Jabasukuriputo Wang

While the concept looks helpful, the way you proposed will only work under a harsh requirement, where any extension must not share any part of the dependency tree with others....

There is not much you can do with these ones containing C extensions.

It's kinda easy you can just snip in ```python torch.backends.cudnn.enabled = False torch.backends.cuda.enable_flash_sdp(False) torch.backends.cuda.enable_math_sdp(True) torch.backends.cuda.enable_mem_efficient_sdp(False) ``` at somewhere around https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/master/modules/initialize.py#L57

You can add it below https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/master/modules/initialize.py#L14 instead.

btw you should use `--opt-sdp-attention` if you are not using it

>RuntimeError: CUDA error: CUBLAS_STATUS_NOT_SUPPORTED when calling `cublasSgemm( handle, opa, opb, m, n, k, &alpha, a, lda, b, ldb, &beta, c, ldc) That means the code is not correctly applied.

Did you correctly replace the cublas dll

## Smart mover The smart mover does something similar to forge, and it only move tensors from CPU to GPU, but not coming back. At some point, I was somehow...

The broken images seems to be caused by not synchronize back the after usage to the creation stream. Fixed. Also changed to layer-wise movement.

There might be problem with extra networks. Haven't look into that.