NVIDIA-AI-IOT / CUDA-PointPillars

A project demonstrating how to use CUDA-PointPillars to deal with cloud points data from lidar.
Apache License 2.0
502 stars 148 forks source link

inference with Multiple GPU #109

Open xiaoxifuhongse opened 6 months ago

xiaoxifuhongse commented 6 months ago

I got a server with two GPUs, both NVIDIA GeForce RTX 2080 SUPER , I want to run several lidars, so I want to make use of both , But when I cudaSetDevice(1), there is a bug like:

(pillars_infer:72328): GLib-GIO-CRITICAL **: 15:31:49.133: g_dbus_proxy_new: assertion 'G_IS_DBUS_CONNECTION (connection)' failed

so when the first gpu out of memory how can I choose the other?