You can specify the simulation duration from the "Fight" page. I kept the default value low (30 minutes) because I've tested with some lower end machines and it can take an inordinate amount of time to run a sim with a high value. It also allows you to quickly go back and forth and get a quick idea what a gear/talent change will do. Then after you have narrowed down your choices, you can re-sim for a longer period to validate your quick sims.

Basically what it comes down to is any good software developer will keep the default experience easy to manage for people new to the application. Having a really long sim time works against the responsiveness of the application and may confuse new users. Once you know what to expect, you can adjust the sim duration up.

EDIT: I just realized you were responding to someone else.

No, I don't assume most people are tech savy. Most of the people on this site are, but I'm aiming for a broader audience. Most people that download spreadsheets or something similar will take one look at it, ask "What now?," delete it, and never make use of it. Secondly, even if I wanted to put everything on one page, there simply wouldn't be enough room--there's just too much information.

I used the character in my profile, along with trying a few other ones. No not behind any type of a firewall.

Isn't 30 mins a really short time for a Sim, that can be heavily effected by a RNG.

You're transposing Realm name and Character name. So, that would be "Bonechewer" in the "Realm:" box and "Ermad" in the "Name:" box.

I get pretty stable numbers with 30 minutes. Again, it's in the interest in keeping the sim from taking too long--especially on slower machines. If want more accuracy, you're certainly free to bump the duration up.

Regarding the duration issue, one thing I have considered doing is sacrificing timing accuracy to allow longer duration simulations to smooth the RNG. The simulation works by executing "rounds". There are currently 1000 rounds per second, which gives you a timing accuracy of within 1/1000 of a second--that's probably overkill and not humanly possible. When I built the first prototype, 1000 RPS was fine--you could still sim for a long duration without taking that much real time. As the sim has become more complex, the real time required has sky-rocketed. So what I could do is reduce the RPS to, say 100 (which would give you 10ms accuracy) and allow you to sim for 10x as long for the same amount of real time.

Regarding the SIM. Once you get the bugs and add in a lot of the features you are planning to try. Would it be possible for the Sim to generate a combat log? I could see that being useful if you were able to model new changes in PTRs and then run a sim for dps on WoW prophet rather than see people's 5 min combat log on a target dummy in silvermoon which accounts for nothing. All that assuming the sim runs as accurate as possible when compared to a raid.

Yeah, an option to produce a combat log is something that I want to add. It would have to be an option, disabled by default; and probably a warning about using it. A 100 min combat log would probably take quite a while to run even on a fast machine and take quite a bit of memory.

I went ahead and committed the change I mentioned above. It's consistent within a few tenths of a percent now. I'm not sure I want to sacrifice on timing accuracy any further. Another factor of 10 would put it at only being accurate to within 100 ms--a lot of people get better latency than that.

Just doing a few tests with your character, using a simulation duration of 1000 minutes I saw the following numbers:
2887, 2892, 2892

Then, I changed out a few of your +16 str gems for +16 exp and got the following numbers:
2915, 2912, 2913

That's probably about as consistent as it's going to get. It's probably fair to say you would benefit from being exp-capped. WowProphet assumes you're always attacking from behind--so no parries. If you're ever in the situation of attacking from the front, that would further increase the value of exp.

Current throrycrafting does recommend expertise capping and most the BIS gear does push us towards that. However keep in mind Wow Prophet is not taking into account raid buffs. Once those come into play stat priorities will switch some

Very true. I more or less meant it as a demonstration to show how it could pick up on slight gear changes with the improved sim duration.

The application uses signed 32-bit numbers. If you add to a number enough times, it will overflow the memory allocated and "wrap" around and go negative. So, as long as your damage stays under (2^32) / 2, or about 2 billion, you shouldn't run into this.

This looks very Positive. I'm looking forward into seeing the sections come up (particularly Buffs, then rotations)

Regardless of anything else, there are some issues with the basic simulation that would need to be fixed before it's actually useful.

