One store, ten reads

Ten questions. One store. One copy on disk.

last stored run · last stored run

Ten separate database products would each keep their own copy of this graph and answer their one question over it. Below, the same node-set is answered ten ways over a single Semurg store, each number from the last stored benchmark run on this box on a public dataset you can download and re-run. Every card carries the exact command to reproduce it on your own hardware.

anchor: loading the resident graph
shipped a first-class query surface scan pattern the shared scan or fold, framed as that paradigm not yet load a set on your own node to light it
Reproduce

Do not take our word. Run the same call.

The number on every card above is the number this command returns, from the same in-process measurement. Copy it, run it, read the JSON.

curl -s 'https://one.semurg.io/api/collapse?dataset=orkut'
1
curl -s 'https://one.semurg.io/api/collapse?dataset=orkut'The whole wall as JSON: anchor, ten lenses, the fused finale. Each card links its own per-lens filter.
2
curl -s 'https://one.semurg.io/api/bench/graph_teps?dataset=orkut&hops=4'The graph lens also runs standalone as a whole-graph BFS. saops, reads, and scans have their own /api/bench keys too.
3
sudo ./semurg-install.shInstall on fresh Ubuntu, pull the same public SNAP graph, and point the identical curl at your own localhost:4000. Same units, your hardware, your number.

Every value renders only when the engine returns a positive measured number, with its tier band (warm resident, cold device) and its maturity tag (shipped surface, scan pattern, or not yet). The vector lens reads not yet because no vector set is loaded on this box, so load one on your own node to light it.