I’m working on GPU metrics gathering for my product. On Windows currently we use NVWMI to gather GPU Usage, Encoder Usage, Decoder Usage and Bus Usage. For Linux implementation I’m using NVML to gather metrics, but I’m wondering what are differences between NVML metrics and NVWMI metrics? Can I interpret them as identical data? What interval is used to measure a specific metric? Whether Nvidia WMI metrics and NVML metrics differ from one another or if mentioned tools use different intervals to measure specific metrics? Currently I’m struggling to find descriptive documentation of NVWMI and its correlation with PerformanceCounters and what is proper way to use these PerformanceCounters. Any help would be very appreciated.