search

ApolloAuto / apollo

An open autonomous driving platform
Apache License 2.0
25.25k stars 9.72k forks source link

Docker image pull issue with NVIDIA 4070 ti super #15516

Open MRBXCD opened 2 months ago

MRBXCD commented 2 months ago

Desktop (please complete the following information):

Describe the bug I am trying to build the Apollo 9.0 on my PC, and I encountered a problem with docker image pull. Since my GPU belongs to NVIDIA 40 series, I have modified the docker image version VERSION_X86_64 in docker/scripts/dev_start.sh under the instruction of official document.

However, when I run the dev_start.sh script, an error is raised: image I have run the script for several times under proxy network and direct network, but the error still exist. I noticed that an issue asked the same problem, but there is no solution for the error and the issue was closed. How could I fix this error? Many thanks for the reply.

LOTEAT commented 2 months ago

This may be due to permissions. Have you tried using sudo bash?

MRBXCD commented 2 months ago

This may be due to permissions. Have you tried using sudo bash?

Thank you for the advice, I have tried sudo bash, it works and the issue mentioned above is solved. image In the next step, I executed sudo bash docker/scripts/dev_into.sh to get permission to enter the docker container. After that, it seems that the container is successfully launched, but I can not execute nvidia-smi command to check the GPU. The terminal gives me nothing after I execute that command: image I followed the official instruction and modified each file it mentioned. However, when I tried the command ./apollo.sh build_opt_gpu perception, the terminal showed that No GPU device found and raised an error: image I wonder have you ever encountered this problem? Many thanks for your reply

LOTEAT commented 2 months ago

I haven't encountered this problem. But it seems that there is a problem with the cuda driver. Have you installed the cuda driver on WSL2? If everything works fine when using nvidia-smi on WSL2, you can check the version requirements of the cuda driver in the official documentation. It may be caused by an abnormal WSL2 cuda driver or an outdated version.