Blood Pact: Using SimulationCraft for gear selection, page 2

Scale Factors
So, scale factors, what the devil are they? These are the relative value of each stat; in the same way apples, oranges and bananas don't cost the same as each other, not all stats are equally good at increasing your DPS. Spirit is nice to have and gives a decent damage boost but not as good as spell power. Hit typically gives the highest DPS gain of any stat but only until you reach the cap and then it's worthless. You can pick up many opinions on what are the respective values of each stat on any number of forums and websites but none of them are tailored to you -- the scale factors you just generated are.
If, for example, your scale factors showed that spell power had a value of 2 and crit 1 then SP is twice as important to increasing your DPS as crit. You would need to find an item that increased your crit by 20 to be as valuable to you as one that increased your SP by 10. Hit can be somewhat misleading in these things though. By default SimulationCraft considers the DPS gain from a gain in each stat except hit, where it works the value out by considering the DPS loss from losing some hit. This way you can consider swapping out some hit items or gems for SP, haste etc. and know if it's worth doing. You can get the app to not treat hit this way by selecting to "use Positive Deltas Only" in the scaling tab of the options.
If you have reached the hit cap then gaining more is worthless but losing some of what you have could be a bad move. This can be hard to judge with scale factors alone and this is where having SimulationCraft plot a graph of DPS gains for each stat is frankly, awesome. Call me a bookish nerd if you like but I love the DPS scaling chart, it just makes the whole issue of stat importance so simple.
DPS Gains chart
If you find this chart a little daunting take a little time to look at it. DPS is shown vertically, so the higher up the chart the better the DPS. Left to right is either an increase or decrease in each stat with your current value in the middle. So the chart starts on the left with showing what your DPS would be if you lost 200 points of each stat -- not all of them together, just which ever one you are looking at. So if you look at the purple spell power line and see where it is furthest left, that gives an idea of what your DPS would be if you lost 200 points of SP.If you look at the chart at the top of the page you can see that most of the stat lines are fairly straight. The red one for hit however rises more steeply than all the others and then flattens off. The point at which it flattens is the hit cap -- the point at which gaining more hit (moving right on the chart) doesn't increase DPS (moving up on the chart). You can see that this cap is about 50 points below their current level. Also in that example chart you can see that the steepest line on it is the spell power one - both for gains and losses of the stat. This means that this character would gain the most by gaining SP and lose the most by losing it. Very handy to know when choosing gear upgrades.
One final little tip for helping with gear choices. You may have noticed that links were provided to Wowhead and Lootrank. These links are based on your scale factors and will show a list of gear for each slot and the total value of each item. Both these sites do the job very well, I like Wowhead for it's filtering and tabbing of the slots and I think the way Lootrank allows you to baseline your own gear is fantastic. I would strongly recommend people try them both and see what they offer. Personally I use Lootrank as they attempt to include proc bonuses for trinkets and weapons but you have to be careful here as it assumes they will always proc to the maximum they can, which isn't going to be the case for those that trigger from "your healing spells". Neither of these sites will include set bonuses in their scoring either but don't let that put you off, just factor it into your consideration when reviewing them. Don't forget you can try out upgrades in the Wowhead profiler and then import them into SimulationCraft to try them out -- SimulationCraft does factor in set bonuses.
SimulationCraft offers many more tools and toys to play with and really takes the strain out of number crunching. As Blizzard introduce more proc-based mechanics into the game it is tools like this that will have a better chance of modeling them than static calculators and spreadsheets. It's certainly not a perfect tool and it's definitely not going to magic you into a better player. What it does though is make analysis a whole lot quicker, easier and more measurable.
Blood Pact is a weekly column detailing DoTs, demons, and all the dastardly deeds done by Warlocks. If you're curious about what's new with Locks since the last patch, check out WoW.com's guide to patch 3.3 or find out what's upcoming in Cataclysm from the BlizzCon 2009: Class Discussion Panel.
Patch 5.3 interview with Ghostcrawler
Mystery of the Unborn Val'kyr
The latest patch 5.3 news
All of the latest Mists of Pandaria news





Reader Comments (Page 2 of 4)
Grodu Feb 1st 2010 1:36PM
apologies for wrong section
Motecuhzoma Feb 1st 2010 1:56PM
@Grodu are you REALY trying to open a .exe on a mac?? thats windows, try downloading the .dmg version :)
Grodu Feb 1st 2010 2:09PM
Already been downloaded the program, I ment .app. sorry for the confusion
Dominic Hobbs Feb 1st 2010 2:20PM
There is a glossary of these terms in SimulationCraft -> Results -> Legend.
