High Bandwidth Sensors
summary
A stress‑test application based on TDK Robokit1 shows how Zephyr reaches >10kHz multi‑sensor sampling by moving from blocking reads to the rtio mempool + async sensor APIs and upcoming streaming sensor APIs. The talk names concrete bottlenecks (driver-owned buffers, thread contention), tools used (RTIO, mempool, decoder contract, Q31 + shift, ZDSP helpers), and design choices that reduce blocking, enable delayed or off‑core processing, and preserve sample alignment; reported results show clear performance and correctness improvements.
AI-written summary · about
chapters (11)
- 0:00High bandwidth sensors intro
- 0:49Defining high bandwidth sensors
- 2:58Blocking to async APIs
- 5:07Mempool benefits
- 8:16Processing one-shot data
- 10:21Decoder flexibility
- 13:10Frame and channel iterators
- 14:33Q31 data and DSP library
- 17:01Streaming sensor APIs
- 19:06Triggers in streaming mode
- 20:12Sensor API improvements
related talks
41mZephyr RTOS: Under 1KB of RAM- 41mZephyr RTOS: Under 1KB of RAM
43mFrom Code To Current: Reducing Energy Consumption in Zephyr
29mModernizing Resource Management in Embedded Systems Using eBPF
25mRecent developments in I/O control and testing
42mExtending Container Performance Isolation: Regulating Memory Bandwidth & Cache in the Kernel
40mApplication and Challenges of Using GenAVB/TSN Software Stack for Industrial...
51mDemystifying Memory: A Practical Tutorial on Managing & Optimizing Memory in Zephyr