Using Device CPU issue

[okorlu]

Hi,

I am trying the deploy Dalle-Flow server on my M1Pro chip MacbookPro. The machine has GPU, but Dalle-Flow doesn't use it! In the Dalle-diffusion summary phase, it said "Using device: cpu" How can I change this?

This can be a newbie question, because I am a newbie. Any help is much appreciated.

[JoanFM]

Hey @okorlu,

Can you please show the logs when using JINA_LOG_LEVEL=DEBUG before starting the Flow? Which Executor do you want to have running in GPU?

[okorlu]

Hey @JoanFM
It was right before waiting dalle... part. It looks like this

[JoanFM]

Hey @okorlu,

I do not see any obvious problem in this image, what is the exact problem?

[okorlu]

Hey @JoanFM,

As you can see right after rerank/rep-0©8402, after "Waiting dalle diffusion summary" it says: "Using device cpu". In the hardware requirements, it is specifically mentioned that "CPU-only environment is not tested and likely won't work." First, I don't think this a problem. You know, may it uses it for other purposes? But than I waiting and waiting and waiting... After more than 20 minutes, still nothing happened. I must admit, when first dalle launched, I tried to deploy it on cpu only machine. So I have experience, too :) But this time I am sure, my computer do have GPUs and enough RAM. I tried to used it in virtual environments, Rosetta anything I saw on the forums, the results is not changing. I pretty upset, I cannot do anything.