Semi-static assignment makes use of the near term locality of slowly moving elements but also accounts for the elements eventually moving.
tarabyte
How do programmers figure out how often to recompute the assignment? Is it by experimentation, or are there heuristics for it?
yeq
The reason why a static assignment is not efficient is that as time goes go, not only do the particles move, their statistical distribution also changes dramatically. The dynamic statement is also not feasible because there is some spatial coherence exists for the space evolution so that it's not necessary to recompute the workload distribution from scratch in every frame.
srb
What are common ways to implement a semi-static assignment? Something like redistribution of work every x number of iterations?
Semi-static assignment makes use of the near term locality of slowly moving elements but also accounts for the elements eventually moving.
How do programmers figure out how often to recompute the assignment? Is it by experimentation, or are there heuristics for it?
The reason why a static assignment is not efficient is that as time goes go, not only do the particles move, their statistical distribution also changes dramatically. The dynamic statement is also not feasible because there is some spatial coherence exists for the space evolution so that it's not necessary to recompute the workload distribution from scratch in every frame.
What are common ways to implement a semi-static assignment? Something like redistribution of work every x number of iterations?