Skip to content

Latest commit

 

History

History
102 lines (56 loc) · 3.53 KB

README.md

File metadata and controls

102 lines (56 loc) · 3.53 KB
Raw

Pytorch OSX Build

Unfortunately, the Pytorch team does not release binary package for Mac OS with CUDA support. This project provides off-the-shelf binary packages. Both Python 2.7 and 3.7 are supported now!

很不幸，Pytorch团队不发布 Mac OS CUDA版。本项目提供 Mac OS 上编译好、可直接安装的Pytorch CUDA版本。本项目同时支持Python 2.7 和 3.7 了！

Performance Warning

【2020.02.18】I benchmarked Pytorch 1.3.1 with CUDA 10.1 and CUDNN 7.6.5 on Mac OS X 10.13.6 and Ubuntu 16.04, performance on Mac OS is around 2/3 of that on Ubuntu. In addition, it is more likey to encounter "CUDA OUT OF MEMORY" error on Mac OS since the operating system takes a large amount of GPU memory for display. Be aware of this performance difference and if you have a lot of data to process, you would better turn to Ubuntu!

The following table lists training time of MNIST image classification demo.

Settings Single GPU Dual GPUs
Ubuntu 1080Ti 99.76 s 50.99 s
MacOS 1080Ti 156 s 82 s

Compile Yourself

If you find the releases cannot meet your requirements, you can compile from source youreself.

  • Guides are avaiable:
  • Source pathces are availabe at source_pathes folder of master branch.

Releases

You can find releases in release page.

你可以在Release页面找到发布版本。

Installation for Python 2.7

First, ensure your CUDA driver and cudnn is installed properly, and copy dependencies in folder usr_local_lib to path /usr/local/lib. Also, install OpenMP using Homebrew.

首先，确保CUDA驱动和cudnn正确安装,并且将文件夹usr_local_lib中的依赖项复制到路径/usr/local/lib。也要通过Homebrew安装OpenMP。

sudo mkdir /usr/local
sudo mkdir /usr/local/lib
sudo cp usr_local_lib/* /usr/local/lib/
brew install libomp
brew link --overwrite libomp

Second, uninstall the previous pytorch installtion by

其次，卸载之前版本的pytorch:

pip uninstall torch

Install the wheel package from this project:

安装：

pip install torch*.whl

Install torchvision:

安装torchvision：

pip install -U torchvision

Installation for Python 3

Install Python 3.x from Homebrew first, and then simply follow the guide for Python 2.7 and replace pip command with pip3 and python with python3.

首先从Homebrew安装Python 3.x，然后按照Python 2.7的安装步骤执行，注意将pip替换为pip3，并用python3启动python。

Enjoy!

开始使用Pytorch吧!

Source Code

Source code from: https://github.com/pytorch/pytorch

Related Links

If you need Tensorflow builds for osx, go to this page: https://github.com/TomHeaven/tensorflow-osx-build

If you need MxNet builds for osx, go to this page: https://github.com/TomHeaven/mxnet_osx_build

如果你需要Tensorflow包，请看这个页面：https://github.com/TomHeaven/tensorflow-osx-build

如果你需要MxNet包，请看这个页面：https://github.com/TomHeaven/mxnet_osx_build