comfyanonymous/ComfyUI_TensorRT

Issues

• i need LORA support

  #42 opened a month ago by sev7eny
  3

• ModuleNotFoundError: No module named 'tensorrt_bindings'

  #18 opened 2 months ago by rinaldop
  5

• The first latent space if full of nan, resulting in a black image for the first of each batch

  #5 opened 11 days ago by Extraltodeus
  3

• IP Adapter Support

  #48 opened 24 days ago by Sundragon1993
  4

• Exporting the operator 'aten::scaled_dot_product_attention' to ONNX opset version 17 is not supported. Please feel free to request support or submit a pull request on PyTorch GitHub: https://github.com/pytorch/pytorch/issues.

  #21 opened 2 months ago by koyamakkkkk
  4

• Error occurred when executing DYNAMIC_TRT_MODEL_CONVERSION

  #53 opened 18 days ago by jikeshiganjia
  0

• Is FreeU_V2 support?

  #52 opened 23 days ago by Hamidawaan
  0

• Error occurred when executing FreeU_V2: 'model_channels'

  #51 opened 23 days ago by Hamidawaan
  0

• I have rewrited this and added support for LORAs, FreeU, HyperTile, etc.

  #40 opened a month ago by yusing
  1

• recommend to cancel "connect to gitlab"

  #47 opened a month ago by gaoxuxu110
  0

• add ControlNet support

  #10 opened 2 months ago by geroldmeisinger
  4

• How to run SD3 with Tensor RT?

  #45 opened a month ago by Jay19751103
  2

• Hangs during TensorRT compilation

  #46 opened a month ago by nikolatesla20
  0

• [Request] Example workflows for SD3 models.

  #28 opened a month ago by syddharth
  12

• TypeError: a bytes-like object is required, not 'NoneType

  #43 opened a month ago by Lolagatorade
  7

• tensors are not unloaded from VRAM unless ComfyUI is restarted

  #44 opened a month ago by endertargaryen
  0

• Built engines are placed in the "output" folder

  #13 opened 2 months ago by aikitoria
  5

• [Issue] Stable Diffusion 3 TRT Generates gibberish

  #41 opened a month ago by 1aienthusiast
  2

• [executionContext.cpp::nvinfer1::rt::ShapeMachineContext::resolveSlots::2842] Error Code 3: API Usage Error (Parameter check failed at: runtime/api/executionContext.cpp::nvinfer1::rt::ShapeMachineContext::resolveSlots::2842, condition: allInputDimensionsSpecified(routine) )

  #31 opened a month ago by DarkAlchy
  1

• output (.engine) files go to --output-directory but are expected in default output directory

  #38 opened a month ago by pollockjj
  0

• "Exception during processing!!! ONNX export failed. Could not open file or directory."

  #37 opened a month ago by Reno0vacio
  0

• Feature request: P100/p4 support

  #36 opened a month ago by dirkson
  0

• Results in random black image generation.

  #7 opened a month ago by WeirdScienceX
  27

• all SD3 engines generate black images randomly (almost all of the time)

  #25 opened a month ago by Shaistrong
  14

• [Bug] RuntimeError: The size of tensor a (96) must match the size of tensor b (168) at non-singleton dimension 3

  #29 opened a month ago by yusing
  1

• What do I do to delete the list of built models?

  #27 opened a month ago by alex13by
  2

• Error occured when executing KSampler

  #17 opened 2 months ago by fusillustrator
  1

• Instructions for downloading, installing and using the pre-converted TensorRT versions of SD3 Medium with ComfyUI and ComfyUI_TensorRT

  #23 opened a month ago by MarcusJT
  2

• mat1 and mat2 must have the same dtype

  #22 opened 2 months ago by koyamakkkkk
  1

• FP8 support for TensorRT?

  #24 opened a month ago by wingmrc
  0

• Can the width and height be dynamically adjusted? For example, support 576×768 or 768×576 at the same time?

  #19 opened 2 months ago by petercham
  2

• Efficient Loader Node appears to enable the loading of LoRAs.

  #20 opened 2 months ago by JPW0080
  4

• an illegal memory access was encountered

  #9 opened 2 months ago by GavChap
  9

• IMPORT FAILED

  #12 opened 2 months ago by yotraxxx
  8

• How to load only CLIP from checkpoint

  #15 opened 2 months ago by aikitoria
  3

• Error occurred when executing DYNAMIC_TRT_MODEL_CONVERSION

  #11 opened 2 months ago by ZayshaaCodes
  3

• With the current security level configuration, only custom nodes from the "default channel" can be installed.

  #6 opened 2 months ago by AlexD81
  1

• Can't convert SDXL models 'aten::scaled_dot_product_attention' to ONNX opset version 17 is not supported

  #8 opened 2 months ago by NeoLoger
  2