Performance engineering in a semi-realtime, safety-critical Linux/NVIDIA-based autonomous vehicle software stack

Focus: mixed-criticality heterogeneous compute workload optimization, resource sharing, scheduling, system reliability, latency analysis, long-tail system behavior analysis, Linux observability and tracing

Linux, perf_events, BPF, libbpf, bcc, cgroups, x86, aarch64, NVIDIA, CUDA, DriveOS, ROS, Python, C++, C

System Performance

System Performance

Observing GPU Runtime Behavior Inside a Self-Driving Car