Platform
Hardware Acceleration
Multi-queue NICs, processor affinity, and cache-friendly batching combine to cut per-packet cost by an order of magnitude.
GA
HOW IT WORKS
Walk through a single flow
- 1Worker threads are pinned to specific processor cores for consistent cache residency.
- 2Batched processing amortises per-function overhead across many packets.
- 3Multi-queue steering distributes flows across workers without lock contention.
RELATED