shock
5. Try to cast Lava Burst
6. Try to cast Lightning Bolt
...
is rawr grossly overvaluing hit? I was playing around with my gear in it, and it continued to tell me to upgrade with +hit items, gems, and enchants until i was well over 100 above the hit cap.
Would you mind adding a tooltip to +hit saying how much you need to cap or how much over the cap you are(including buffs if they are checked)? Also the same thing for +haste(show the haste %) including the totem if checked. You could also add the cast time to the tooltips for the individual spells. This would make it much easier to tweak out rotations and such.
01/31/09, 6:12 AM
Similar Threads
