@pies1 Hey, I'm from Ask Mr. Robot. There are 2 reasons it takes longer to run on AMR.
1. Accuracy: we're looking at 1000s of data points for each stat (instead of just 2 like SimC does, which is one of the reasons those weights can be so volatile).
2. Trinkets / Items with procs: we simulate every item with a proc when you run a 'gearing strategy.' Since procs don't rank well (or at all, really) with stat weights, this is an important piece of information needed to rank gear. So that setup is running every item with a proc so that they all get ranked directly from simulations
You can increase speed if you increase (or is it decrease?) the margin of error to be a bit larger. We also have a few ideas to tap into more computing power for future features. Basically, the gearing strat we do is super thorough, so it takes a bit longer. But totally worth it so you get the best gear rankings