Notably, the hit table is wrong (it's showing me having 0.56% miss at 277 hit rating; i'm assuming you're still using 9% as the melee hit cap, when extensive testing has shown that it's 8% in WotLK), and it's showing exactly the same crit rate for spells and melee, which shouldn't be the case due to melee/spell crit contributions from agility/intellect respectively.

Anyone else having problems with loading the page at all?

Yeah, I seem to be having problems with the host. I'm looking into the problem.

I've seem to have run into a problem where when I import my character from the armory, everything else loads fine - but it gets stuck on 'loading' the stats on any of the weapons.

This leads to the point where the simulator says my DPS is about ~1000 without any white swings taken into account.

Items are downloaded from Wowhead; check your connectivity to Wowhead.

While more diversity in simulation tools is not a bad thing, isn't what you plan more or less another Simulationcraft, only web-based and with an userfriendly UI?
Are there any other differences (other than silverlight does not support c++) i am not aware of, esp. in the philosophy or inner workings of the applications?

Looks very classy! Always good to see another interesting simulation effort.

A few recommendations regarding the model:

(1) Consider fixed time fights and scaling the number of iterations to get accuracy. Fight length is a very interesting input parameter (due to cooldowns, etc). Related to this is the issue of target health and how it affects things like Execute and Molten Fury. simulationcraft is actually target-health limited, but I use each iteration to refine the target health such that we get the desired fight length.

(2) You've referred to "rotations" a couple of times. If this is your model, consider changing to a priority list of actions with ready() methods. This will greatly help you when you start modeling other classes.

(3) Generating scale factors (EPs, stat-weights, whatever you call it) from a simulation engine is extraordinarily painful. Consider your example above. DPS1={2887,2892,2892} and DPS2={2915,2912,2913} gives a DPS_Delta={28,20,21}. Dividing the DPS_Delta by the stat change would result in scale factor changes that vary 40% between min and max. It can be done, but it is difficult to do in a runtime efficient manner. So that you might skip much of the pain-and-suffering I went through, consider looking at sc_rng.cpp. When "smooth_rng" is specified (which happens during scale factor gen) we instantiate roll-specific not-so-random RNG packages. This improved our convergence by a factor between 10x and 100x depending upon the class.

Best of luck! If you have any questions don't hesitate to ask.

Thanks, I very much appreciate the feedback.

1. One of the things that I'm really shooting for is accessibility--I want WowProphet to be accessible to the average Wow player. I understand why fight length is important but I'm not sure average players would. Or, maybe a better question is "How do I design the simulation setup so that average players understand what they're simulating?" With the current model, it should be pretty straight-forward (hopefully): the longer you sim, the more accurate the results. But that's definitely something that I would like to do--maybe put it in an advanced tab on the simulation setup page or possibly allow people to define scenarios on the opponent setup page.

2. Rotations probably isn't the best word for it (especially paladins) but it's part of the accepted vernacular, so that's just the word I used. The current implementation is a priority list like you mentioned. The highest priority item that is available is executed. One of the things I have envisioned with the rotation builder is that you could add conditionals. eg. Only execute an ability if some buff or proc is active. Currently, the only "conditional" is if the ability is not on cooldown.

3. Yeah, stat weights are hard to do with a simulator--they can't be solved like simple closed-form equations. One of the things that I have considered doing is taking peoples' results and doing a statistical analysis. Yet another benefit of having this web-based is the huge amount of data available. I haven't given this much thought on how I would go about this, but I'm sure it's possible.

Accessibility is important. It is the difference between creating a tool people can use and a pet project you do for fun (cough-simulaitoncraft-cough). That being said, there are some very important modeling aspects to fight length that even an average player will consider important. Extending a fight over a huge period of time means that you effectively end up with a steady-state with regards to resource gain and loss. If you consider initial mana pools in the same bucket as regen, they would represent a significant portion of the mana sources in most fights. Fight length also effects how target-health-dependent mechanics are modeled. You might get an odd result if your DPS Warrior was Executing for 20hrs of a 100hr fight.

I don't mean to press and these aren't issues with any real immediacy...... but I wanted to touch on a few things that cost ME time to go back an re-architect oh-so-long ago. Again, I wish you the best and look forward to monitoring your progress.