Skip to content
Performance with context

A number is stronger with its evidence.

Understand Sigil’s committed-throughput measurements, the workload behind them, and the difference between a benchmark and current network health.

BenchmarksIllustrated study
SHARED RULES. SEPARATE ACTORS.
Shared rules. Separate actors.

Illustrated coordination. No live network activity.

Benchmarks / An illustrated overview
01 / Built for a purpose

Measure the work that reached commitment.

Ingress counters can make a system look fast before it has completed the work. Sigil’s Velocity measurement model counts signed user transactions that execute into lane receipts and roots, with the environment and observation window kept alongside the result.

01

A clear numerator

Committed signed transactions are distinct from requests received, synthetic counter increments, rejected transactions, or dropped work.

02

Topology belongs beside throughput

The worker count, lane count, slot configuration, and commitment hierarchy describe the resources that produced a result.

03

Latency has a definition

Slot latency, block interval, and end-to-end finality are different measurements. Each should name its start, finish, percentile, and observation window.

04

Failures stay in the report

Invalid, dropped, and lagging work affect the interpretation. A performance summary should expose those outcomes rather than count only the successful path.

05

Readiness is a separate question

The same historical report distinguished functional runtime performance from incomplete public-launch and independent-operator evidence. Throughput alone does not establish either.

The details matter

Good questions.
Clear answers.

Is the historical Velocity result the speed of every Sigil transaction?

No. It describes a particular workload and deployment. Contract execution, contended state, different hardware, and different network conditions need their own measurements.

Where do I look for current network health?

Use the network status surface for current observations. Treat stale or unavailable telemetry as unknown rather than replacing it with a historical benchmark.

Keep exploring

One capability.
A connected platform.

01Velocity architecture02Current network status03Consensus model