❓ [Question] Is it ok to build v1.1.0 with cuda10.2 not default cuda11.3?

[wikiwen]

❓ Question

Is it ok to build v1.1.0 with cuda10.2 not default cuda11.3?
It's hard to upgrade latest gpu driver for some machine which is shared by many people. So cuda10.2 is preferred.

Environment

Build information about Torch-TensorRT can be found by turning on debug messages

• PyTorch Version: 1.11.0
• CUDA version: 10.2
• TensorRT: 8.2.4.2

[narendasan]

You can build Torch-TensorRT with 10.2 as long as you change the dependencies for CUDA, cuDNN and TensorRT, LibTorch to pull 10.2 based builds.

• TensorRT/WORKSPACE

  Line 44 in db61e90

  path = "/usr/local/cuda-11.3/",

• TensorRT/WORKSPACE

  Line 61 in db61e90

  urls = ["https://download.pytorch.org/libtorch/cu113/libtorch-cxx11-abi-shared-with-deps-1.12.0%2Bcu113.zip"],

• TensorRT/WORKSPACE

  Line 69 in db61e90

  urls = ["https://download.pytorch.org/libtorch/cu113/libtorch-shared-with-deps-1.12.0%2Bcu113.zip"],

• TensorRT/WORKSPACE

  Line 82 in db61e90

  "https://developer.nvidia.com/compute/cudnn/secure/8.4.1/local_installers/11.6/cudnn-linux-x86_64-8.4.1.50_cuda11.6-archive.tar.xz",

• TensorRT/WORKSPACE

  Line 92 in db61e90

  "https://developer.nvidia.com/compute/machine-learning/tensorrt/secure/8.4.1/tars/tensorrt-8.4.1.5.linux.x86_64-gnu.cuda-11.6.cudnn8.4.tar.gz",