Server performance: a lighter voyage
Matched tests show about 41% less outgoing world-update data and less server routing work on long journeys. Average total server processing remains essentially unchanged.
Live · 0.13.7 · 2026-09-23T01:15:15.318Z
- In a matched 50-player long-route test, p95 server routing work fell from 4.15 ms to 3.09 ms, a 25.5% reduction.
- Outgoing world-update data was about 41% smaller. Average total server processing time stayed essentially unchanged in both matched tests.
- Sailing routes are planned by your client and checked by the server before your ship follows them.
- Compact movement and cannonball updates let your client reconstruct the visual journey between confirmed positions.
- Inventory rearrangement shows your pending layout immediately, then confirms or corrects it using the server response.
- Discord linking delivers its title reward congratulations and Claim all instructions through your in-game Mail, without a chat notification.
- The server still controls sailing speed, collisions, combat outcomes and item ownership.
Game 0.13.7 is live. The figures below compare the previous 0.13.6 release with this update under the same controlled workload. Restart your client or reload the browser game to load the new version.
Route checks reject illegal steps and diagonal shortcuts through land. Stale positions reconcile from the server’s confirmed location. Tutorial and arena restrictions remain enforced.
Compact updates are negotiated with compatible clients. Server checks and confirmed updates remain the authority for what happens in the world.
The existing account-link reward letter is private to the linked player. Its once-only reward and claim checks are preserved.
The harbour artwork illustrates sailing. It is not a screenshot of a benchmark, and this note makes no claim about internet ping or frame rate.
Measured results: routing and outgoing data
Average total server processing was essentially unchanged. The measured improvements are lower server routing work for long journeys and smaller outgoing world updates.
| Metric | 0.13.6 | 0.13.7 | Result |
|---|---|---|---|
| Total server processing, short routes · mean | 32.263 ms | 32.336 ms | Essentially unchanged |
| Total server processing, long routes · mean | 31.066 ms | 31.051 ms | Essentially unchanged |
| Server routing work, long routes · p95 | 4.146 ms | 3.090 ms | 25.5% lower |
| Outgoing world data, short routes · per player | 20.56 kB/s | 12.08 kB/s | 41.2% smaller |
| Outgoing world data, long routes · per player | 20.83 kB/s | 12.31 kB/s | 40.9% smaller |
Two matched local tests, each with 50 players, 152 NPCs, 200 warm-up updates and 1,000 measured updates. Baseline and candidate alternate in the same process with identical maps, balance and commands. Short and long clicks span 300 and 1,500 world units. Timings are server work per update; p95 is the value at or below which 95% of samples fall. Final player states matched.
Measured on an Intel Core i9-10885H with Node 20.18.0. Client route calculation is excluded from server time. Client proposals use current positions without network delay; live snapshots may be older and require server replanning. These controlled tests exclude database, socket, internet and GPU time and include synthetic visual cannonballs. They do not establish lower internet ping, higher FPS or lower total server processing time.