CUDA driver version is insufficient for CUDA runtime version under NVIDIA Jetson Nano

[collabnix]

jetson@192:~/docker/darknet/gpu$ sudo docker run -it jitteam/devicequery ./deviceQuery
Unable to find image 'jitteam/devicequery:latest' locally
latest: Pulling from jitteam/devicequery
f9cb56a5c671: Already exists
87de231a2e6c: Pull complete
36c5d6eb2688: Pull complete
40093f2ec581: Pull complete
064d691990ab: Pull complete
cce7e876d241: Pull complete
35e6373a9792: Pull complete
edeab4125290: Pull complete
adf5dfa1dfb2: Pull complete
e6c0ad7be811: Pull complete
03b16bee77ae: Pull complete
74f8a5cf7c0d: Pull complete
37439fc13cc0: Pull complete
1bb694f9307d: Pull complete
430801120376: Pull complete
4f4fb700ef54: Pull complete
d62079597f19: Pull complete
72f994d81758: Pull complete
3fc1be53c6d7: Pull complete
Digest: sha256:e88bc1f5938a04ead500fd1486d95ba6340f2896b9789cacca6fd6d39a28bf09
Status: Downloaded newer image for jitteam/devicequery:latest
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

cudaGetDeviceCount returned 35
-> CUDA driver version is insufficient for CUDA runtime version
Result = FAIL
jetson@192:~/docker/darknet/gpu$

[loretoparisi]

Hello @collabnix this could be due to the old FROM nvidia/cuda:8.0-cudnn6-devel-ubuntu16.04. You should upgrade this a more recent CUDA / cudnn version.

I don't know if jetson nano supports the latest cuda that are: CUDA 10.2 / cudnn7 running on Ubuntu18.04. In that case you should change the like here

toFROM nvidia/cuda:10.2-cudnn7-devel-ubuntu18.04

See here for a comprehensive list: https://gitlab.com/nvidia/container-images/cuda/blob/master/doc/supported-tags.md