SimulationCraft And Stat Weights

[Lyones Profile]

Hi, long time reader, first post though.

I decided to download simulationcraft program and I can`t make it work. When i try to mesure my Weapon DPS weight, it always appear something unreal, negative values or something really low. i would like to fix it, and i though to ask some help here =D

Also, if you can simulate for me:
http://us.battle.net/wow/en/character/n ... nes/simple

Thanks ;D

[aggixx Profile]

There's a bug in the most recent precompiled version of SimC, it's since been fixed but there hasn't been a release since then. I'll sim it for you in a few mins.

Edit: Here you go:
http://dl.dropbox.com/u/1086121/simc_pr ... 41055.html

[Lyones Profile]

Thanks Aggixx, Really Kind of your part.

[Helistar Profile]

Aggix was faster

BTW I see you with the Wrap of Unity on your back.....

Not that I disagree with the choice, it was on my back as well for a couple of first HM downs of our guild

[aggixx Profile]

You had the right cloak on in the simulation though

[sporadic Profile]

Doing my own sims, I've noticed the same bug as Lyone, however I also have experienced something very odd. My scale factor results are bizarre:
Agi 3.91
Exp .74
Hit .91
Crit 1.28
Haste 1.24
Mastery 1.17
Wdps -.02

As you can see, I'm having my secondaries valued Crit > Haste > Mastery, which is quite confusing, as the sims you present show Haste being valued even below Hit and Expertise. In addition, I would expect Hit and Expertise to be valued equally, but they're clearly not.

I'm running SimCraft 510-9, which has given me accurate results in the past, so I'm left quite stumped. Do you have any ideas why I might be getting these results?

My armory is http://us.battle.net/wow/en/character/s ... c/advanced

[aggixx Profile]

Hit Expertise not being equally valued would be a result of having hit and expertise over the cap and/or not running enough iterations.

There was a major bug in 510-9 where all the results generated by anything but your first CPU thread were discarded, so that could cause you to get very inaccurate results if you're running a lot of threads and/or not enough interations. If you want to make sure your results are accurate I would upgrade to 510-10 (which was just released today) and make sure you're running at least 10k iterations, 25k is a lot better if you can though. 510-10 also includes the fix to the issue with WDps scale factors.

[sporadic Profile]

That might be the problem. I did run at 50k iterations, but I did it with 8 threads. If I run with 510-10, can I use multiple threads, or is that bug still present?

Edit:
I ran 50k iterations again, this time with 510-10, and got better results. My factors now properly show Mastery > Crit > Haste, however I still have the odd Hit and Expertise values (1.17 and .74 respectively).

Thanks for pointing out the bug, I doubt I would have figured that out quickly at all.

[aggixx Profile]

That bug is fixed in 510-10.