The ones you specifically ask about are:
DPE: average damage-per-execute
DPET: average damage-per-execute-time, the amount of damage generated divided by the time taken to execute the action, including time spent in the GCD
DPR: average damager-per-resource
M%: percentage of executes that resulted in misses
D%: percentage of executes that resulted in dodges
P%: percentage of executes that resulted in parrys
G%: percentage of executes that resulted in glances
T-M%: percentage of ticks that resulted in misses
mousewrites Feb 1st 2010 1:54PM
Agreed... so is there any yardstick/sim program that will tell me how i'm doing on 5 mans? Or are the 5 mans just too varied for sims to work correctly?
Dominic Hobbs Feb 1st 2010 2:14PM
You can make SimulationCraft model something closer to your 5 man circumstances by making sure you 'switch off' all the buffs and boss debuffs that will not be present in the fight you want to model. Also, reduce the fight length if appropriate and if the fight is chaotic (lot of moving about etc) then select the 'Helter Skelter' option from the Fight Style option.
You might also want to play about with the Player Skill option. I'm not sure what the code does to implement a skill setting but things like clipping DoTs and fouling your rotation (priorities) will have a profound impact on your actual DPS.
If you want to see if you can match SimulationCraft numbers then set up a simulation that would match you on a target dummy (no buffs or debuffs you can't have there) and try to meet that. You will also have a calmer environment in which to get your casting right or see if you mess it up.
Then there is also the option of reverse engineering your performance by uploading your combat log to a site such as World of Logs. From there you can see if you cast as often, DoTs ticked as much, your crit rates and power of your spells. All of which you can compare with your SimulationCraft report.
mousewrites Feb 1st 2010 2:36PM
Thank you! that will help a lot. I've tweaked SimCraft some, and it's closer to what i'm doing, so your suggestions are most welcome.
Grizzleton Feb 1st 2010 2:57PM
I'd also like to mention Rhadatip here... Using your own Simcraft scale factors Rhadatip will do a dps comparison for you on the fly of any piece of gear, taking into account the existing gear you're wearing, your enchants and your favorite gem choices.
Grizzleton Feb 1st 2010 2:58PM
SimulationCraft, sorry. I too WTB edit button.
richardjarrell Feb 5th 2010 12:26PM
I have been a huge fan of SimCraft for a while. Things got so much easier when you could just import a character as well. Used to have to input all your stats and priority list yourself.
Racci Feb 1st 2010 4:15PM
I have been looking to get into SimCraft, this was such a great article to get started. It's pretty overwhelming trying to do on your own, thank you for the well written guide.
Brouck Feb 1st 2010 7:07PM
I have been using RAWR for my optimization needs to relatively good results. Should I think about switching to simulationcraft?
Dominic Hobbs Feb 1st 2010 7:27PM
I haven't been following the rawr dev team progress with regards locks for a while now though I recall they were struggling a while back, that may no longer be an issue though.
The big selling point in SimulationCraft is that it simulates the fight rather than using flat calculations. By doing so it can better model proc events and the like. Also, the whole project is very open to peer reviewing of code, theory and fomulas. This means that it is very quick to adapt to new mechanics within the game and to get them right. That's not to say Rawr or any other calculator doesn't but that the mechanisms in SimulationCraft are likely to be stronger.
From a personal point of view I have never gotten on too well with Rawr, not because it did what it does badly, simply because the means of achieving the results were uncomfortable for me. I got on better with the old spreadsheet and now with SC. Though having it open to review is a big seller for me.
SC used to be pretty inaccessible but huge strides have been made in that direction recently (mainly with a GUI interface). The core of solid maths is still there but the knobs and dials are much easier to use than they were and there are more of them. I would suggest you link this post to all your buddies and get everyone playing with the app and learn from each other. This is simply an intro to the tool with a single goal in mind. There's so much more you can do with it than I have shown.
TL:DR version:
I would say yes, but if Rawr works for you then maybe try both side-by-side for a while and see if they match, if they do then you choose. SC has an import function from Rawr so this should be pretty easy.
Jen Feb 1st 2010 8:18PM
RAWR is still struggling with Locks. My Tree hubby lives by RAWR and finally convinced me to try it out a couple of weeks ago and it still doesn't even take into account pet damage.
Aaron Feb 1st 2010 8:11PM
It's important to understand that SC isn't meant to tell you the exact dps you are going to do.
