exporter之nvidia-gpu监控
NVIDIA 的 dcgm-exporter 是一个开源工具,基于 **NVIDIA Data Center GPU Manager (DCGM)**,用于收集 GPU 的运行指标,并以 Prometheus 格式暴露这些数据,方便进行监控和分析。
DCGM Exporter: NVIDIA’s Data Center GPU Manager (DCGM) exporter, often used to collect GPU metrics for Prometheus.
Metrics include: memory usage, temperature, utilization, power draw, etc.
Installation typically involves running a containerized exporter on systems with NVIDIA GPUs.
