Ubuntu:name trimesh is not defined

[formoree]

I just rented a 4090 server with torch and python installed on the server itself, and I pip-installed genesis-world as requested.But I got this error when running the hello_world demo file.

[formoree]

I found that switching to another cloud server rental platform and using a higher version of Cuda can solve this problem.

[formoree]

After trying to run the hello genesis file with no error, I tried to run my own project file.On my local 4060ti computer, each step takes about 4~5 seconds, of which the collection time is about 4 seconds, but it takes 140 seconds here. I don't know why. I have rented multiple cloud servers and all have similar problems.

[Kashu7100]

Is your simulation using GPU (I suspect it is working with CPU)? Can you check the initial log for the backend?
For your first comment it says backend is gs.cpu (although you mentioned you are using 4090 server).

[formoree]

Thank you very much for your reply. The first hello_world demo is specified as gs.cpu in the file, so it shows as CPU. In the third comment later, I specified it as GPU, but the GPU call is 0, and there is no traditional log output from gs.init either.

[Kashu7100]

Can you try gs.gpu on the first? Also what is your cuda version and nvidia-driver version?