Parameters for GPU-local population dynamics: destruction and spawn-as-copy over a seed's fixed, over-allocated capacity. More...
#include <PopulationProcessor.hpp>
Collaboration diagram for MayaFlux::Buffers::PopulationConfig:Public Member Functions | |
| void | declare_fields (const std::shared_ptr< NetworkGeometryBuffer > &buffer) const |
| Declare the two state fields population dynamics needs. | |
Public Attributes | |
| SpatialHashConfig | hash |
| uint32_t | live_count |
Detailed Description
Parameters for GPU-local population dynamics: destruction and spawn-as-copy over a seed's fixed, over-allocated capacity.
hash.particle_count here is expected to already be live_count plus whatever reserve capacity SpatialFieldConfig::reserve_fraction asked for: the caller overrides SpatialHashConfig::particle_count to that total before constructing this, so every hash/claim stage already dispatches over the full reserve range. live_count is the boundary this struct itself adds: everything at or past it starts dead, and is the only region PopulationSpawnProcessor is allowed to write a new particle into.
This mechanism is not reflected to the CPU. Whatever PhysicsOperator simulates is the authoritative population; what exists here is GPU-local bookkeeping about which of the seed's records currently render as alive, rebuilt from scratch on the next reseed. ClaimAccumulateProcessor is the one enforcement point: it never lets anything past live_count reach PhysicsOperator's own bond/mass tracking, however far spawn has grown the live-looking population beyond it.
Definition at line 32 of file PopulationProcessor.hpp.
The documentation for this struct was generated from the following files:
