Even relatively small gear differences can make serious changes in the value of those trinkets, using the 502 version on my paladin and my warrior, there's a gear gap but the real difference is my haste levels on the two toons. I can autoattack a dummy on my paladin and it will get to 2 stacks eventually. My warrior has to have good luck just to maintain 1 stack.
Or compare Grog's toon (on undermine) with Nytro. Nytro has a fairly significant gear advantage, but his haste really pales compared to Grog's. He actually has more than me, and my ilvl is even higher still. I'm not sure if Anaxie is just doing hard math, or if simcraft is properly turning out results for the RPPM trinkets (shaman threads lead me to believe it may be, or at least the enhance module may be), but I'd imagine that simming our three toons would produce somewhat different results in the value of spark vs primordius. It looks like I only have to do a little juggling on my toon to make it possible to do direct comparisons to see what kind of results it gives.
Sims going now, using me, Grog, and Nytro (I think these three toons are really good testing points if simcraft is modeling correctly, since Grog and I have very similar haste levels and 4t15, our primary difference is going to be AP, and Nytro and I have similar gear levels but significantly different haste levels) since we all have spark and none of us have reforged the haste on it away it means I don't have to do any math to maintain caps, and changed my talents to match. I'll change Grog's spark to the same itemid as mine and Nytro's, and use TF Primo, so that all six sim runs are comparing the same ilvl trinkets on each toon. 50k iterations, patch, elite, for consistency.
Now that it occurs to me, I'm actually going to do four sims on Grog, one set of 2/2 TF Spark vs 2/2 TF Primo, and one set of 2/2 RF Spark vs 2/2 RF Primo to see if the dps spread is similar on the differing ilvls.
Grog on both sets were pretty consistent, and on all three toons Primordius was a very minor gain. On a whim I simmed a comparison using Anaxie, and 543 Spark did also sim slightly lower (1k@226k) than 543 Primordius. When they're that close, the one with the more favorable mechanics tends to be the one you want, and Primordius is just more reliable. It always procs high on pull, and always procs high during bloodlust.
Purely conjecture but with the way that it's simming consistently across various gear and haste levels, there could have been a problem with previous versions of either simcraft's ret model or the trinket modeling in the base program and we've just been basing our opinions on that, similar to back in LK when there was a problem with the sim's valuation of haste and it just wasn't caught for a long time.
Edited by Cayse on 6/16/2013 2:20 PM PDT