Oct 102010

During WOTLK, regular readers will remember that I frequently advocated the use of simulators, such as Toskk’s, Simulationcraft, and Rawr, to help definitively answer gear/rotation questions for your character. Well, the first simulator to support 4.0 has arrived, and it’s a doozy. Welcome to Mew.

Using Mew is very simple; the interface is based off of Toskk’s. Plug in your character’s stats (unbuffed caster form) and proceed through the windows, selecting your set bonuses, any trinket procs, talents, glyphs, consumables, and raid buffs. Once done, click “Calculate.” The program will take your inputs and spit back what it considers to be your estimated DPS, plus loads of stats if that’s your thing.

Now, in and of itself, this is great, but not super-special. What’s really awesome about Mew is how Yawning and Tangedyn have incorporated both Toskk’s formulation engine and a version of Simcraft’s simulation engine in the same package. (Formulation uses models to predict your character’s average DPS; simulation actually simulates an encounter, for a given number of attempts, and calculates actual DPS. Simulation is slower, but probably a better measure. (For more, see Yawning’s Formulation vs. Simulation wiki page.) Furthermore, the simulation engine does not use hard-coded logic, but like Simcraft, allows for custom rotations to be written and loaded in. I’ve dabbled quite a bit, and it’s a very powerful and effective tool. (Yawning has graciously incorporated some of my suggestions into the default simulation script.)

“Okay, enough pictures,” you say. “Where can I get this hotness?”

Three options, in increasing levels of freshness. (All require Java.) First, an online applet version of the program is located here. Second,  the downloadable binary is located here. Finally, developers (with more skills than me) can grab the code and compile it themselves directly off SVN (link here). Please remember that this is still alpha code, and things may change drastically. Many new features are planned, including Armory integration once 4.0 goes live.

I’ll be writing up my initial findings from this tool and posting them tomorrow, and I’ll also write a third post going into more detail on the custom simulation logic (the current default script can be obtained here). I’m also creating a forum thread to discuss Mew: feel free to ask questions or discuss there.

 Posted by at 7:44 am

  11 Responses to “Introducing Mew: The new feral druid simulator for Cataclysm”

  1. Thanks for the excellent writeup.

    Just to let you know, I put out a new build today that fixes some issues. Baring anything else major (either bugs, or “Oh hey, the actual 4.0.1 patch has totally different coefficients than those used for all of beta”) the codebase should be relatively solid now, though I say that every version. Hopefully I can work on improving the UI a bit now.

    As per your suggestion, the downloadable archive has a copy of the default script included as well, to make tweaking things easier (Under “Mew-20101010/Samples”).

  2. Hmmm I did a run with some values pulled out from memory (and so probably wrong).
    I see that it correctly handles DoT refreshes, which cab bring uptimes in the 100% range.
    For the stat scaling, I’m not surprised that agility is better the anything (it already almost was before doubling the AP it provides), but I find strange that mastery is so good, considering that the result places the total bleed damage at around 37%.
    (BTW I didn’t test recently if shred is affected by mastery, if it is then I definitely understand the result….)

  3. Ooh, awesome. Thanks!

    Shred isn’t affected by mastery, but bleed damage is significantly higher than what this shows; I think the coefficients have changed since this was last updated. Also, mastery is just better than haste with how low shred damage is.

  4. I tried with my real stats at lvl 85 and the Toskks model (which usually overestimates DPS) is significantly lower than what I can put out on the dummy. Significantly = 10%, and the calculation uses berserk while I don’t.
    Calculated DPS 4.9k, I can manage 5.4k without the optimal uptimes. The scaling coefficients are also weird (AP is very very high). Is the lvl 85 model final?

  5. I’d need to see your damage breakdown, what stats you fed into mew (the save files are plain text so you can put it on pastebin or whatever), and if your character had any procs effects equipped that aren’t modeled.

    In addition to that:
    How long did you DPS the dummy for?
    How many times did you DPS the dummy?
    Were you making special effort to apply bleeds with TF? (The Toskk model does not.)

  6. Yawning, the master coder and customer support pro. :)

    I’ll try to get you some logs this week to verify against Mew, but no promises (I’m taking my CCNA on Friday…I’ve procrastinated enough on the final cram for that already).

  7. @Alaron: Thanks, I appreciate it.

    In the famous last words department: “Oh hey, the actual 4.0.1 patch has totally different coefficients than those used for all of beta”

    I guess the patch hasn’t shipped yet, but the AP coefficients on cata.wowhead.com and mmo-champion.com are incorrect. There’s a shiny new binary build available that corrects this, and for those that use it from a web browser, updating the copy there is next on my list. At level 85, this pushes self buffed DPS up by ~500. If you were testing with full raid buffs the changes aren’t going to be all that noticeable since I also adjusted the value of Horn of Winter.

    Sorry for the inconvenience, and I appreciate the feedback.

    • Ick. Thanks for logging the changes. If I’m seeing them right, they’ve buffed the scaling portion of Rip and Rake (both Direct and DoT) by 50%, and FB by 10%. What change to HoW did you have to make? Did they roll it back from the 155 Str/Agi (I think it was)?

  8. Yep, nailed the changes I found. Was somewhat annoying to work out, and I have a spreadsheet filled with damage numbers now. :/

    The HoW change was “I was using really old values from early in the beta that were hilariously high”.

  9. […] Blog post and discussion thread on fluiddruid.net (full disclosure: my site) […]

  10. Ok, I tried again and I must have done something wrong the 1st time. This time I was also using the downloaded version, the applet has some weird interface behaviour (like tab not working at all). Now I get 5.7k, which compares a lot better with my 5.4k-without-berserk.

 Leave a Reply