How to set specific GPU-ID to run TRT engine model with Polygraphy?

[PythonImageDeveloper]

TensorRT Version: 7
NVIDIA GPU: RTX 3060
NVIDIA Driver Version: 470
Operating System: Ubuntu 18.04

Hi there,
I've used docker tensorrt-20.11 image and I optimized two models with Polygraphy with default args and I have 2 same GPUs, but I want to run each converted TRT models in separated GPU.
Q1- When I Convert the model to TRT with GPU0, is it possible to run on GPU1? Two GPUs are same.
Q2- How I can specific GPU-ID to run TRT model?

[ttyio]

@PythonImageDeveloper ,

1. yes the engine can be reused if 2 GPUs are the same.

2. you can CUDA_VISIBLE_DEVICES to control the GPU
   https://developer.nvidia.com/blog/cuda-pro-tip-control-gpu-visibility-cuda_visible_devices/

Thanks

[ttyio]

close since no activity for more than 3 weeks, please reopen if you still have question, thanks!