I searched and read documents on TensorRT API but i cannot find information on how to bind a device to an inference execution. I want to ask if TensorRT support device binding.
Take example :
-App1 : Bind GPU0 to execute an engine file (A.trt) to detect face
-App2 : Bind GPU1 to execute an engine file (B.trt) to classify image
So CUDA cores on GPU0 only use to run A.trt and CUDA cores on GPU1 will only use to run B.trt
Please refer to below link in case it helps:
Thank you, that is what i am looking for,