When using GFD (or the device plugin) on Tegra-based systems, we do not use NVML to extract device information.
Furthermore, we rely on heuristics to detect the platform we are working on.
We need to update the logic in the device plugin to use https://github.com/NVIDIA/go-nvlib/pull/28 to align with other components. Adding a --mode switch (default: auto) also makes sense to allow users to explicitly override it.
When using GFD (or the device plugin) on Tegra-based systems, we do not use NVML to extract device information.
Furthermore, we rely on heuristics to detect the platform we are working on.
We need to update the logic in the device plugin to use https://github.com/NVIDIA/go-nvlib/pull/28 to align with other components. Adding a
--mode
switch (default:auto
) also makes sense to allow users to explicitly override it.