Benchmark Methodology
Throughput and latency depend on the benchmark setup. We define the node count, validator distribution, network latency, transaction size, account contention, and consensus parameters before measuring.
Service
Performance Assurance
Don't trust the TPS number. We reproduce performance claims under controlled conditions: defined hardware, workload, transaction size, topology, latency, and concurrency.
Performance vs. Correctness
A protocol that sacrifices finality, censorship resistance, or safety to hit a TPS target is not faster — it is broken. We measure correctness under load, not just peak throughput.
Resource Modelling
CPU, memory, bandwidth, disk, and energy use are tracked during tests. We identify bottlenecks, resource exhaustion points, and the cost to operate at claimed scale.
Stress & Saturation
We push the protocol beyond its claimed limits to find where it degrades, stalls, or fails. Saturation behaviour is often more revealing than the headline number.