# simulating setup-system because install-docker currently broken
$ ./data-science-stack install-base
...
$ ./data-science-stack install-driver
...
$ reboot
...
$ ./data-science-stack diagnostics
...
###NV### Wed Feb 10 12:32:15 UTC 2021 #### Driver detected (0 means not installed): 0
###NV### Wed Feb 10 12:32:15 UTC 2021 #### NVIDIA SMI:
nvidia-smi not found, NVIDIA GPU Driver not installed correctly.
###NV### Wed Feb 10 12:32:15 UTC 2021 #### CUDA detected (0 means not installed): 0
###NV### Wed Feb 10 12:32:15 UTC 2021 #### Docker detected (0 means not installed): 0
$ lsmod | grep nvidia
nvidia_drm 57344 0
nvidia_modeset 1224704 1 nvidia_drm
nvidia 34086912 1 nvidia_modeset
drm_kms_helper 217088 1 nvidia_drm
drm 557056 3 drm_kms_helper,nvidia_drm
nvidia-smi is part of xorg-x11-drv-nvidia-cuda (fedora; rpmfusion) and nvidia-driver-cuda (rhel; cuda-rhel from nvidia) or xorg-x11-drv-nvidia-cuda (rhel, non-free rpmfusion), not the driver, and isn't available until install-cuda.
install-cuda needs be part of setup-system if nvidia-smi is going to be part of the driver identification.
nvidia-smi is part of xorg-x11-drv-nvidia-cuda (fedora; rpmfusion) and nvidia-driver-cuda (rhel; cuda-rhel from nvidia) or xorg-x11-drv-nvidia-cuda (rhel, non-free rpmfusion), not the driver, and isn't available until install-cuda.
install-cuda needs be part of setup-system if nvidia-smi is going to be part of the driver identification.