Stable Diffusion is broken : ( #80
-
|
It's not a problem just making you aware, Neither DiffusionBee or Stable Diffusion from the command line are not working as they should. Similar issue to #71 also, lower than expected bench, previously 175K. Can I access earlier commits from around June 14?, (I tried the version from #71), Stable Diffusion was working in Sequoia I can't remember which commit it was.
|
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 3 replies
-
|
You are on macOS Sequoia, this is a beta OS and expected to have bugs. In addition, for Sequoia on AMD, there are two pairs of PAT patches; you will need to switch both to use Shaneee's for better performance. |
Beta Was this translation helpful? Give feedback.
-
|
Also you are using ForgedInvariant on a desktop AMD CPU. Remove it. |
Beta Was this translation helpful? Give feedback.
-
|
Removed, thanks. Unfortunately, Stable Diffusion is still broken, but the Metal Bench went up to 95K!. On my 4700G build with latest NootedRed, Stable Diffusion works well under macOS Sequoia. Like you say, Sequoia is still Beta so we can expect things not to work. |
Beta Was this translation helpful? Give feedback.
-
|
Same here in Sonoma and RX6700XT, Intel CPU. |
Beta Was this translation helpful? Give feedback.
-
|
Stable Diffusion and DiffusionBee will produce bad results because of PyTorch, which ended support for X86 in January 2024. You can roll back to 2.0.2 for x86 support. @VisualEhrmanntraut this thread can be deleted as it's nothing to do with NootRX. |
Beta Was this translation helpful? Give feedback.
Stable Diffusion and DiffusionBee will produce bad results because of PyTorch, which ended support for X86 in January 2024. You can roll back to 2.0.2 for x86 support. @VisualEhrmanntraut this thread can be deleted as it's nothing to do with NootRX.