Performance of YoloV3 pytorch implmentation on Tesla V100 and GTX 1050 Ti

I am using YoloV3 pytorch implementation https://github.com/ayooshkathuria/pytorch-yolo-v3 to process a clip of two minutes which mostly is detecting vehicles.

TensorRT is not used to optimize here in both cases

While running the YoloV3 on the GTX 1050 Ti

Running the original script https://github.com/ayooshkathuria/pytorch-yolo-v3/blob/master/video_demo.py on the video takes around 4 minutes 20 seconds

Running the an altered code from the original script https://github.com/ayooshkathuria/pytorch-yolo-v3/blob/master/video_demo.py on the same video clip takes around 6 minutes 20 seconds

While running the YoloV3 on the Tesla V100

Running the an altered code from the original script https://github.com/ayooshkathuria/pytorch-yolo-v3/blob/master/video_demo.py on the same video clip takes around 6 minutes 20 seconds

How can both cases (GTX 1050 Ti Vs Tesla V100) take the same time to process the same clip?

I expected code to run faster on the Tesla V100 than on GTX 1050 Ti