Hello, I am trying to run DeepStream application with nvinferserver and simple IoU tracker.
When I enable my tracker with gpu-id=0, running deepstream-app fails right away with the following messages:
** INFO: <create_primary_gie_bin:154>: gpu-id: 0 in primary-gie group is ignored, only accept in nvinferserver's config
max_fps_dur 8.33333e+06 min_fps_dur 2e+08
Civetweb version: v1.16
Server running at port: 9000
gstnvtracker: Loading low-level lib at /opt/nvidia/deepstream/deepstream/lib/libnvds_nvmultiobjecttracker.so
GPUassert: invalid device ordinal src/modules/NvMultiObjectTracker/context.cpp 61
when I set [tracker]/enable=0 in the app txt config, the app runs perfactly fine.
Environment: I am running an nvinferserver model described as here: Custom Detection parser error with nvinferserver and custom python model with > 1 streams.
My app runs in a docker container on a dual RTX A4000 , Ubuntu 22.04 machine, using DeepStream 6.3 container. Driver Version: 535.104.05 CUDA Version: 12.2;
The container has only 1 gpu assigned (using docker compose: deploy:resources:reservations:devices).
I am attaching the relevant configuration files:
deepstream-app.txt (1.5 KB)
config_tracker_IOU.yml (1.9 KB)
config_infer_primary.pbtxt (3.1 KB)