Parabrick Deepvariant run error

When I run deepvariant on HPC 8GPUs by a docker container limited to 6 gpus. It shows the error

Can you help me?

Thanks for your interest in Parabricks and sorry for your troubles. Although there are some hints to this in the snapshot you provide; but could you please provide a summary of the Driver and CUDA versions (perhaps the output of nvidia-smi). Also, can you share the hardware configuration of the server in question (GPU type, memory etc)?

thank you for your replay. I attach the driver and cuda version here. My HPC server is HPE Appolo Gen 10 which uses 8 GPUs x 32GB, 512GB RAM and installed ubuntu1804. I install parabricks v2.4.2 and v2.5.0 by docker. But both of them show same error. Cần you @aquraini help me?

@dongoctuan.0101 there is a message asking you to check if you have enough disk space in the output and temp (/tmp y default) directories. Can you confirm that you are not running out of disk space?

@mvella I am sure. Output will written to NFS 13TB. I think the problem is cudnn. Maybe it can not initialize cudnn.