In the old implementation, packet processing was limited to a fixed number per frame (max 8). If more packets arrived than could be processed in a single frame, they would accumulate in the buffer, increasing latency over time.

In the new implementation, the limit is no longer based on packet count, but on a time budget. Each frame processes packets for up to 15 ms.

This means that:

At 30 FPS (≈33 ms per frame), roughly half of the frame time (15 ms) can be spent processing packets.
During this time, the system drains the buffer as much as possible instead of being artificially capped at 8 packets.
As a result, packets are processed faster and are less likely to accumulate, reducing latency and network spikes.

In short:
we moved from a fixed packet-per-frame limit to a time-based processing budget, allowing more efficient and consistent packet handling.

Packet Improvement

Description

Download Resource

Details

How to get a token?