On the Containerization and Orchestration of RISC-V architectures for Edge-Cloud computing
summary
Claims a working port of the container and orchestration stack to RISC‑V as KubeEdge‑V, built on runC and GCC, and tested on a SiFive Monte Cimone board versus an Nvidia Jetson Xavier. Memory and most app benchmarks match ARM, but system workloads show a ~21% slower context‑switch and an overall ~4% average slowdown (versus ~1% on ARM). Root cause traces to the container runtime layer (not the kernel); deployment guide and scripts are openly available.
AI-written summary · about
chapters (12)
- 0:00RISC-V containerization intro
- 1:11Lack of RISC-V orchestration
- 2:16CubeEdge for Edge orchestration
- 3:22Porting stack to RISC-V
- 4:31Performance vs ARM Edge
- 7:50RISC-V benchmark slowdowns
- 10:39Container overhead causes
- 11:27Performance findings summary
- 12:34Contributions and limits
- 13:14Q&A: Performance delays
- 15:05Q&A: CubeEdge's role
- 18:19Q&A: European initiatives
related talks
14mRISC-V: The Lowest Layer of the Cloud-Native Landscape
20mIntegrate RISC V to build Open Common Automotive Platform
22mEmpowering Edge Software Stack on Arm with High Performance Container Networking Tech
25mThe State of RISC V Software
19mAccelerating Edge AI with the Help of RISC-V
07mUbuntu on RISC-V
39mIoT Application Running on KubeEdge + ARM Platform
25mValidating RK3588 for KubeEdge: Scalable ARM64 Edge Node Simulation Without Hardware
