Hacker News Reader

It is 50x times slower on M1 than on RTX 3090.

M1 takes ~4.2s per iteration, 3.5 minutes per image [0].

RTX 3090 takes ~4.7s per image (all 50 iterations) [1].

[0] - https://wandb.ai/morgan/stable-diffusion/reports/Running-Sta...

[1] - trust me bro

Btw that's the kind of perf I see on my M1, but I keep seeing "0.00G VRAM used" for each generation. I wonder what that's about. In Activity Monitor I do see the GPU being used.

SD measures VRAM usage by calling a specific pyTorch method which usually wraps CUDA call.

I guess whomever ported that to M1 just haven’t implemented that method.

OK, I must have misread some comments. Thanks for the update.

Reply on news.ycombinator.com