In search of the cut-off point for what duration left on Sunfire that makes it not worth refreshing, I have discovered that removing the refresh at peak condition in the simcrafting script sims into a significant DPS increase up towards one percent, depending on fight type and length. The precise change:
Code:
Default:
actions.single_target+=/sunfire,if=remains<7|(buff.solar_peak.up)
Better deeps:
actions.single_target+=/sunfire,if=remains<8.5
8.5 seconds on Sunfire seems to be the top point, although values as far away as 7 and 10 are not much lower and still represent an increase over the default command. All the simming is done with my current 686 gear, that has well balanced stats, crit/haste trinkets and tier bonuses. (I Might switch to Haste later today for H&F.)
I've tried with different threshholds for going into or out of Solar eclipse, and various ways of conditioning the refresh on the presence of the peak buff, but I haven't been able to script something that beats the simple 8.5 rule.
In the Simcraft log, there's a miniscule drop in Sunfire uptime and several fewer Sunfire casts. While trying to play according to that rule on a targed dummy I frequently find myself refreshing Sunfire at start of Solar.
Is Simulationcraft a valid way to test rotations? Is this finding an artifact? Or is really minimizing the number of GCDs spent on refreshing DoTs worth losing a few seconds uptime on, and the peak bonus actually not worth going for?