Middleware By Example

I've two 1080ti GPUs , I would like to use only one of the device for DeepLearning Training purpose,Most of the google suggestions advising to use CUDA_VISIBLE_DEVICES environment variable.Setting CUDA_VISIBLE_DEVICES either in windows environment variables are as programmatic way dint help much. But below approach helped to control the GPU usage for CUDA operations. Open NVIDIA Control Panel" by right clicking on the desktop screen. And select CUDA-GPUs to the desired gpu you want to use for DL training, In my case: I chose 2nd GPU Peace!!