Given a set of item candidates for each slot, and normal talents/buffs input, the sim should compare the performance of each combination of items.
This is similar to EP calculation, in which randomness should be controlled for each iteration for more consistent comparisons. The output should be a ranking of item sets, with avg/median DPS for each.
The combinations do get out of hand rather quickly, and it will likely be necessary to distribute this workload across some relatively capable machines. EC2 c5a machines are probably the best price per vCPU/hr.
This will not be remotely feasible to run via the web version, but the output should be consumable from the UI.
Given a set of item candidates for each slot, and normal talents/buffs input, the sim should compare the performance of each combination of items.
This is similar to EP calculation, in which randomness should be controlled for each iteration for more consistent comparisons. The output should be a ranking of item sets, with avg/median DPS for each.
The combinations do get out of hand rather quickly, and it will likely be necessary to distribute this workload across some relatively capable machines. EC2 c5a machines are probably the best price per vCPU/hr.
This will not be remotely feasible to run via the web version, but the output should be consumable from the UI.