Despite the multi-threaded nature of tyFlow, a complex flow can still take a long time to simulate. To help troubleshoot why a simulation is taking so long, you can turn on tyFlow’s simulation profiler by enabling “Print simulation details” within a tyFlow object’s “Debugging” rollout. The profiler will print the total execution time for each step of the simulation to the MAXScript listener, allowing you to quickly identify simulation bottlenecks. More information about using the profiler can be found here.
This question is scene-specific, but some general guidelines can be used to optimize every flow:
Keep simulation time steps as low as possible. Only increase time steps when more simulation accuracy is required.
Keep input geometry as low-res as possible. Even though tyFlow can handle hi-res input geometry with relative ease, a lot of time may still be wasted parsing unnecessarily hi-res meshes.
Setup your flows with lower particle counts, only increasing the number of particles when you’re ready to export/render.
Use Birth Flow operators to chain multiple flows together, allowing for some flows to generate the initial state of the simulation (requiring only one computation of that state), while other flows (that are continually updated) work from that initial state.
Use PRT partitions to export huge numbers of particles into multiple passes simultaneously, across multiple machines, rather than trying to export all particles at once.