I have installed fastai and Pytorch on a Win 11 desktop with an Intel CPU and Nvidia RTX card.
As I run various training steps, how can I be sure that the GPU is being invoked?
Is there a parameter setting to enable the GPU usage?
I also tried fastai.show_install() and got error messages.
Thanks in advance.
I have used Windows Task Manager for a quick way to get a pop-up window that shows the GPU. Go to the Performance tab and you can quickly see how much GPU memory is being used and the utilization. You can also run nvidia-smi -l (-l stands for loop) to see the details in a terminal like anaconda powershell.
Appreciate the guidance. Both good ideas, and both worked. While training the small model, GPU utilization went up to 35%. So FastAI found the GPU on its own. Thanks.