I believe those are CPU specific, it disregards my AMD GPU. It's too slow a proces.
For what it's work, a sort of hybrid CPU GPU approach is giving me workable results with automatic111, about 30-50s to generate a 512x512 image ataround 1-2 it/s.
Not enough for me to want to use local-run txt2img for daily use (why, when Bing/Poe exists and give you high quality image gen for free in seconds), but enough that when I want some granular control or don't want to give my data to a company (i like making AI portraits of myself and friends), I can run locally.
If I could improve performance, I'd love to get into video production
