Forked from https://github.com/lzhengchun/matrix-cuda.
https://www.notion.so/justin0u0/GKE-GPU-Time-Sharing-7ba904b3cf9848d9896645f9da4651c0
Change the command
in the job.yaml the apply the manifest.
kubectl apply -f ./manifests/job.yaml
# optional
kubectl apply -f ./manifests/cuda.yaml # can execute `nvidia-smi` in the pod
# optional
kubectl apply -f ./manifests/devicequery.yaml # can execute `deviceQuery` in the pod