AWS SimSpace Weaver
A managed service that executes large-scale spatial simulations with millions of entities distributed across multiple servers
Overview
AWS SimSpace Weaver is a managed compute service that enables running large-scale spatial simulations with millions of simultaneous entities by distributing the computation across multiple servers. It handles the complex infrastructure of spatial partitioning, entity migration between partitions, and clock synchronization, allowing simulation developers to focus on entity behavior logic rather than distributed systems engineering.
Spatial Partitioning and Distributed Execution Model
SimSpace Weaver divides the simulation world into spatial partitions, each running on a separate compute worker. As entities move through the simulated space, they are automatically migrated between partitions. The service manages partition boundaries, ensuring entities near boundaries can interact with entities in adjacent partitions through a subscription mechanism. Clock synchronization across all workers maintains simulation consistency, with configurable tick rates determining the simulation's temporal resolution. The spatial partitioning strategy can be customized based on entity density patterns - uniform grids for evenly distributed entities or adaptive partitioning for scenarios with clustering. Each worker runs a simulation app that processes entities within its partition, communicating state changes through the service's messaging layer.
Game Engine Integration and Visualization
SimSpace Weaver integrates with Unreal Engine and Unity for visualization of running simulations. A separate viewer app connects to the simulation and renders the current state, enabling real-time observation of millions of entities. The viewer receives entity position and state updates through a dedicated streaming channel optimized for rendering. This separation of simulation logic from visualization allows the simulation to run at full speed regardless of rendering complexity. Custom visualization clients can be built using the SimSpace Weaver SDK, enabling domain-specific views such as heat maps, flow diagrams, or statistical dashboards. Recording and playback capabilities allow capturing simulation runs for later analysis without the computational cost of re-running.
Use Cases and Design Considerations
Primary use cases include urban traffic simulation (modeling millions of vehicles and pedestrians for city planning), crowd simulation (event safety planning, evacuation modeling), defense and training scenarios (large-scale battlefield simulation), and telecommunications network modeling. Design considerations include entity interaction radius (determining how far entities can sense and interact), update frequency requirements (high-frequency for physics-heavy simulations, lower for strategic simulations), and state size per entity (affecting memory requirements per partition). Cost is based on the number of compute workers and simulation duration. For development, smaller simulations can run on fewer workers, scaling up for production runs. The service automatically handles worker failures by redistributing partitions, providing resilience for long-running simulations.