GPU Tail Latency Diagnosis for Serverless and HPC Workloads using eBPF

Diagnosing GPU tail latency spikes in serverless and multi-tenant cloud environments is critical for maintaining service-level objectives, yet existing monitoring tools lack the granularity for root cause analysis in guest-only settings. We introduce an eBPF-based telemetry system that provides unified host-side monitoring of GPU workloads, correlating eBPF-derived host metrics with GPU-internal events for holistic system observability. The system achieves 81–88% diagnostic accuracy, detects spikes within 5 seconds, and completes root cause analysis in 6–8 seconds, operating with 1.21% CPU overhead at 100Hz sampling. Evaluated on multi-tenant GPU inference workloads, the system identifies root causes including NIC contention, PCIe pressure, and CPU interference, enabling production debugging for elastic GPU platforms without requiring fabric privileges.