Does YOLOv5 dual GPU training requires SLI? (Nvidia GPUs)

I have set up yolov5 for training custom dataset. It works for one GPU. But When I tried the command given in the manual for multi-gpu training, it raised error:
Dual GPU training attempt:

However, when I run the training command usually used for single GPU, it works, and in my hardware monitor both GPUs are doing some work: Run with the command for regular training

I’m not sure did both gpu divide the task and work as intended. The weight was successfully generated though. I have 1080 and 1070ti on msi Z270-a pro motherboard. I don’t have the SLI bridge set up.

CUDA => 11.6

cuDNN => 8.4.1

python => 3.9.12

pytorch => 1.11.0

Thank you!