Active development — graph comparisons use a synthetic 10K-triple dataset (or Schema.org profile); the platform suite covers Sentinel, streaming, and SHACL from qualia-cli bench. Scale and concurrency may differ in production. Run the open-source harness locally to verify.
COMPARATIVE HARNESS v2 • 512 MB ON-DEVICE EDGE

Comparative
Benchmarks

Graph engine shoot-out plus platform suite metrics.
Graph · Sentinel · streaming · ontology.

engines
n = 10,000 triples
synthetic-10k
Engine
Loading results…
Live benchmark hub Browser harness
Performance under 512 MB constraint
Engine Dataset file
MB on disk
Ingestion
ms · triples/s
Point
p50 · p95
Two-hop
p50 · p95
Filter
p50 · p95
Throughput
point ops/s
RSS
MB
Ingestion Time
Lower is better • ms to load 10k triples
Point Query Latency (p50)
Lower is better • indexed lookup
Filter Scan Latency (p50)
Lower is better • predicate scan across all triples
Peak Memory (RSS)
Lower is better • server/process RSS under load
Reproduce these results

Run the harness locally. Results are committed for full reproducibility and auditability.

Click to copy the selected profile command