Which Pytorch/Onnx version can work with TensorRT6.0?

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Description

I have been trying to get a model in Pytorch to run on my Xavier with TensorRT6.0+cuda10.0, and I keep getting errors like [TRT] (Unnamed Layer* 0) [Convolution]: at least 4 dimensions are required for input. I have tried pytorch 1.10 and 1.2

After reading a couple of posts in the forum it seems one way to fix it is to upgrade to TensorRT7.0, but TensorRT7.0 has a dependency on cuda 10.2, which I cannot upgrade to since it slows down other parts of our program (normal cuda kernels) by a significant amount as compared to cuda 10.0.

The question is, when TensorRT6.0 was released, what was the pytorch version it could work with using Onnx?

Environment

TensorRT Version: 6.0
GPU Type: Volta in Xavier
Nvidia Driver Version: Jetpack 4.3
CUDA Version: 10.0
CUDNN Version: 7
Operating System + Version: ubuntu 18.04.6
Python Version (if applicable):
TensorFlow Version (if applicable):
PyTorch Version (if applicable): 1.2
Baremetal or Container (if container which image + tag):

Relevant Files

Please attach or include links to any models, data, files, or scripts necessary to reproduce your issue. (Github repo, Google Drive, Dropbox, etc.)

Steps To Reproduce

Please include:

  • Exact steps/commands to build your repro
  • Exact steps/commands to run your repro
  • Full traceback of errors encountered
2

Hi,

Please refer PyTorch Release Notes :: NVIDIA Deep Learning Frameworks Documentation.
But we recommend you to please use latest TensorRT version, as many issues have been fixed. You can upgrade Jetpack version.

Thank you.

1 Like