AITemplate
AITemplate copied to clipboard
How we can run stable diffusion models as parallel at the same time?
Hi guys, I compiled the SD model with batch size = 1 and achieve inference time = 2 sec. Then I compiled the same model with Batch size = 4, I achieve inference time = 10 sec for 4 images. This shows us the models don't process as parallel and do sequential way. Why? How I can solve the problem?