Engine benchmark
Client round-trip latency across 6 queries × postgres / indx. Every call is a direct, fresh engine query — no cache.
| query | engine | round-trip | upstream | hits |
|---|---|---|---|---|
| headless | postgres | — | — | — |
| indx | — | — | — | |
| cli | postgres | — | — | — |
| indx | — | — | — | |
| publish | postgres | — | — | — |
| indx | — | — | — | |
| plugin | postgres | — | — | — |
| indx | — | — | — | |
| deploy | postgres | — | — | — |
| indx | — | — | — | |
| barkpark | postgres | — | — | — |
| indx | — | — | — | |
| median | postgres | — | — | |
| median | indx | — | — |
round-trip = client-perceived latency, median of 6 direct calls per cell. upstream = the route handler → API fetch time. The finder shows per-query engine compute + upstream + round-trip live.