Nvidia driver: 515.86
CUDA: 11.6
Docker: 20.10.21
1. Nvidia container toolkit 설치(Installation Guide — NVIDIA Cloud Native Technologies documentation)
1-1. package repository 추가
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
&& curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
&& curl -s -L https://nvidia.github.io/libnvidia-container/$distribution/libnvidia-container.list | \
sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
1-2. 설치
sudo apt-get update
sudo apt-get install -y nvidia-docker2
sudo systemctl restart docker
1-3. 설치확인
sudo docker run --rm --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi
output으로 nvidia-smi 정보가 출력이 되어야 한다.
2. TensorRT container 확인 및 실행
2-1. 버전 확인 링크
Container Release Notes :: NVIDIA Deep Learning TensorRT Documentation
Container Release Notes :: NVIDIA Deep Learning TensorRT Documentation
The NVIDIA container image for TensorRT, release 22.02, is available on NGC. Contents of the TensorRT container This container includes the following: The TensorRT C++ samples and C++ API documentation. The samples can be built by running make in the /work
docs.nvidia.com
2-2. 실행 코드 정보 링크
Container Release Notes :: NVIDIA Deep Learning TensorRT Documentation
Container Release Notes :: NVIDIA Deep Learning TensorRT Documentation
About this task On a system with GPU support for NGC containers, when you run a container, the following occurs: The Docker engine loads the image into a container which runs the software. You define the runtime resources of the container by including the
docs.nvidia.com
뼈대
docker run --gpus all -it --rm -v local_dir:container_dir nvcr.io/nvidia/tensorrt:<xx.xx>-py<x>
2-3. 실제 코드 예제
sudo docker run --gpus all -it --network=host --ipc=host --shm-size 8G --rm -v /:/tensorrt_docker nvcr.io/nvidia/tensorrt:22.02-py3
2-4. TensorRT 테스트(Container 내부)
yolov5(TFLite, ONNX, CoreML, TensorRT Export · Issue #251 · ultralytics/yolov5 (github.com))
TensorRT 변환 예제를 테스트 해 보고자 pytorch 및 관련 패키지 설치 후 export.py 를 실행함
# upgrade pip
pip install --upgrade pip
# install pytorch
pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116
# cv2 구동 패키지 설치
apt update
apt-get install ffmpeg libsm6 libxext6 -y
# yolov5 git clone
git clone https://github.com/ultralytics/yolov5.git
Yolov5 폴더 내부에서
# yolov5 구동 패키지 설치
pip install -r requirement.txt
# pt 파일을 tensorrt 파일로 변환
python export.py --weights yolov5s.pt --include engine
끝
'컴퓨터 > 머신러닝 (Machine Learning)' 카테고리의 다른 글
| AMD GPU MIGraphX docker 사용 정리 (0) | 2022.12.22 |
|---|---|
| Super resolution 모델, HAT, inference 사용 정리 (0) | 2022.12.19 |
| 3080, Radeon vii, 6900xt, 딥러닝 (image classification) 학습 성능 비교 (0) | 2022.12.12 |
| Ubuntu, ROCm, AMD GPU, Docker, Pytorch 환경에서 딥러닝 정리 (0) | 2022.12.10 |
| Pytorch, grad-cam 사용 정리 (0) | 2022.12.10 |
