Distributed Embedded Systems Using Zephyr
summary
Zephyr integrates Pigweed RPC and .proto APIs so services compile as local vs remote implementations; example builds a two‑service cache with a custom ring buffer channel output and an HDLC path. Benchmarks run 1,000 set/get roundtrips: a control loop costs ~33 µs per call, the PW RPC pipeline about 116 µs per call; profiling pins serialization and destructor overheads and suggests optimizations that could cut RPC latency toward 14–45 microseconds per call. Tests and parallel development practices prove feasible.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:32Portable microservices in Zephyr
- 3:20Pigweed RPC modes
- 6:09Pigweed in Zephyr projects
- 7:55Headers to Protobuf services
- 9:18Header-based API limits
- 11:03Cache service example
- 12:27Channel output abstractions
- 13:33Performance test setup
- 15:19RPC performance bottlenecks
- 17:05Q&A
related talks
34mZephyr and OpenAMP
38mRemote Processor Framework in Zephyr RTOS
31mRPC Frameworks Overview
55mForget about HTTP
43mZbus 2030: Community-Driven Directions for Zbus
11m"RPC Frameworks Overview"
18mHypervisor-less Virtio: Assembling Multi-OS Systems Using Standards-based Pro...
24mGluing Clojure Microservices