Questions about running Carla on GPU servers

opendilab / InterFuser

[CoRL 2022] InterFuser: Safety-Enhanced Autonomous Driving Using Interpretable Sensor Fusion Transformer

Apache License 2.0

514 stars 42 forks source link

Questions about running Carla on GPU servers #83

Open YilinGao-SHU opened 6 months ago

YilinGao-SHU commented 6 months ago

Thanks so much for your work! But I got some puzzle problems when I try to run Carla on a GPU Server with 8 GPUs, when I use the following commands to start a Carla server, all the process run on gpu 0 like this. CUDA_VISIBLE_DEVICES=0 ./CarlaUE4.sh --world-port=20000 -carla-server & CUDA_VISIBLE_DEVICES=1 ./CarlaUE4.sh --world-port=20002 -carla-server & Have u ever met this mistake?

deepcs233 commented 6 months ago

Hi! Could you show the output of nvidia-smi? I can not understand the screenshot :)

YilinGao-SHU commented 6 months ago

Hi! Could you show the output of nvidia-smi? I can not understand the screenshot :)

Of course! The usage of the GPUs looks like this. It can be found that all Carla process run on GPU 0. Looking forward to your reply~

deepcs233 commented 6 months ago

Hi, this is my nvidia-smi output and it looks ok:

the corresponginf script:

CUDA_VISIBLE_DEVICES=0 ./CarlaUE4.sh --world-port=2000 -opengl  &
sleep 1
CUDA_VISIBLE_DEVICES=1 ./CarlaUE4.sh --world-port=2002 -opengl  &
sleep 1
CUDA_VISIBLE_DEVICES=2 ./CarlaUE4.sh --world-port=2004 -opengl  &
sleep 1
CUDA_VISIBLE_DEVICES=3 ./CarlaUE4.sh --world-port=2006 -opengl  &
sleep 1

YilinGao-SHU commented 6 months ago

Thanks so much! I will have a try!