How to parse the compilation of pytorch-gpu source code
This article introduces how to parse pytorch-gpu source code compilation, the content is very detailed, interested friends can refer to reference, I hope to help you.
Software and hardware environment ubuntu 18.04 64bitNVidia GTX 1070Tianaconda with python 3.7CUDA 10.1cuDNN 7.6python 1.8python and GPU environment
This is not much to say, not installed, you can refer to the link below
anaconda basic use ubuntu installation CUDA and cuDNN compilation steps
Installation Base Dependency
conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi typing_extensions future six requests dataclasses
Since you want to use GPU, you also need to install LAPACK support, and install the corresponding software package according to CUDA version.
# Add LAPACK support for the GPU if needed
conda install -c pytorch magma-cuda101 # or [ magma-cuda101 | magma-cuda100 | magma-cuda92 ] depending on your cuda version
Then you can start cloning code.
git clone --recursive https://github.com/pytorch/pytorch
cd pytorch
# if you are updating an existing checkout
git submodule sync
git submodule update --init --recursive
Once the preparations are complete, you can start compiling.
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../ "}
python setup.py install
CMAKE_PREFIX_PATH is actually the installation directory of anaconda, such as/home/xugaoxiang/anaconda3
pytorch
It takes a long time to compile, just wait patiently.
test
Open a new terminal for testing.
(base) xugaoxiang@1070Ti:~$ ipython
Python 3.7.6 (default, Jan 8 2020, 19:59:22)
Type 'copyright', 'credits' or 'license' for more information
IPython 7.19.0 -- An enhanced Interactive Python. Type '? ' for help.
In [1]: import torch
In [2]: torch.__ version__
Out[2]: '1.8.0a0+46d846f'
In [3]: torch.cuda.is_available()
Out[3]: True
In [4]:
Pytorch has been installed in the anaconda environment, and is the gpu version, done!
The whole compilation process is still very smooth, before we compiled opencv, tensorflow, caffe, compared with them, pytorch in the dependency processing is very good, in the form of sub-projects into their own projects, the whole compilation process did not appear error, the document is also written simple and clear, no wonder its market share will be higher and higher.
About how to parse pytorch-gpu source code compilation to share here, I hope the above content can be of some help to everyone, you can learn more knowledge. If you think the article is good, you can share it so that more people can see it.
