4. Use of triton c++

4.1 use of tritonclient c++

4.2 use of triton c++

4.3 Dependency installation

1. Install minconda

        Minconda is a lightweight alternative to Anaconda. By default, only python and conda are installed, but you can install the required packages through pip and conda

1) download

Download the version of Miniconda that matches your own system from the official website — conda documentation

wget https://repo.anaconda.com/miniconda/Miniconda3-py38_4.12.0-Linux-x86_64.sh

bash Miniconda3-py38_4.12.0-Linux-x86_64.sh

一路enter或yes

The installation location is PREFIX=/root/minanaconda3

export CONDA_EXE='/root/miniconda3/bin/conda'
export _CE_M=''
export _CE_CONDA=''
export CONDA_PYTHON_EXE='/root/miniconda3/bin/python'

2) Set environment variables

echo export PATH=/root/miniconda3/bin:$PATH >> ~/.bashrc
source ~/.bashrc

Check the version information, if the value of PYTHONPATH is set, please unset pythonpath

 3) Change source

When using conda install, it is downloaded from the conda-forge channel by default.

Tsinghua source can be added

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --set show_channel_urls yes

 4) Use the conda virtual environment

Step 1: Create a virtual environment

conda create -n myenv python=3.8 -y

Step 2: Activate the virtual environment

conda activate myenv

The reason for this problem is that there is a problem with the activation command of conda. The solution is as follows

source activate

You can see that it was successfully activated and entered the myenv environment 

2. Install faiss gpu

conda install faiss-gpu

Ask the next question

 The solution is

conda config --add channels conda-forge

The faiss-gpu version is 1.7.2, and some installed software will be automatically updated at the same time.

numpy, faiss, cudatoolkit and other software are installed, conda, openssl and other software are updated

Install the specified version of faiss-gpu, faiss-cpu

conda install faiss-gpu==1.7.0

The available versions of faiss-gpu are as follows 

 

Install the specified version of faiss-cpu

conda install faiss-cpu==1.7.1

3. Install pytorch

conda install pytorch -y

##卸载
conda uninstall pytorch

The latest version of pytorch will be installed, and the specified version of pytorch will be installed, such as

conda install pytorch=0.1.10

Install pytorch that conforms to the cuda version, the official website Start Locally | PyTorch

 The installation command is as follows

conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia

You can see that cuda is installed

 

4. Install opencv

conda install opencv

##卸载
conda uninstall opencv

The latest version of opencv4.6 is installed by default, and ImportError: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version `GLIBCXX_3.4.30' not found (required by /root/blender/3.2/python/ lib/python3.10/site-packages/../../libopencv_gapi.so.406)

And strings /usr/lib/x86_64-linux-gnu/libstdc++.so.6 | grep GLIBCXX found that GLIBCXX_3.4.30 is not supported, the best way is to downgrade opencv, and upgrading libstdc++.so.6 will be troublesome and may will not succeed

Install the specified version of opencv 

 conda install opencv-python==4.5.1.48 -y

 Note the available opencv versions

5. Install skimage

conda instal scikit-image

6. Install kafka

conda install kafka-python