easy-tensorrt-sys 0.2.1

Rust binding to NVIDIA TensorRT, forked from tensorrt-rs-sys
docs.rs failed to build easy-tensorrt-sys-0.2.1
Please check the build logs for more information.
See Builds for ideas on how to fix a failed build, or Metadata for how to configure docs.rs builds.
If you believe this is docs.rs' fault, open an issue.

easy-tensorrt-sys

Rust binding to NVIDIA TensorRT.

This is a fork of tensorrt-rs-sys, which seems no longer activately maintained.

What's new

  1. Replaced the dependency cuda-rs with cudarc for cuda interaction, which is actively maintained and offers better compatibility.
  2. Add init_libnv_infer_plugins to address initialization issues with TensorRT plugins.

Installation

To use easy-tensorrt-sys, ensure you have the following installed:

  • CUDA Toolkit (e.g., CUDA 11.8)
  • TensorRT SDK (e.g., TensorRT 8.6.1.6)
  • Rust (with cargo)

Linux

  1. Install CUDA and TensorRT (e.g., via .deb packages or tarballs).
  2. Set environment variables if your install paths are non-standard:
export CUDA_INCLUDE_PATH=/usr/local/cuda/include

export TENSORRT_INCLUDE_PATH=/usr/local/TensorRT-8.6.1.6/include

export TENSORRT_LIB_PATH=/usr/local/TensorRT-8.6.1.6/lib

Windows

Install CUDA and TensorRT from NVIDIA’s website.

Set environment variables:


$env:CUDA_INCLUDE_PATH = "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\include"
$env:TENSORRT_INCLUDE_PATH = "C:\Program Files\NVIDIA GPU Computing Toolkit\TensorRT\include"
$env:TENSORRT_LIB_PATH = "C:\Program Files\NVIDIA GPU Computing Toolkit\TensorRT\lib"

Notes

  • Tested on Linux, with TensorRT 8.6.1.6, CUDA 11.8.

  • Other platforms or versions are not guaranteed to work.

Contributions

Contributions are welcome! Feel free to open issues or submit pull requests.