Runpod Running Slower Than Local Machine
I conducted a benchmark test on stable diffusion image-to-image. My pipeline involves using ControlNet(inpaint + Canny), I read each frame of a video and pass it through the pipeline.
On Runpod, I utilized the image nvidia/cuda:12.2.2-cudnn8-devel-ubuntu22.04 with RTX4090, 12 vCPUs, and 30GB of RAM (community GPU cloud). I only achieved 15fps. However, when I ran the same code on my local machine, equipped with an RTX 4070, 16GB of RAM, i7 13700, and Ubuntu 22.04, it reached 19fps.
I'm curious if anyone can explain why this happend. Could there be some overhead with the Docker container on Runpod that is affecting performance?
0 Replies