mirror of
https://github.com/CloudNebulaProject/wayray.git
synced 2026-04-10 21:20:40 +00:00
Ensure today's CPU-first design doesn't block a future zero-copy GPU encoding path (DMA-BUF → VAAPI/NVENC, sub-5ms encode latency). Key design constraints: - FrameSource enum: Cpu and DmaBuf variants (build Cpu now, DmaBuf later) - FrameEncoder trait: accepts both, encoder picks preferred source - Pipelined render→encode→send (no synchronous blocking) - Adaptive frame cadence: OnDamage (desktop) vs Fixed (gaming) - Damage tracking forks: pixel diff for CPU, Wayland hints for GPU No GPU code built now -- just trait interfaces that preserve the path. |
||
|---|---|---|
| .. | ||
| adr | ||
| plans | ||