Dedicated server benchmark · 20 August 2026
Net Butter Sync
Replicated state instead of per-tick RPCs. Measured on a two-core dedicated server at roughly 120 NPCs, against the per-frame RPC approach it replaces.
- Cost per NPC
- −38%0.0618 → 0.0385 ms
- Frame time p99
- −36%12.54 → 7.98 ms
- Worst frame
- −69%33.80 → 10.38 ms
- Frames missing 30 Hz
- −70%2.3% → 0.7%
Server cost per NPC
Game-thread frame time
All measurements
| Measure | Per-frame RPC sync | Net Butter Sync | Change |
|---|---|---|---|
| Cost per NPC | 0.0618 ms | 0.0385 ms | −38% |
| Game thread, mean | 7.86 ms | 5.32 ms | −32% |
| Game thread, p99 | 12.54 ms | 7.98 ms | −36% |
| Worst frame | 33.80 ms | 10.38 ms | −69% |
| Frames missing 30 Hz | 2.3% | 0.7% | −70% |
| Stalls > 15 ms per 3,900 | 23 | 0 | eliminated |
Test conditions. Unreal Engine 5.8 dedicated server, Linux
container limited to --cpus=2 --memory=8g, 30 Hz server tick.
Roughly 120 server-authoritative NPCs, zero players connected.
93 seconds of steady state after a 40 second warm-up, three runs per
configuration.
Noise floor. Repeating one configuration three times gives ±4% on the mean and ±3% on p99. Every difference shown here is well outside that.
Not measured here. Bandwidth. With no clients connected nothing is sent, so the per-client byte cost — where replicated state gains most over multicast RPCs — is not represented in these figures.