Skip to content

❓ [Question] How to run Torch-Tensorrt on JETSON AGX ORIN? #1166

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
krmayankb opened this issue Jul 5, 2022 · 5 comments
Closed

❓ [Question] How to run Torch-Tensorrt on JETSON AGX ORIN? #1166

krmayankb opened this issue Jul 5, 2022 · 5 comments
Assignees
Labels
channel: linux-jetpack question Further information is requested

Comments

@krmayankb
Copy link

❓ Question

Not able to run Torch-Tensorrt on Jetson AGX ORIN
As per the release note, it is mentioned that current release doesn't have support for Jetpack 5.0DP but ORIN only supports Jetpack 5.0DP (I might be wrong but inferring from this Jetpack Archives.. Is there a way to run Torch-Tensort on ORIN? if not what's the possible timeline for new release with this support?

What you have already tried

I did tried building for python, as suggested in the repo, it enables import torch_tensorrt but doesn't supports any attributes.

Environment

  • PyTorch Version (e.g., 1.0): 1.11
  • CPU Architecture: arm64
  • OS (e.g., Linux): Ubuntu 20.04
  • How you installed PyTorch (conda, pip, libtorch, source): tried both, wheels provided here and building from source(instruction from here).
  • Build command you used (if compiling from source): python3 setup.py --use_cxx11_abi (however, this refers to jetpack 4.6 by default)
  • Python version: 3.8
  • CUDA version: 11.4
  • GPU models and configuration: Jetson ORIN
  • Any other relevant information:
@krmayankb krmayankb added the question Further information is requested label Jul 5, 2022
@krmayankb krmayankb changed the title ❓ [Question] How do you ....? ❓ [Question] How to run Torch-Tensorrt on JETSON AGX ORIN Jul 5, 2022
@krmayankb krmayankb changed the title ❓ [Question] How to run Torch-Tensorrt on JETSON AGX ORIN ❓ [Question] How to run Torch-Tensorrt on JETSON AGX ORIN? Jul 5, 2022
@peri044
Copy link
Collaborator

peri044 commented Jul 6, 2022

We are currently working on that and plan to release a version compatible with Jetpack 5.0 soon. Here is a working PR you can try #1158

@haichuanwang001
Copy link

@krmayankb Were you able to run torch_tensorrt on Orin successfully?

@krmayankb
Copy link
Author

@krmayankb Were you able to run torch_tensorrt on Orin successfully?

@haichuanwang001 No, I couldn't. I stop trying based on feedback received.

Currently, I am working with the Python API of TensorRT. It's a bit of work to wrap your head around but it works. Let me know if you need more info about this.

@narendasan
Copy link
Collaborator

torch-tensorrt 1.1.1 was released with Orin support https://github.com/pytorch/TensorRT/releases/tag/v1.1.1

Follow the instructions here to install https://pytorch.org/TensorRT/tutorials/installation.html

@krmayankb
Copy link
Author

krmayankb commented Jul 20, 2022

@narendasan good to see this release.
image

Please refer above image. I don't see any change to setup.py it considers Jetpack 4.5 and 4.6. refer this
line. Is it fine to modify setup.py or file need an update?


UPDATE: library build and functionalities are working on ORIN with Jetpack 5.0DP, while the version of torch_tensorrt is: 1.1.0+c2e396a8 . Thanks for the support.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
channel: linux-jetpack question Further information is requested
Projects
None yet
Development

No branches or pull requests

6 participants