error when running test.sh

error when running test.sh  

  By: vernayb on Feb. 24, 2022, 11:53 p.m.

I ran in this error at the end of running the test.sh script (the previous step with process.py EXECUTE_IN_DOCKER = False yielded the correct results )

Successfully built 71c8050df253 Successfully tagged conic-inference:latest docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]].

I think it comes from the factthat we don't use the GPU for inference/composition. Is there some code to change in the docker file:

LABEL nl.diagnijmegen.rse.algorithm.hardware.cpu.count=4 LABEL nl.diagnijmegen.rse.algorithm.hardware.cpu.capabilities=() LABEL nl.diagnijmegen.rse.algorithm.hardware.memory=16G LABEL nl.diagnijmegen.rse.algorithm.hardware.gpu.count=1 LABEL nl.diagnijmegen.rse.algorithm.hardware.gpu.cudacomputecapability=11.1 LABEL nl.diagnijmegen.rse.algorithm.hardware.gpu.memory=12G

or do I need to have nvidia-docker, cuda and cudnn installed anyway to be able to export the docker file?

Thank you

Bertrand

Re: error when running test.sh  

  By: simongraham73 on Feb. 25, 2022, 11:03 a.m.

Like you suggested, you should have docker engine and nvidia docker container as we instruct in our videos to test your container.

Simon