Run Stable Diffusion on Your M1 Mac’s GPU

#1

Post by bfirsh » Thu, Sep 01, 2022, 4:19 PM UTC

Re: Run Stable Diffusion on Your M1 Mac’s GPU

#2

Post by ThrowawayTestr » Thu, Sep 01, 2022, 4:48 PM UTC

That was fast.

#3

Post by omginternets » Thu, Sep 01, 2022, 4:50 PM UTC

Is there a way to get it to run on an an Intel-based Mac? I've attempted several times, but quickly ran into dependency issues and other quirks.

#4

Post by amelius » Thu, Sep 01, 2022, 4:50 PM UTC

I'd rather see someone implemented glue that allows you to run arbitrary (deep learning) code on any platform.

I mean, are we going to see X on M1 Mac, for any X now in the future?

Also, weren't torch and tensorflow supposed to be this glue?

#5

Post by keepquestioning » Thu, Sep 01, 2022, 4:52 PM UTC

Gamechanger!

#6

Post by keepquestioning » Thu, Sep 01, 2022, 4:53 PM UTC

One beautiful thing I realized about all this progress in AI.

We will still need people to do the hard yards, and get dirt between their fingernails. I am firmly in the camp of those people.

Fancy algorithms won't dig holes, or lay out rail tracks of over hundreds of miles.. or build houses all across the world.

#7

Post by avereveard » Thu, Sep 01, 2022, 4:56 PM UTC

How fast is it on a m1?

#8

Post by ebiester » Thu, Sep 01, 2022, 4:57 PM UTC

avereveard wrote: post #7

How fast is it on a m1?

It takes 1-2 mins for a 512x512 image. It's been a lot of fun since I did this last night.

#9

Post by nathas » Thu, Sep 01, 2022, 4:58 PM UTC

avereveard wrote: post #7

How fast is it on a m1?

I haven't ran this fork yet, but about 1.3 sec/iter. Usually ~30-50 iters/sample (image).

#10

Post by ebiester » Thu, Sep 01, 2022, 4:59 PM UTC

Note: I ran this and haven't yet been able to get img2img working yet. I borked it up trying to get conda working.

It's been a lot of fun to play with so far though!