https://github.com/ce-dric/tensorrt-batch

TensorRT capable of processing batch units
https://github.com/ce-dric/tensorrt-batch

docker docker-container python pytorch tensorrt tensorrt-engine tensorrt-inference

Last synced: about 23 hours ago
JSON representation

TensorRT capable of processing batch units

• Host: GitHub
• URL: https://github.com/ce-dric/tensorrt-batch
• Owner: ce-dric
• Created: 2024-07-08T03:10:58.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2024-07-08T07:52:09.000Z (3 months ago)
• Last Synced: 2024-09-26T02:05:18.476Z (about 23 hours ago)
• Topics: docker, docker-container, python, pytorch, tensorrt, tensorrt-engine, tensorrt-inference
• Language: Jupyter Notebook
• Homepage:
• Size: 145 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
### Requirements

- Docker

- Nvidia GPU

### Getting Started

1. make container (docker)

```shell

> docker run --rm -it --gpus all --shm-size=16G -p 8888:8888 -v %cd%:/workspace nvcr.io/nvidia/pytorch:20.12-py3 bash

> pip install pycuda==2024.1

```

or 

```

> docker build -t trt:latest .

> docker run --rm -it --gpus all --shm-size=16G -p 8888:8888 -v %cd%:/workspace trt bash

```

2. make onnx model

```shell

> python export.py

```

3. build tensorrt model

```shell

> trtexec --onnx=fcn-resnet101.onnx --workspace=64 --minShapes=input:1x3x256x256 --optShapes=input:4x3x256x256 --maxShapes=input:80x3x256x256 --buildOnly -saveEngine=fcn-resnet101.engine

# &&&& PASSED TensorRT.trtexec  ~

```

4. launch notebook

```shell

> jupyter notebook --ip 0.0.0.0 --allow-root --no-browser

```