Splunk Technology Add-on
TA-Nvidia
GPU metrics
Free for personal use · commercial license required
GPU
nvidia-smi on PATH (NVIDIA driver installed) · Splunk 9.x
Why TA-Nvidia exists
TA-Nvidia runs nvidia-smi as scripted inputs and flattens the full query into per-GPU metrics (temperature.gpu_temp, utilization.gpu_util and more) with dashboards for fleet and per-GPU views.
What it collects
- Full nvidia-smi query, flattened to nvidia.* metrics
- Per-GPU temperature, utilization, ECC, clocks, power
- Running compute processes
Runs on stock tools
No daemons, no agents, no exotic collectors - it shells out to well-known CLIs or reads sysfs, and parses the result into Splunk events.
External tools used: nvidia-smi on PATH (NVIDIA driver installed) · Splunk 9.x.