Skip to content

Conversation

@yondonfu
Copy link

Changes in this PR:

  • Add a .gitignore to exclude a few files generated during usage.
  • I ran into division by 0 errors here and anywhere where there is a division operation with self.curr_split_batch. I was using a EmptyLatentImage node with batch_size = 1 connected to a KSampler node that used the TensorRT engine built from your branch as an input. The problem seemed to go away when I set the batch_multiplier to 1 in a4dceae. I noticed that during ONNX export the batch size is also multiplied by 2, but was not sure if that should be left as-is.
  • The static/dynamic build nodes were complaining if the onnx_model_path input was not provided which is fixed by setting a default None value in the nodes. It seems that even if the input is marked as optional at the ComfyUI node level, you still need to set a default value for the corresponding param in the function invoked by the node.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants