Optimizing Multi-GPU Performance for Video Transcription Tasks

[CatherineXX]

I currently have two A5500 GPUs for video transcription tasks and have set up two video queues, each corresponding to a GPU. However, while GPU0 performs well, GPU1 is much slower and not performing to its full potential.

I am facing the same issue even when executing directly using two command lines.
Here is an example command:
whisper 0.mp4 --language English --model medium --device cuda:0
whisper 1.mp4 --language English --model medium --device cuda:1