search

OpenVisualCloud / Smart-City-Sample

The smart city reference pipeline shows how to integrate various media building blocks, with analytics powered by the OpenVINO™ Toolkit, for traffic or stadium sensing, analytics and management tasks.
BSD 3-Clause "New" or "Revised" License
189 stars 81 forks source link

"GPU" name is not registered in the InferenceEngine #736

Open Gsarg18 opened 3 years ago

Gsarg18 commented 3 years ago

I am not able to run inference model of smart city on the GPU device

xwu2git commented 3 years ago

More details please. What GPU device do you use? Currently the sample only supported the VCAC-A accelerator. To support other GPU devices, you have to extend the sample to use different base analytics images, for example, replace xeon-ubuntu1804-analytics-gst with xeone3-ubuntu1804-analytics-gst under analytics/common. There are other changes to the pipeline configuration to utilize VAAPI for decoding.

Gsarg18 commented 3 years ago

Currently I am using Intel® UHD Graphics 630 (CFL GT2) graphics card

Gsarg18 commented 3 years ago

Is xeone3-ubuntu1804-analytics-gst image support Intel® UHD Graphics 630 (CFL GT2) graphics card? I am using Intel® Core™ i7-8700 CPU @ 3.20GHz × 12 processor.

xwu2git commented 3 years ago

You have to try it. It should work.

xwu2git commented 3 years ago

I have a branch here. Not yet successful enabling inference. Something is missing.

nnshah1 commented 3 years ago

@xwu2git Do you have a system that reproduces the issue? We recently ran into an issue with Comet Late GPUs and the XeonE3 container. I will send the details to you offline.

whbruce commented 3 years ago

See issue #662 I posted on OVC Dockerhub repo, but it is specific to Comet Lake. Issue shows how to check for OpenCL device.