It's meant to show how a different rotation/gear set can affect your potential dps. Often times it'll model a perfect scenario (although this can be augmented with the player skill bar) and can be vastly superior to one's own skill.
However, it is still constantly being worked and reworked when new information becomes available/is discovered, which is common for most theorycrafting tools. One thing you can't really do is compare a simulation to something like a spreadsheet because most spreadsheets average the damage of each ability; therefore, making the dmg per execution static where as SC will have dynamic values. Spreadsheets also average talent/gear procs toward your dmg; therefore, you'll get discrepancies between SC and spreadsheets/web apps that are modeled the same way as a spreadsheet (I'm looking at you femaledwarf.com).
In the end, SC is a great tool to help you decide on upgrades as well as spell priority/rotation; however, it shouldn't be used to compare absolute values.
Zomgdotz Mar 23rd 2010 4:43PM
Just to add to your comment, there are several assumptions that SC will make unless you tell it otherwise...
Like you have appropriate buffs for a 25-man raid. These can massively change the dps you see. Further it assumes you have a high skill level (read as your always time things properly to maximize the benefit: e.g. cooldowns with bloodlust, cooldowns as soon as they're available again, never clipping your immolate, keep up lifetap buff, etc..etc..etc.). Truthfully, even the most attentive person can screw these things up occasionally. I recently installed powerauras just to make my lack of a lifetap buff more apparent to me. It also assumes, unless you tell it otherwise, that you're fighting a patchwerk style fight -- e.g. no movement and no debuff. Ever ran with a tank who moved the boss too much (or erratically)? Your dps drops a lot. Similarly, if you're able to stand still and just blow something up, it shoots up. If you want to see something a little different, change the fight type to Helter skelter.
Please understand, this is not at all a criticism of SC. I *heart* SC.
It is a great tool to compare the relative value of 1 item vs. another.
One other tip. If you're a true max/min person wanting to squeak out every last bit of DPS, take some time to understand how monte carlo (I assume it is monte carlo based?) or other simulation based software works. For example, simulate the exact same character with the exact same gear/spec etc 3 times and you'll get 3 slightly different results. The reason why is that the software is simulating the fight a few hundred times....and then giving you the average. This is the power of simulation based software. So why do you need to know this? Be wary up grades that give you < ~ 20 dps changes. I would suggest increasing the number of simulations in the options to max to give yourself a little tighter average and then take a look again. Sometimes the upgrade will only really be a sidegrade.
Jen Feb 1st 2010 8:16PM
OMG...I think I love you!
Seriously though, thank you for breaking this down into terms the general populace can understand. I have spent hours trying to figure out the EJ site and the 20 mins it took me to read this finally made things click. And now I don't have to rely on them, I can test things out for myself.
Lhock Feb 1st 2010 8:23PM
Plain and simple, if you had to login to the armory website, you aren't on Blizzard's armory, and you just got phished/scammed.
Brouck Feb 1st 2010 8:43PM
Dominic,
You convinced me, I will keep RAWR and use it for my other toons, but I'm going to give SC a try and see what results I can get out of it. Thanks for being such a great contributer to WoW.com and the Lock community.
Cheers
Brouck
zenocide Feb 1st 2010 10:42PM
If you download SimulationCraft from the main site it will never ask you for your username or password.
However we have heard recently that some other sites have had keyloggers installed which is rather annoying.
As mentioned earlier, the only official site is at: http://code.google.com/p/simulationcraft/
It's not a large download and Google isn't going to be running out of bandwidth anytime soon so there's no reason to use any mirrors.
We are considering adding MD5 or SHA1 hash sums etc. to the downloads but so long as people download from the official site they shouldn't be needed anyway.
The only "personal" information it might ever ask for is your character's name (and server) so it can download your profile from the wowarmory. Or wowhead profile id etc. None of which can be used in anyway to compromise your account. Again, we will NEVER ask for your username or password.
Furthermore, while it's certainly not everyone's cup of tea, all of the source is available on the google site and can be downloaded either manually or with SVN and you can choose to audit the code yourselves (if you find any bugs we'd love to hear feedback!!!) and compile from the source.
As for the web browser in simcqt.exe it's using the QT framework (which other programmers will have heard of) rather than being written by scratch by us.
For everyone that does give SimulationCraft a try we'd love to hear from you if you find any Issues or Bugs etc. There's an issues page at: http://code.google.com/p/simulationcraft/issues/list
Also if there are any good documenters out there who can understand C++ code, I know that I for one would welcome a bit more documentation for SimulationCraft (but I'm too lazy to write my own ;P)