Split Rendering TV Applications Using WASM
summary
BBC R&D builds a prototype that runs the same Rust wasm32-unknown no_std module in-browser and on server-side Wasmtime, using WebGL/TypeScript in TVs and SDL/OpenGL + GStreamer on Linux with an Nvidia A40 in the cloud. Input travels over WebRTC libdatachannel, frames use a zero-copy GPU→encoder pipeline, and end-to-end tests show a 230 ms max latency with ~50–55 ms attributed to frame capture/encode/network; prototype meets smooth 30–60fps UX.
AI-written summary · about
chapters (12)
- 0:00BBC R&D and split rendering
- 3:11TV device ecosystem challenges
- 7:41Why split rendering with WASM
- 10:11Prototype architecture for universal access
- 13:21Performance and latency analysis
- 16:10WebAssembly standards and API needs
- 19:22Personalized TV experiences and limits
- 21:31Q&A on server and latency
- 24:19Accessibility in WASM TV apps
- 27:06Non-WASM devices and browser limits
- 29:14Browser-bound WebAssembly and runtime alternatives
- 30:38Closing remarks
related talks
27mWebAssembly for Performant Interactivity
47mRewrite the web with Rust and Web Assembly
24mVideo Editing in the Browser
02h 58mWebassembly: Expectation vs. Reality
12mExploring Video Editing in the Browser
25mLessons in WebAssembly: Client Side Video Editing
41mLet's build with WebAssembly!
37mBeyond the browser: Moving forward with a more Declarative Future with WebAssembly
