Appearance
Insurance Monte Carlo Benchmark
This benchmark measures distributed execution of an insurance loss simulation model.
Workload
Monte Carlo insurance loss simulation.
Parameters include:
- claim frequency
- severity distribution
- catastrophe probability
- policy limits
Simulation size:
100,000,000 iterations
Network Topology
Agents: 20
Shard size: 5,000,000 iterations
Measured Result
Iterations: 100,000,000
Wall time:
≈ 2.8 seconds
Derived Metrics
Aggregate throughput:
≈ 35.7 million iterations/sec
Per-agent throughput:
≈ 1.79 million iterations/sec
Interpretation
The insurance model performs significantly more work per iteration than the ForgeCAT workload.
Consequently throughput is lower but remains highly scalable across the network.
Scaling Projection
Assuming similar per-agent performance:
| Agents | Throughput | Time for 1B |
|---|---|---|
| 20 | ~35.7M/s | ~28 s |
| 200 | ~357M/s | ~2.8 s |
| 560 | ~1B/s | ~1 s |
Boundary
Results represent a simplified insurance simulation and should not be interpreted as representative of all actuarial workloads.