Xen Summit 2026

As automotive platforms shift to centralized compute, the key challenge is consolidating mixed-critical workloads while ensuring isolation, predictability, and visibility into cross-domain interference.
This session presents a Xen-based implementation on Renesas R-Car Gen5 (X5H) using a Dom0-managed architecture to run Android IVI, Linux service/AI workloads, and real-time control with defined domain partitioning.
We cover system architecture, bring-up, and integration challenges, including CPU allocation and device assignment trade-offs. We also demonstrate on-device AI inference using LLaMA 3.2 and Qwen models in dedicated Xen domains, focusing on coexistence with IVI and real-time workloads.
Finally, we present a cross-domain observability framework using xentop and a custom dashboard, providiāng visibility into CPU usage and interference patterns. The session shares practical insights and measured behavior for deploying mixed-critical workloads on X5H